article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
the nuclei heavier than iron are produced in stars by the three processes called @xmath0- , @xmath9- , and @xmath10-process @xcite .
the first two involve neutron capture reactions and @xmath11-decays .
the main difference is the neutron density and therefore the time scale of the neutron capture . for the @xmath0-process
the capture time scale is much longer ( typically one capture per year ) than the @xmath11-decay half - lives , therefore the reaction flow runs along the valley of stability up to @xmath8bi . in the @xmath9-process
the neutron density is so high ( @xmath1210@xmath13 @xmath14 ) that subsequent neutron captures drive the material far away from the line of stability close to the neutron drip - line .
when the neutron density drops and the temperature decreases ( freeze out " ) , the material decays back to stability via long @xmath11-decay chains . for about 30 proton - rich isotopes between @xmath15se and @xmath16hg other production mechanisms involving high temperatures and charged particles are required .
these so - called @xmath10-nuclei " are responsible for only less than 1% of the total solar abundance in mass range above iron , and are mainly produced in a mechanism called @xmath1 process " .
the nuclear data needs for all heavy ion production processes involve mainly cross sections and half - lives of the participating nuclei . in the stellar @xmath0-process environment temperatures of up to several hundred million
k are attained .
this requires the knowledge of cross sections in the energy range of a few tens of ev up to a few hundred kev for the important reactions , which are mainly @xmath18 but also some @xmath19 and @xmath20 reactions .
however , the involved nuclei can also exist in excited states , whereas the measurements in the laboratory include only nuclei in the ground - state ( with exception of the quasi - stable @xmath21ta@xmath22 ) , and thus an additional correction is needed .
also the stellar decay rates may be different from terrestrial ones , due to thermal excitation of the nuclei and , in the case of electron capture , due to ionization .
special cases with respect to @xmath0-process nucleosynthesis are @xmath23dy and long - lived radioactive isotopes like e.g. @xmath24se .
@xmath23dy is stable under terrestrial conditions , but can undergo bound - state @xmath11-decay when it is fully ionized ( @xmath23dy@xmath25 has a half - life of 47 d @xcite ) . among the nuclei involved ,
the long - lived isotopes like @xmath26ni and @xmath24se assume key positions , because their stellar @xmath11@xmath27-decay rate becomes comparable to the neutron capture rate ( @xmath28@xmath29@xmath30 ) .
the resulting competition leads to branchings in the @xmath0-process nucleosynthesis path .
these branching - point isotopes can be used for an in - situ diagnosis of the temperature and neutron density .
it was noticed that the effective branching ratios for @xmath24se and @xmath31kr are almost equal although their terrestrial half - lives differ by more than four orders of magnitude .
the solution are @xmath11-decays from thermally populated excited states . for @xmath24se
the @xmath11-decay rate is almost constant for @xmath32@xmath3310@xmath34k , but drops quickly to the order of a few years when low - lying first excited states become thermally populated @xcite . since the temperature dependence of the half - life
is well known from calculations @xcite , the branching at @xmath24se can be interpreted as an @xmath0-process thermometer . another important ingredient for a reliable @xmath0-process simulation are accurate cross sections . in the weak @xmath0-process ( @xmath35=60 - 90 )
uncertainties of single cross sections can produce a long - range propagation effect which affects the whole reaction flow up to the next shell closure at @xmath36=50 .
the reason is that no local equilibrium exists in the weak component ( the product abundance times cross section for neighboring isotopes in the reaction path is no longer approximately the same ) .
this was first demonstrated for @xmath37ni(n,@xmath1 ) @xcite where until five years ago experimental values at stellar temperatures ( @xmath2=30 kev ) ranged between 12 and 37 mb .
the higher cross section lead to a 45% higher yield of all isotopes between @xmath35=65 and 90 in stellar models compared to the previously recommended value of 12 mb .
a careful analysis including new measurements have led to a new recommended value of [email protected] mb at @xmath2=30 kev ( see kadonis v0.3 , www.kadonis.org ) , which is also in good agreement with the most recent n_tof result @xcite . also for the neighboring radioactive branching isotope @xmath26ni first experimental results are now published @xcite which lead to a factor of 2 higher cross section at @xmath2= 30 kev than previously recommended .
other important nuclei in the @xmath0 process , where accurate cross sections are indispensable , are isotopes at the neutron shell closures and the so - called `` @xmath0-only isotopes '' .
the latter are of pure @xmath0-process origin and thus represent the most direct probes of the @xmath0-process abundance distribution .
the desired uncertainty at stellar temperatures is @xmath381% which is up to now only achieved for some isotopes between @xmath35=110 - 176 , the average uncertainty at @xmath2=30 kev is @xmath383.8% .
the isotopes at the shell closures are directly responsible for the peaks in the solar @xmath0-abundance distribution because their small capture cross sections are hindering the reaction flux towards higher masses .
the average uncertainty at @xmath2=30 kev for all 19 stable isotopes at @xmath36=28 , 50 , 82 , and 126 is presently @xmath388.3% .
another important group of nuclei are the so - called
neutron poisons ``
. these can be all isotopes with large abundances and/or neutron capture cross sections which reduce the available neutrons per seed nuclei . in massive stars
the reaction @xmath39n@xmath19@xmath39c has the largest influence on the neutron economy .
other prominent cases are @xmath40c and @xmath41o due to their large abundances . in the case of @xmath40c
the neutrons are recovered via @xmath40c@xmath18@xmath42c@xmath43@xmath41o .
however , the large abundance of @xmath41o makes it again to a neutron poison ( ' ' self - poisoning " ) . a similar behavior can be seen also with the second neutron production reaction , @xmath44ne@xmath43@xmath45 mg . since @xmath45 mg is produced in large amounts , part of the newly created neutrons are immediately consumed by the @xmath45mg@xmath18 reaction .
all three capture reactions exhibit up to now rather large uncertainties in the order of 10% .
last but not least the neutron source reactions itself , @xmath42c@xmath43@xmath41o and @xmath44ne@xmath43@xmath45 mg , need improvements . despite all efforts in the last years these reactions are still rather uncertain in the relevant energy regions .
for a detailed discussion , see @xcite .
for the @xmath1 process at temperatures of @xmath32= 2@xmath463 gk photon - induced reactions on pre - existing seed nuclei dominate .
detailed reaction network calculations involve thousands of nuclei and ten thousands of reactions , mostly on unstable nuclei .
unfortunately , these calculations still fail to reproduce the observed solar abundance for all @xmath10-nuclei within a factor of 3 @xcite , especially for the most abundant nuclei @xmath47mo around @xmath36=50 .
the nuclear physics input for these calculations comes mainly from hauser - feshbach model calculations . to test these predictions systematic measurements in the relevant mass and energy range
are needed .
experimentally it is easier to investigate the inverse radiative capture reactions and to calculate the photo - induced reaction rate by applying the principle of detailed balance @xcite .
additionally , the experimental ground - state rate for capture provides a larger contribution to the stellar rate than for photodisintegration ( with a few exceptions , such as @xmath48gd(@xmath49)@xmath50sm ) .
the improvement of the predictive power for reaction cross sections is crucial for further progress in @xmath10-process models , either by directly replacing theoretical predictions by experimental data or by testing and improving the reliability of statistical models , if the relevant energy range , the so - called gamow window , is not accessible by experiments .
until recently no dedicated database for the @xmath10 process existed , where @xmath5 , and @xmath6 reactions for isotopes between @xmath7ge and @xmath8bi within the respective gamow window were collected .
this energy range is typically between 1 - 6 mev for proton - induced and 3.5 - 13.5 mev for @xmath51-induced reactions .
the pioneering compilation for stellar neutron capture cross sections was published in 1971 by allen and co - workers @xcite .
this paper reviewed the role of neutron capture reactions in the nucleosynthesis of heavy elements and presented also a table of recommended ( experimental or semi - empirical ) maxwellian averaged cross sections ( macs ) at @xmath2= 30 kev ( macs30 ) for nuclei between carbon and plutonium .
the idea of an experimental and theoretical stellar neutron cross section database was picked up by bao and kppeler @xcite in 1987 for @xmath0-process studies .
this compilation included cross sections for @xmath18 reactions between @xmath40c and @xmath8bi , some @xmath19 and @xmath20 reactions for isotopes between @xmath52s and @xmath53ni , and also ( n,@xmath1 ) and ( n , f ) reactions for long - lived actinides . a follow - up compilation was published in 1992 @xcite , and in 2000 the compilation @xcite was extended to big bang nucleosynthesis .
it now includes a collection of recommended macs30 values for isotopes between @xmath3h and @xmath8bi , and semi - empirical recommended values for nuclides without experimental cross section information .
these estimated values are normalized cross sections derived with the hauser - feshbach code non - smoker @xcite , which account for known systematic deficiencies in the nuclear physics input .
additionally , the database provided stellar enhancement factors and energy - dependent macs for thermal energies between @xmath2= 5 kev and 100 kev . the first version of the `` karlsruhe astrophysical database of nucleosynthesis in stars '' ( kadonis )
was released in 2005 @xcite .
this online database was an updated sequel to the previous compilations .
kadonis was updated three times until now , the fourth update is in progress .
the current version v0.3 @xcite contains recommended macs values at @xmath2=30 kev for 359 isotopes , out of which 80 are semi - empirical estimates based on theoretical data and experimental values of nearby isotopes .
the average uncertainties for the ( experimental ) macs at @xmath2= 30 kev are @xmath386.5% .
experimental information exists for almost all stable isotopes with exception of @xmath54o , @xmath55ar , @xmath56k , @xmath57v , @xmath7zn , @xmath58ge , @xmath59se , @xmath60ru , @xmath61xe , @xmath62la , @xmath63dy , and @xmath64pt .
the @xmath0-process database of kadonis provides not only recommended values but lists also the available experimental data as well as additional information from some evaluated databases , namely endf / b , jendl , and jeff .
the latter data can also be retrieved from www.nndc.bnl.gov/astro/ @xcite in the present update , the previously missing datasets for @xmath65li , @xmath66b , and @xmath54o were included , increasing the number of isotopes to 363 .
from the above mentioned 16 stable isotopes without experimental information in the stellar energy range , new measurements on @xmath57v , @xmath7zn , @xmath58ge , @xmath59se , and @xmath64pt were included .
a further extension to more neutron - rich isotopes ( 2 - 3 mass units away from stability ) is under discussion , but only semi - empirical estimates will be available for these cases .
the values from most recent evaluated libraries endf / b - vii.1 @xcite and jendl-4.0 @xcite were included , and were used in many cases instead of older time - of - flight measurements . as mentioned before , only reactions on nuclei in their ground states ( g.s . )
are measured in laboratory experiments . in the stellar environment , however , also reactions on thermally excited states of the nuclei contribute to the reaction rate . in the past , experimental g.s .
macs and g.s .
rates have been converted to stellar ones using a stellar enhancement factor ( sef ) , given by the theoretically predicted ratio between stellar rate and g.s
this approach has been found to be inadequate @xcite and will be replaced by the procedure given in @xcite , which also allows to properly define uncertainties in the stellar rates from a combination of theoretical and experimental uncertainties .
once the update with new experimental data is complete , semi - empirical values for those isotopes without experimental information will be recalculated as in @xcite .
the new data will then be included in state - of - the - art @xmath0-process models to investigate the influence of changes on the calculated @xmath0-abundance distribution .
the aim of the @xmath10-process database is the collection of all available experimental cross sections in or close to the respective gamow window and the comparison with various theoretical models . in 2006
a first draft for such a database within the kadonis project was launched , but it lasted until 2010 before this part was improved in collaboration with atomki debrecen , hungary @xcite .
the new @xmath10-process compilation started essentially from the exfor database @xcite , which has the advantage that it contains ( almost ) all available experimental cross sections and is regularly updated .
exfor was carefully scanned for experimental data from @xmath5 , and @xmath6 reactions for targets heavier than @xmath7ge and energies down to 1.5 times the upper end of the respective gamow window for @xmath32= 3 gk .
this upper cut - off energy was arbitrarily chosen because most available data were measured above the astrophysically relevant gamow windows . in the present database
309 datasets for 181 different charged - particle reactions between @xmath7ge and @xmath8bi are available . the revised version of kadonis - p is available online at www.kadonis.org/pprocess ( fig . [ fig1 ] ) since march 2013 .
for a more detailed description , please refer to ref .
the kadonis project ( www.kadonis.org ) is providing astrophysical cross sections for @xmath0- and @xmath10-process nucleosynthesis .
the main aim is to maintain user - friendly and easily accessible compilations of the available ( experimental ) data in the relevant energy ranges .
comparisons to some theoretical predictions are available , as well as to data from evaluated libraries .
this work is supported by the helmholtz association via the young investigators projects vh - ng-627 and vh - ng-327 , the hungarian scientific research funds otka nn83261 ( eurogenesis ) and otka k101328 , the hungarian academy of sciences , the masche collaboration in the esf eurocores program eurogenesis , the thexo collaboration within the 7th framework programme of the eu , and the swiss national science foundation
. 9 m.e .
burbidge , g.r .
burbidge , w.a .
fowler , and f. hoyle , rev .
* 29 * , 547 ( 1957 ) .
et al . _ ,
lett . * 69 * , 2164 ( 1992 ) .
n. klay and f. kppeler , phys .
c * 38 * , 295 ( 1988 ) .
k. takahashi and k. yokoi , nucl .
a * 404 * , 578 ( 1983 ) . c. lederer _
et al . _ ( the n_tof collaboration ) , neutron capture reactions on fe and ni isotopes for the astrophysical s - process , these proceedings .
c. lederer _
et al . _ ( the n_tof collaboration ) , phys .
lett . * 110 * , 022501 ( 2013 ) .
h. nassar _
lett . * 94 * , 092504 ( 2005 ) .
m. wiescher , f. kppeler , and k. langanke , ann .
astrophys . * 50 * , 165 ( 2012 ) .
t. szcs , i. dillmann , r. plag , and zs .
flp , kadonis - p : the astrophysical p - process database , these proceedings .
m. arnould and s. goriely , phys .
rep . * 384 * , 1 ( 2003 ) .
t. rauscher , n. dauphas , i. dillmann , c. frhlich , zs .
flp , and gy .
gyrky , accepted for publication in reports on progress in physics ( 2013 ) ; available at arxiv:1303.2666 .
t. rauscher _ et al .
c * 80 * , 035801 ( 2009 ) .
t. rauscher , int .
e * 20 * , 1071 ( 2011 ) .
b.j . allen _
et al . _ ,
* 4 * , 205 ( 1971 ) .
z.y . bao and f. kppeler , at .
data nucl .
data tables * 36 * , 411 ( 1987 ) .
h. beer , f. voss , and r.r . winters , astrophys . j. suppl . *
80 * , 403 ( 1992 ) .
z.y . bao _
et al . _ , at .
data nucl .
data tables * 76 * , 70 ( 2000 ) .
t. rauscher and f .- k .
thielemann , at .
data nucl .
data tabl .
* 75 * , 1 ( 2000 ) .
i. dillmann , m. heil , f. kppeler _
_ , aip conf . proc . * 819 * , 123 ( 2006 )
. i. dillmann , r. plag , f. kppeler , and t. rauscher , proc .
`` efnudat fast neutrons - scientific workshop on neutron measurements , theory & applications '' , geel/ belgium , p.55 ( 2009 ) .
available online : http://paperc.de/11411-efnudat-fast-neutrons-9789279117053#!/pages/55 t. rauscher , p. mohr , i. dillmann , and r. plag , astrophys .
j. * 738 * , 143 ( 2011 ) .
t. rauscher , astrophys . j. lett . * 755 * , l10 ( 2012 ) .
et al . _ , nucl .
data sheets * 112 * , 2887 ( 2011 ) .
k. shibata , t. kawano , t. nakagawa _ et al .
_ , journal of nuclear science and technology * 48 * , 1 ( 2011 )
. b. pritychenko and s.f . mughabghab , at .
data nucl .
data tables * 113 * , 3120 ( 2012 ) .
t. szcs , i. dillmann , r. plag , and zs .
flp , proceedings of science , pos(nic xi)247 ( 2010 ) .
t. szcs , i. dillmann , r. plag , and zs .
flp , j. phys .
: conf . ser .
* 337 * , 012033 ( 2012 ) .
experimental nuclear reaction data database ( exfor ) , www-nds.iaea.org/exfor/exfor.htm . | the kadonis ( karlsruhe astrophysical database of nucleosynthesis in stars ) project is an astrophysical online database for cross sections relevant for nucleosynthesis in the @xmath0 process and the @xmath1 process .
the @xmath0-process database ( www.kadonis.org ) was started in 2005 and is presently facing its 4th update ( kadonis v1.0 ) .
the @xmath1-process database ( kadonis - p , www.kadonis.org/pprocess ) was recently revised and re - launched in march 2013 .
both databases are compilations for experimental cross sections with relevance to heavy ion nucleosynthesis . for the @xmath0 process
recommended maxwellian averaged cross sections for @xmath2= 5 - 100 kev are given for more than 360 isotopes between @xmath3h and @xmath4bi .
for the @xmath1-process database all available experimental data from @xmath5 , and @xmath6 reactions between @xmath7ge and @xmath8bi in or close to the respective gamow window were collected and can be compared to theoretical predictions .
the aim of both databases is a quick and user - friendly access to the available data in the astrophysically relevant energy regions . |
entanglement is a remarkable feature of quantum mechanics and plays a fundamental role in quantum computation and quantum information processing . due to its fundamental importance for quantum information processing ,
it has attracted much attention in recent years @xcite .
the entanglement can be produced either by direct interactions between qubits @xcite , or by indirect interactions between the qubits through a third party bose .
however , both of the above processes are constrained to the closed system where the influence of the environment is neglected . in real situations , a quantum system can never be isolated and will inevitably interact with its environment . a severe effect of this unwanted interaction is decoherence which generally leads to the degradation of quantum coherence and entanglement .
up to now , decoherence remains to be the main obstacle to the practical implement of quantum information processing .
recently , the influences of the environment on the entanglement have been investigated extensively @xcite .
it was found that in certain cases the environment can act as a third party and be constructive to the production of the entanglement , in contrast to its destructive role in the ordinary situations .
that is to say , depending on different coupling mechanisms between the environment and quantum system in different models , the environment has dual nature influenced on the entanglement between the qubits . in the present work
, we will show that this dual influences can be present simultaneously even in one model under certain given coupling mechanism between the environment and system .
for this purpose , we model a quantum register by two or three two - level atoms with dipole - dipole interactions .
we study its entanglement dynamics ( i.e. the time evolution of entanglement quantity @xcite ) under the influence of a common environment . the direct interactions between the qubits have not been considered in the previous works @xcite .
the environment used here is treated as a reservoir of quantum harmonic oscillators with infinite degrees of freedom @xcite , in contrast to the single - mode model used by kim et al .
@xcite . by introducing a collective mode consisting of all the qubit modes and eliminating the enormous irrelevant degrees of freedom of the environment , a quantum master equation of the atomic system
can be obtained under the born - markovian approximation .
we use the quantum jump approach @xcite to study the time evolution behaviors of bi- and tri - partite entanglement defined by the concurrence @xcite and the negativity @xcite , respectively .
originally , the quantum jump approach @xcite was proposed to study the transition dynamics of a single atom system ( for a review , see , e.g. @xcite ) and has been used to obtain the operator solution of the master equation @xcite .
the merit of using quantum jump approach is that it supplies a simple way to obtain the solution of master equation in the multipartite system with a very large hilbert space .
the dynamical behavior of the bi- and tri - partite entanglement shows clearly that the environment has dual nature influenced on the entanglement in the system studied . on the one hand ,
it destroys entanglement induced by the dipole - dipole interactions , which shows the decoherence nature of the environment . on the other hand ,
it induces stable entanglement in an incoherent way , which shows the third party role of the environment .
different from the previous works @xcite , the dual nature of the environment is shown simultaneously in the same system in our situation .
the stable entanglement induced by the environment is very different from the simple oscillation behavior of entanglement with time obtained by kim _
@xcite , where the memory effect of the single - mode environment was considered . here
we point out that the environment in our model as a reservoir has no memory effect under the markovian approximation .
the environment induces entanglement in an incoherent way since with infinite degrees of freedom it introduces an irreversibility to the dynamics of the qubit system .
the paper is organized as follows . in section
ii we introduce the model and the quantum jump approach . in section
iii we study the entanglement dynamics of a two - qubit system .
the entanglement dynamics of a three - qubit system is discussed in section iv . finally , section v is devoted to a brief summary .
we consider a quantum register composed of @xmath0 ( = 2 and 3 in the following discussions ) identical two - level atoms coupled by dipole - dipole interactions .
these atoms are assumed to be located very near , so that they feel a common environment and interact collectively with the environment .
the environment is modeled by an electromagnetic field with infinite degrees of freedom which is assumed at vacuum state .
if the collective interaction between atoms and the common environment is very weak and the characteristic time of correlation function of the environment is very short compared with the time scale of system evolution , then born and markovian approximations are valid . by the standard method of quantum optics @xcite , the master equation of the reduced density matrix of the system
can be derived @xmath1+\frac{% \gamma } { 2}(2j_{-}\rho \left ( t\right ) j_{+}-j_{+}j_{-}\rho \left ( t\right ) -\rho \left ( t\right ) j_{+}j_{- } ) , \label{master}\]]where @xmath2 is the decay constant of the collective mode represented by @xmath3and @xmath4describes the dipole - dipole interactions of the qubits . the first term on the right - hand side of eq .
( [ master ] ) generates a coherent unitary evolution of the density matrix , while the second term represents the decoherence effect of the environment on the system and generates an incoherent dynamics of system .
it should be noted that in general case the strengths of dipole - dipole interactions are dependent on the positions of the atoms .
but in our model , the separation of the atoms are very small compared to the wavelengths of the most relevant field modes , so that the fields induce a global dipole - dipole interactions(in the long wavelength limit ) , i.e. @xmath5 . thus the interaction hamiltonian becomes @xmath6by using individual mode representation eq .
( [ master ] ) can be rewritten as @xmath1+\frac{% \gamma } { 2}\sum_{i , j=1}^{n}(2\sigma _ { -}^{i}\rho \left ( t\right ) \sigma _ { + } ^{j}-\sigma _ { + } ^{j}\sigma _ { -}^{i}\rho \left ( t\right ) -\rho \left ( t\right ) \sigma _ { + } ^{j}\sigma _ { -}^{i } ) .
\label{masterin}\]]it should be pointed out that the terms with @xmath7 in the sum on the right - hand side of eq .
( [ masterin ] ) denote the individual dissipation of each atom due to the environment , while the @xmath8 terms describe the couplings between the atoms indirectly induced by the common environment .
the master equation ( [ master ] ) has been used for many years to study the superradiance effect involving the interaction of collective atomic systems with the radiation field @xcite . the collective interaction between the atoms and
the radiative field can induce two typical atomic states with an enhanced ( the superradiant state ) and a reduced ( the subradiant state ) spontaneous decay rate .
when the atoms are confined in a very small region , the subradiant state is completely decoupled from the environment , and therefore can be regarded as a decoherence - free state . a related experiment on the superradiant spontaneous emission of two trapped ions
has been reported @xcite .
the common feature of the quantum master equations is the existence of the sandwich terms , where the reduced density matrix of the system is in between some quantum excitation and de - excitation operators .
in general , the master equation is converted into c - number equation @xcite . in this paper
, we use the quantum jump approach to obtain the exact operator solution of the master equation ( [ master ] ) .
as mentioned above , the quantum - jump approach was proposed to describe single experimental realizations of simple quantum systems ( i.e. two- and three - level atoms ) @xcite and has been extensively used in quantum optics plenio98 . by using the quantum jump approach , it is straightforward to obtain the exact operator solution of the master equation .
the advantage of using this method is that one does not need to deal with each element of the reduced density matrix individually , the time - dependent solution of the reduced density matrix can be obtained directly from the formal solution under the actions of operators . by the quantum jump approach
the reduced density matrix @xmath9 can be written as a sum of the conditional density matrices @xmath10 as follows @xmath11where @xmath12 is the total density matrix of the system and the quantized environment , @xmath13 is the partial trace over the modes of the environment , and @xmath14 is the projection operator on the state of the quantized environment field that contains @xmath15 photons .
the @xmath16 denotes the reduced density matrix of the system with @xmath15 photons detected in the environment .
the time evolution equation of @xmath17 can be derived readily , @xmath18where @xmath19 .
( 3 ) reflects that the time evolution of the reduced density matrix consists of smooth evolutions which are interrupted by instantaneous quantum jumps .
it is clear to see that the hierarchy of the conditional density matrices @xmath20 is terminated by instantaneous quantum jumps automatically . due to the dissipative effect of the environment ,
the effective hamiltonian @xmath21 governed the smooth evolution is a non - hermitian operator .
it is straightforward to write down the formal operator solution of eq .
( conequ ) @xmath22where @xmath23 is the initial state .
thus , the total reduced density matrix of the system can be obtained by the summation over all the conditional density operators @xmath17 . in the following
we explicitly consider the two - qubit and three - qubit systems .
to simplify the following calculations , we use the collective state representation @xcite .
the representation space is spanned by the common eigenstates of the complete set of commuting operators @xmath24 for two - qubit system .
this collective state representation can also be obtained by the angular momentum additive rules and the clebsch - gordan coefficients from the product state representation of the individual ones @xcite .
the two - qubit basis denoted by @xmath25 includes the spin singlet and triplet @xmath26where @xmath27 are the eigenstates of @xmath28 with elgenvalus @xmath29 , respectively .
the @xmath30 act on the collective basis by @xmath31 when a total system is composed of two subsystems described by a two - dimensional hilbert state , the general measure of entanglement between these two subsystems is the entanglement of formation @xcite .
this quantity can be analytically calculated as a function of the concurrence @xmath32 @xcite , which can also be taken as a measure of the entanglement .
the concurrence is defined as @xmath33 where @xmath34 are eigenvalues of the matrix @xmath35^{1/2}$ ] . here
@xmath36 is the complex conjugation of @xmath37 .
the concurrence varies from @xmath38 for a disentangled state to @xmath39 for a maximally entangled state .
we shall use the concurrence to quantify the degree of entanglement in the two - qubit system . in the following numerical investigations ,
our guide line is to explore the following topics of entanglement dynamics : \(1 ) two kinds of entanglement generation mechanisms and their different roles played : direct coherent generation due to the atomic dipole - dipole interaction and indirect incoherent generation due to the common environment coupling to the qubits . from eq .
( [ master ] ) we can see that the atomic dipole - dipole interactions in @xmath40 produce a dynamical coherence which oscillates in time , and the common environment induces an effective interaction which produces the entanglement in steady states .
besides , the common environment also produces a dissipative term which makes all the eigen states of the operator of the master equation decay and finally die out except the steady states with zero eigen values @xcite .
thus in the dissipative dynamics , the dynamical coherence as a transit phenomenon will be destroyed finally by the dissipation , only the environment induced coherence and entanglement in the decoherence - free steady states survive . in the following numerical study ,
the two kinds of entanglement generation show clearly in the figures .
\(2 ) how does the environment produce an entangled steady state from a separable initial state ?
this issue is closely related to the generation of an entanglement source and is thus very significant .
\(3 ) how does the entanglement measure evolve and change in time from an initial entangled state under the influence of the common environment ?
this issue is related to the decoherence of a multi - qubit system and is thus practically helpful in the understanding of the decoherence dynamics .
\(4 ) as the decoherence - free subspace has more than one dimensions , the steady state will be a probabilistic mixture of the decoherence - free states .
how does one choose a proper initial state to get a desired steady state ? in order to show explicitly the entanglement dynamics of the two - qubit system , we take the initial state @xmath41 as our first example . expanding this state in the collective basis of eqs .
( twobas ) and substituting it into eqs .
( [ solu ] ) , we can obtain analytically the solution of the master equation ( [ master ] ) after some calculations @xmath42the time evolution behavior of the entanglement measure , i.e. the concurrence , can be determined by @xmath43 from eqs .
( [ solut ] ) and ( [ tconcu ] ) we can see that the steady solution of the atomic system under the long time limit is @xmath44 which is an equal probabilistic mixture of the extremal states @xmath45 and an entangled mixed state with @xmath46 .
the extremal states @xmath47 is the so - called subradiant state @xcite and maximal entangled pure state .
this means that due to the symmetry of the interaction between the system and the environment , the system induces a collective entangled state which is immune to the decoherence .
this entangled decoherence - free state contributes to the stable entanglement produced by the environment . from the purification scheme proposed in ref .
@xcite many pairs of this states can be used to distill out a maximal entangled state @xmath48 with probability 1/16 .
the time evolution behavior of the concurrence is showed in fig .
we consider three cases of the dipole - dipole interaction : absence ( @xmath50 , dotted line ) , weak ( @xmath51 , dashed line ) , and strong ( @xmath52 , solid line ) . in the case of @xmath50 ,
the entanglement increases monotonously and approaches to a stable value .
since there are no dipole - dipole interaction between the qubits , it is quite clear that this entanglement is induced purely by the environment .
this is a typical example that the environment plays a constructive role in the entanglement formation between qubits .
when the dipole - dipole interaction is switched on , the entanglement induced by the environment is not affected but an oscillating dynamical contribution added to the total entanglement is observed .
the stronger the dipole - dipole interaction is , the faster the entanglement oscillates , as seen easily from eq . ( [ tconcu ] ) . to see the contribution of the dipole - dipole interaction to the entanglement , in fig .
[ purec2 ] we plot the quantity of @xmath53 as a function of time .
it is shown that the entanglement induced by the dipole - dipole interaction decays rapidly as shown by solid line , which indicates that the environment plays a destructive role to the entanglement induced by the coherent interaction . to confirm this
, we switch off the environment , i.e. , @xmath54 . in this situation , the entanglement induced by the coherent interaction oscillates without dissipation , as the dashed line shown in fig .
[ purec2 ] . to summarize ,
the environment can act as a third party to induce the entanglement between the qubits incoherently .
it also plays an usual role of dissipation to destroy the entanglement induced coherently by the dipole - dipole interaction .
the dual nature of both construction and destruction roles played by the environment on the entanglement generation can be seen clearly by the above numerical analysis . to explore the dependence of the steady state entanglement on initial states , in the following we take the werner states as the initial states .
the werner states read @xmath55where @xmath56 is a @xmath57 identity matrix , @xmath58 and @xmath59 defined by eq .
( [ twobas ] ) , and @xmath60 $ ] denotes the fidelity of @xmath61 to @xmath62 .
it is known that @xmath63 are entangled states for @xmath64 and @xmath65 . in a similar way
, one can obtain the time evolution behavior of @xmath66 as @xmath67}{4}|1,-1\rangle \langle 1,-1| , \notag \\ & & \rho _
{ -}(t)=\frac{(1-p)e^{-2\gamma t}}{4}|1,1\rangle \langle 1,1|+\frac{% ( 1-p)e^{-2\gamma t}(1 + 2\gamma t)}{4}|1,0\rangle \langle 1,0| \notag \\ & & + \frac{(1-p)[3 - 2e^{-2\gamma t}(1+\gamma t)]}{4}|1,-1\rangle \langle 1,-1|+% \frac{1 + 3p}{4}|0,0\rangle \langle 0,0| .
\label{sow}\end{aligned}\]]because only the diagonal forms of collective states are involved in the werner states ( [ werner ] ) , the coherent interaction @xmath40 of eq .
( master ) has no actions on the werner states .
the concurrence can be calculated in a similar way . before discussing the dynamical behavior of the concurrence , it is useful to consider the long time behavior of the concurrence . with @xmath68 , @xmath69
become @xmath70with the steady entanglements as @xmath71 and @xmath72 , respectively .
the above results show that different initial states lead to different steady states with different entanglements . also from the limit case with @xmath73 one
can see that @xmath74 is sensitive to the decoherence , while @xmath75 is not .
below we discuss entanglement dynamics in the general case .
fig . @xmath76 and fig .
@xmath77 show the concurrences of @xmath78 as functions of @xmath79 and @xmath80 .
when @xmath81 , a common characteristic of @xmath82 is that they have no entanglement initially . with time
a stable entanglement is formed through the environment .
when @xmath64 , the concurrences of @xmath78 show apparently different time evolution behaviors .
for @xmath83 , the entanglement contained in the initial state @xmath84 is firstly destroyed by the environment due to the decoherence effect of the environment .
the entanglement experiences a sudden death@xcite . after a finite time duration
the entanglement is reconstructed due to the constructive effect of the environment .
physically , the initial contribution to the entanglement in @xmath85 comes from the component of @xmath86 . with time
, @xmath87 ( i.e. @xmath88 ) approaches to the product state @xmath89 is unchanged and the corresponding entanglement gradually dominates the entanglement of @xmath90 and finally , it reaches the steady state value in the long time limit , i.e. , @xmath91 , as mentioned above .
different from @xmath83 , the initial entanglement in @xmath92 does not decay .
this is because the main contribution to the initial entanglement in this case comes from the decoherence - free state @xmath93 .
the entanglement induced by the environment is superposed on the initial entanglement .
when @xmath73 , the werner state reduces to the pure state @xmath93 which is independent of time and free from decoherence . from the special and general analysis
above , one sees that the entanglement of the steady state induced by the environment is rooted in the seeds of the entangled decoherence - free states which as components have already been contained in the initial states .
this property can be used in choosing a proper initial state to obtain a desired entangled steady state .
the three - qubit basis can be constructed from the addition of angular momenta @xmath94 and @xmath95 .
the representation space is spanned by the common eigenstates of the complete set of commuting operators @xmath96 .
the collective basis @xmath97 can be obtained by virtue of the corresponding clebsch - gordan coefficients @xcite @xmath98the @xmath30 act on the collective basis as follows @xmath99which are independent of the quantum numbers @xmath100 .
to discuss the entanglement dynamics in the three - qubit system , we use the negativity proposed by vidal and werner to quantify the degree of entanglement @xcite .
the idea of this measure of the entanglement comes from the peres - horodecki criterion for the separability of bipartite systems @xcite .
the negativity was originally introduced to an arbitrary two - qubit state @xmath37 and defined as @xcite @xmath101where @xmath102 is the negative eigenvalue of the partial transpose of @xmath103 with respect to the @xmath104th qubit , i.e. @xmath105 .
given a bipartite state one can calculate the partial transpose @xmath105 of the density operator .
the state is exactly separable if @xmath105 is again a positive operator .
however , if one of the eigenvalues of @xmath106 is negative , the state is entangled @xcite . from this viewpoint ,
the negativity is used to quantify the degree that @xmath105 fails to be positive and to represent the strength of quantum correlation between the two subsystems .
it was proved that the relation between @xmath107 and @xmath108 is @xmath109 $ ] for any two - qubit state @xcite .
the merit of using the negativity to quantify the entanglement is that it allows us to investigate the entanglement properties between part @xmath110 and the sum of other components in the multipartite system .
similarly , we choose an explicit initial state , e.g. , @xmath111 to discuss the entanglement dynamics in this case . expanding this initial state in terms of the coupled basis of eqs .
( [ bas ] ) @xmath112and using the quantum jump approach , the time - dependent solution of the master equation can be obtained analytically @xmath113 \notag \\ & & + \frac{e^{-\frac{5\gamma t}{2}}}{\sqrt{6}}[e^{-i3gt}|1,\frac{3}{2},\frac{1% } { 2}\rangle \langle 0,\frac{1}{2},\frac{1}{2}|+e^{i3gt}|0,\frac{1}{2},\frac{1% } { 2}\rangle \langle 1,\frac{3}{2},\frac{1}{2}| ] \notag \\ & & -\frac{e^{-\gamma t}}{\sqrt{12}}[|1,\frac{1}{2},\frac{1}{2}\rangle \langle 0,\frac{1}{2},\frac{1}{2}|+|0,\frac{1}{2},\frac{1}{2}\rangle \langle 1,\frac{% 1}{2},\frac{1}{2}|]+\frac{1-e^{-\gamma t}}{2}|0,\frac{1}{2},\frac{-1}{2}% \rangle \langle 0,\frac{1}{2},\frac{-1}{2}| \notag \\ & & + \frac{4(e^{\gamma t}-1)e^{-4\gamma t}}{3}|1,\frac{3}{2},\frac{-1}{2}% \rangle \langle 1,\frac{3}{2},\frac{-1}{2}|+\frac{1-e^{-\gamma t}}{6}|1,% \frac{1}{2},\frac{-1}{2}\rangle \langle 1,\frac{1}{2},\frac{-1}{2}| \notag \\ & & -\frac{2e^{-\frac{5\gamma t}{2}}(e^{\gamma t}-1)}{\sqrt{18}}[e^{-i3gt}|1,% \frac{3}{2},\frac{-1}{2}\rangle \langle 1,\frac{1}{2},\frac{-1}{2}% 2}| ] \notag \\ & & + \frac{2e^{-\frac{5\gamma t}{2}}(e^{\gamma t}-1)}{\sqrt{6}}[e^{-i3gt}|1,% \frac{3}{2},\frac{-1}{2}\rangle \langle 0,\frac{1}{2},\frac{-1}{2}% 2}| ] \notag \\ & & -\frac{1-e^{-\gamma t}}{\sqrt{12}}[|1,\frac{1}{2},\frac{-1}{2}\rangle \langle 0,\frac{1}{2},\frac{-1}{2}|+|0,\frac{1}{2},\frac{-1}{2}\rangle \langle 1,\frac{1}{2},\frac{-1}{2}| ] \notag \\ & & + \frac{1 + 3e^{-4\gamma t}-4e^{-3\gamma t}}{3 } |1,\frac{3}{2},\frac{-3}{2}% \rangle \langle 1,\frac{3}{2},\frac{-3}{2}| .
\label{threeso}\end{aligned}\ ] ] in fig . @xmath114 and fig .
@xmath115 we show the time evolution behavior of the negativity @xmath116 and @xmath117 corresponding to the partial transpose with respect to the first atom @xmath118 and second atom @xmath119 , respectively . in the present case , @xmath120 where @xmath32 denotes the third atom .
one can see that the system shows a similar property of entanglement production as the two - qubit system . in absence of the direct interactions ,
the entanglement induced by the environment increases monotonously , then approaches to a stable value at a large time scale for both @xmath116 and @xmath117 .
after the interactions are switched on , the oscillation occurs .
moreover , during the evolution of time , this oscillation is suppressed gradually by the environment . from eq .
( @xmath121 ) it is readily seen that the time - dependent solution asymptotically tends to the steady state @xmath122which is a probabilistic mixture of decoherence - free state @xmath123 and @xmath124 @xcite . in the present three - qubit case ,
there are two entangled decoherence - free states which contribute to the entanglement of the steady state , i.e. @xmath125 and @xmath126 .
it is interesting to note that @xmath127 is not a simple mixture of three decoherence - free states .
it still involves quantum coherence between @xmath125 and @xmath126 . by comparison of the initial state eqs .
( [ trein ] ) with the steady state eq .
( [ steadt ] ) , we see that although the jump has occurred , i.e. from latexmath:[$% @xmath124 by the action of @xmath129 , the superposition coefficients between @xmath130 and @xmath131 are not changed .
this is because that the subspaces of @xmath132 and @xmath133 are degenerate under the actions of both the coherent interaction @xmath40 and the dissipative operators @xmath30 .
anyway , the steady state @xmath127 is still an entangled state due to the entangled decoherence - free states contained in @xmath124 , which is a linear superposition of the two entangled decoherence - free states .
the above discussion in the three - qubit case further confirms the conclusion that the environment has dual nature on the entanglement .
different from the two - qubit case , the hilbert space of the three - qubit system reduces into three decoupled subspaces @xmath134 , @xmath130 , and @xmath131 , each of which has its own decoherence - free state @xcite .
only two of the decoherence - free states are entangled . under the time evolution , each component in the three subspaces of the initial state approaches its corresponding decoherence - free state .
the necessary condition to induce stable entanglement is that the initial state contains the components in the two subspaces with entangled decoherence - free states , i.e. in the subspaces of @xmath130 and @xmath131 .
in conclusion , we have investigated a model of quantum register , which is composed of two or three atoms and coupled to a common environment . using the quantum jump approach ,
the time - dependent solution of the master equation and the entanglement dynamics of the system are studied analytically and numerically .
the dual nature of the common environment on the entanglement and its dissipative dynamical origin are explored in detail based on the eigen solutions of the operator of master equation . on the one hand , due to its induced dissipative term in the master equation ,
the environment destroys the entanglement induced coherently by the atomic dipole - dipole interactions . on the other hand , due to its induced effective atomic interactions in the master equation , the environment can incoherently induce the entanglement among qubits in the decoherence - free space .
the entanglement dynamics studied in the present paper addresses the constructive role of the environment on the entanglement production and suggests to make use of the above positive role to construct an environment - assisted entanglement production in quantum information processing .
j.h.a . and s.j.w are grateful to the center for theoretical sciences at cheng kung university and professor w.m .
zhang for their kind hospitality during the visit .
thanks the financial supports of the fundamental research fund for physics and mathematic of lanzhou university under grant no lzu05 - 02 and the nsc grant no .
95 - 2816-m-006 - 001 .
the work is also partially supported by the nnsf of china under grants no 10604025 , 10375039 , and 90503008 . | we use the quantum jump approach to study the entanglement dynamics of a quantum register , which is composed of two or three dipole - dipole coupled two - level atoms , interacting with a common environment .
our investigation of entanglement dynamics reflects that the environment has dual actions on the entanglement of the qubits in the model . while the environment destroys the entanglement induced by the coherent dipole - dipole interactions , it can produce stable entanglement between the qubits prepared initially in a separable state .
the analysis shows that it is the entangled decoherence - free states contained as components in the initial state that contribute to the stable entanglement .
our study indicates how the environmental noise produces the entanglement and exposes the interplay of environmental noise and coherent interactions of qubits on the entanglement . and and quantum jump approach , master equation , entanglement dynamics 03.65.ud , 03.67.-a |
network calculus provides an elegant way to characterize traffic and service processes of network and communication systems . unlike traditional queueing analysis
in which one has to make strong assumptions on arrival or service processes ( e.g. , poission arrival process , exponential service distribution , etc ) so as to derive closed - form solutions@xcite , network calculus allows general arrival and service processes . instead of getting exact solutions , one can derive network delay and backlog bounds easily by network calculus .
deterministic network calculus was proposed in @xcite @xcite@xcite@xcite , etc .
however , most traffic and service processes are stochastic and deterministic network calculus is often not applicable for them .
therefore , stochastic network calculus was proposed to deal with stochastic arrival and service processes @xcite@xcite@xcite@xcite@xcite@xcite@xcite .
there have been some applications of stochastic network calculus@xcite@xcite@xcite@xcite .
however , little effort has been made on applying it to multi - access communication systems . in the paper , we take the first step to apply stochastic network calculus to an 802.11 wireless local network ( wlan ) . in particular , we address the following questions : * under what situations can we derive stable backlog and delay bounds ? * how to derive the backlog and delay bounds of an 802.11 wireless node ? *
how tight are these bounds when compared with simulations ? in this paper , we answer these questions and make the following contributions : * we derive the general stability condition of a wireless node based on the theorems of stochastic network calculus . from this
, we give the specific stability condition of an 802.11 wireless node . *
we derive the service curve of an 802.11 node based on an existing model of 802.11@xcite . from the service curve
, we then derive the backlog and delay bounds of the node .
* the derived bounds are loose in many cases when compared with ns-2 simulations .
we discuss the reasons and point out future work .
this paper is organized as follows . in section [ sec : snc ] , we give a brief overview of stochastic network calculus . in section [ sec : model ] , we present the stochastic network calculus model of a wireless node . in section [ sec : stab ] , we derive the general stability condition of a wireless node . in section [ sec:802_11 ]
, we derive the backlog and delay bounds and the stability condition for an 802.11 node . in section [ sec :
simulation ] , we compare the derived bounds with simulation results .
related work is given in section [ sec : related ] and finally , section [ sec : conclusion ] concludes the paper and points out future directions .
in this section , we first review basic terms of network calculus and then cite the results of stochastic network calculus which we will use in this paper .
there are various versions of arrival and service curves .
we adopt _ virtual backlog centric ( v.b.c ) stochastic arrival curve _ and _ weak stochastic service curve _ in our analysis .
we consider a discrete time system where time is slotted ( @xmath0 ) .
a process is a function of time @xmath1 . by default
, we use @xmath2 to denote the _ arrival process _ to a network element with @xmath3 .
@xmath2 is the total amount of traffic arrived to this network element up to time @xmath1 .
we use @xmath4 to denote the _ departure process _ of the network element with @xmath5 .
@xmath4 is the total amount of traffic departed from the network element up to time @xmath1 .
let @xmath6 ( @xmath7 ) represents the set of non - negative wide - sense increasing ( decreasing ) functions . clearly , @xmath8 and @xmath9 .
for any process , say @xmath2 , we define @xmath10 , for @xmath11 .
we define the backlog of the network element at time @xmath1 by @xmath12 and the delay of the network element at @xmath1 by @xmath13 fig .
[ fig : curves_eg ] illustrates an example of @xmath2 and @xmath4 with @xmath14 and @xmath15 at @xmath16 . ,
@xmath4 , @xmath14 and @xmath15,title="fig:",scaledwidth=80.0% ] + in deterministic network calculus , @xmath2 can be upper - bounded by an arrival curve .
that is , for all @xmath17 , we have @xmath18 where @xmath19 is called the _ arrival curve _ of @xmath2 . a _ busy period _ is a time period during which the backlog in the network element is always nonzero . for any busy period
@xmath20 $ ] , suppose we have @xmath21 which means that the network element provides a guaranteed service lower - bounded by @xmath22 during the busy period .
we can let @xmath23 be the beginning of the busy period , that is , the backlog at @xmath23 is zero or @xmath24 .
therefore , @xmath25 the above equation infers @xmath26}$ ] , which can be written as @xmath27 where @xmath28 is called the operator of _ min - plus convolution _ and @xmath29 is called the _ service curve _ of the network element .
we cite the following definitions and theorems from @xcite@xcite except that we define definition [ def : stability ] by ourselves .
[ def : ac ] a flow is said to have a virtual - backlog - centric ( v.b.c ) stochastic arrival curve @xmath30 with bounding function @xmath31 , denoted by @xmath32 , if for all @xmath33 and all @xmath34 , there holds @xmath35>x\ } \leq f(x).\end{aligned}\ ] ] originally , in deterministic network calculus , we have @xmath36 for all @xmath37 .
however , there is usually some randomness in stochastic arrival processes and @xmath38 may not be upper - bounded by any arrival curve deterministically ( e.g. , traffic arrivals in @xmath39 $ ] can be arbitrarily large in poisson process ) .
thus , v.b.c stochastic arrival curve is proposed to tackle this problem .
roughly speaking , @xmath38 can exceed @xmath40 by @xmath41 , but its probability is upper - bounded by @xmath42 which is a decreasing function of @xmath41 .
[ def : sc ] a server @xmath43 is said to provide a weak stochastic service curve @xmath44 with bounding function @xmath45 , denoted by @xmath46 , if for all @xmath33 and all @xmath34 , there holds @xmath47 in deterministic network calculus , we have @xmath48 , which means that there is a service guarantee denoted by the service curve @xmath29 .
however , there is usually some randomness in stochastic service process and thus a server may not always provide a guaranteed service curve deterministically .
thus , weak stochastic service curve is proposed to tackle this problem .
roughly speaking , @xmath49 can be less than @xmath50 , but its probability is upper - bounded by @xmath51 which is a decreasing function of @xmath41 .
the utility of the above definitions is that if we can characterize the traffic by a v.b.c stochastic arrival curve and the server s service process by a weak stochastic service curve , then we can calculate backlog and delay bounds of the network element by theorem [ theo : backlog_delay ] .
[ theo : backlog_delay ] consider a server fed with a flow @xmath52 .
if the server provides a weak stochastic service curve @xmath46 to the flow and the flow has a v.b.c stochastic arrival curve @xmath53 , then \(i ) the backlog @xmath14 of the flow in the server at time @xmath1 satisfies : for all @xmath33 and all @xmath34 , @xmath54}).\end{aligned}\ ] ] \(ii ) the delay @xmath15 of the flow in the server at time @xmath1 satisfies : for all @xmath33 and all @xmath34 , @xmath55}).\end{aligned}\ ] ] by definition @xmath56 when @xmath57 in this theorem .
note that as noticed recently by researchers of network calculus , the formula of delay bound in this theorem often returns trivial results , which we will see in section [ sec:802_11 ] .
by now , we have reviewed the key results of stochastic network calculus .
next , we will show how to calculate v.b.c stochastic arrival curve and weak stochastic service curve .
in @xcite , the author presented a theorem to facilitate calculation of stochastic arrival curves . before showing the theorem
, we first introduce @xmath58__-upper constrained _ _ @xcite .
[ def : theta_uc ] a process @xmath52 is said to be @xmath59-upper constrained ( for some @xmath60 ) , if for all @xmath17 , we have @xmath61 this definition is equivalent to @xmath62 , which means that @xmath38 s moment generating function is upper - bounded .
two related concepts are defined as follows .
[ def : theta - mer ] a process @xmath52 s _ minimum envelope rate ( mer ) with respect to @xmath63 _ ( @xmath63-mer ) , denoted by @xmath64 , is defined as follows : @xmath65 we say that @xmath52 has an _ envelope rate ( er ) with respect to @xmath63 _ ( @xmath63-er ) , denoted by @xmath66 , if @xmath67 .
the following theorem expresses the relationship between @xmath63-er and @xmath59-upper constrained .
[ theo : relation_theta ] + ( i ) if @xmath52 is @xmath59-upper constrained , then @xmath66 is @xmath63-er of @xmath52 .
+ ( ii ) if @xmath52 has @xmath63-er @xmath68 , then for every @xmath69 there exists @xmath70 so that @xmath52 is @xmath71-upper constrained .
now we have two kinds of traffic characterization : v.b.c stochastic arrival curve and @xmath59-upper constrained .
the following theorem establishes the connection between them .
[ theo : ac_theta ] suppose @xmath2 is @xmath72-upper constrained , then it has a v.b.c stochastic arrival curve - upper constrained , which is wrong as noticed recently by researchers of network calculus .
however , one can easily see the theorem holds for v.b.c stochastic arrival curve . ]
@xmath73 , where @xmath74 for any @xmath75 and @xmath34 .
this theorem indicates that if we can show that the traffic is @xmath59-upper constrained , then we can get its v.b.c stochastic arrival curve by eq .
( [ eq : theo_ac_theta ] ) .
we now introduce the concept of _ stochastic strict server_. this concept was inspired by the observation that a wireless channel can be described by an ideal service process and an impairment process .
as we will see in section [ sec : model ] , a wireless node can be modeled as a stochastic strict server .
a server @xmath43 is said to be a stochastic strict server providing stochastic strict service curve @xmath76 with impairment process @xmath77 to a flow iff during any backlogged period @xmath39 $ ] , the output @xmath78 of the flow from the server satisfies @xmath79 we can easily find the weak stochastic service curve of a stochastic strict server by the following theorem .
[ theo : strict_sc ] consider a stochastic strict server @xmath43 providing a stochastic strict service curve @xmath80 with an impairment process @xmath77 .
if the impairment process has a v.b.c stochastic arrival curve , or @xmath81 , and @xmath82 , then the server provides a weak stochastic service curve @xmath46 with @xmath83 so far , we have cited all results of stochastic network calculus which we will use in this paper .
finally , we define stable backlog and stable delay .
a natural definition is to check whether the expectation of backlog ( or delay ) is finite .
[ def : stability ] the backlog @xmath14 is stable , if @xmath84 similarly , the delay @xmath15 is stable , if @xmath85 we say that the backlog ( or delay ) bound of stochastic network calculus is stable if they can derive stable backlog ( or delay ) .
in this section we model a wireless node ( not restricted to 802.11 ) by stochastic network calculus . in general , we can define one slot ( @xmath86 ) to be any duration of time and measure traffic amount in any unit ( e.g. bits , bytes or packets ) .
we consider a wireless node .
let @xmath2 denote the traffic arrived at the node from the application layer .
suppose @xmath52 is @xmath87-upper constrained . from theorem [ theo : ac_theta ] , we have @xmath88 , where @xmath89 for any @xmath90 .
we can model a wireless node by a stochastic strict server .
let the channel capacity be @xmath91 traffic unit per slot .
the departure process @xmath92 during any backlogged period @xmath93 $ ] , where @xmath94 is the ideal service curve and @xmath77 is the impairment process due to backoff , channel sharing and transmission errors .
since @xmath95 , @xmath77 has a finite @xmath63-mer . from theorem [ theo : relation_theta ] , there exist @xmath96 and @xmath97 so that @xmath77 is @xmath98-upper constrained .
based on theorem [ theo : ac_theta ] , we have @xmath99 , where @xmath100 for any @xmath101 . from theorem [ theo : strict_sc ] , the node provides a weak stochastic service curve @xmath46 , where @xmath102 for any @xmath103 .
furthermore , from theorem [ theo : backlog_delay ] , we must have @xmath104 , or equivalently , @xmath105 thus , @xmath106 . otherwise , if @xmath107 , we get a trivial backlog bound , @xmath108 .
one fundamental question we need to address is under what condition we can get _ stable _ @xmath14 and @xmath15 from stochastic network calculus , i.e. , @xmath109 and @xmath110 . before presenting our result
, we first define the concept of _ envelop average rate_. [ def : er ] the envelop average rate of a process @xmath52 , denoted by @xmath111 , is defined as @xmath112 let @xmath113 and @xmath114 be the envelop average rate of @xmath52 and @xmath77 , respectively .
the following proposition shows the stability condition .
[ prop : stab ] a wireless node has stable backlog and stable delay if @xmath115 _ proof : _ we have shown that @xmath116 if eq .
( [ eq : stab_1 ] ) holds .
thus , for any @xmath1 , @xmath117 since @xmath42 and @xmath51 are exponentially decreasing functions according to eq .
( [ eq : a_ac ] ) and eq .
( [ eq : i_ac ] ) , @xmath118 is an exponentially decreasing function .
thus we have , for any @xmath1 , @xmath109 .
it is easy to see that for any @xmath1 , @xmath110 .
otherwise , the service time is @xmath119 and thus @xmath120 which contradicts eq .
( [ eq : eb ] ) .
now we examine eq .
( [ eq : stab_1 ] ) . from eq .
( [ eq : a_ac ] ) and eq .
( [ eq : i_ac ] ) , @xmath121 and @xmath122 for any @xmath123 .
thus , eq . ( [ eq : stab_1 ] ) is equivalent to @xmath124 from theorem [ theo : relation_theta ] , we have @xmath125 and @xmath126 for any @xmath127 , where @xmath128 and @xmath129 are @xmath63-mers of @xmath52 and @xmath77 , respectively .
equivalently , we have @xmath130 using taylor expansion on @xmath128 , we have @xmath131}}}.\end{aligned}\ ] ] therefore , @xmath132 similarly , @xmath133 therefore , there exist @xmath134 and @xmath135 so that @xmath136 and @xmath137 for any enough small @xmath138 .
( [ eq : stab_4 ] ) is satisfied if @xmath139 since @xmath140 , @xmath141 and @xmath142 can be arbitrarily small , the above equation is satisfied when @xmath143 + * remarks : * since the proof is based on theorems of stochastic network calculus , it indicates that we can get stable backlog and delay bounds by stochastic network calculus as long as the condition of eq .
( [ eq : stab ] ) holds .
since this condition is very general , we conclude that theoretically stochastic network calculus is effective .
in this section , we apply the results in the previous section to calculate backlog and delay bounds for an 802.11 wlan node . for simplicity , we assume there are @xmath144 _ identical _ stations ( or nodes ) sending packets to an access point .
all nodes operate in distributed coordination function ( dcf ) mode with rts / cts turned off@xcite .
we consider an ideal channel , that is , transmission errors are only caused by collisions .
two packets are collided if their transmissions overlap in time . besides , we assume that all data packets are of the same size . a node with a data packet ( or simply packet ) to transmit first monitors the channel activity .
if the channel is idle for a period of time equal to a distributed interframe space ( difs ) , the node transmits .
otherwise , if the channel is sensed busy ( either immediately or during the difs ) , the node backs off , in which the node defers channel access by a random number of _ idle slots _ within a contention window ( @xmath145 ) , ranging from 0 to @xmath146 .
when the backoff counter reaches zero and expires , the node can access the channel . during the backoff period ,
if the node detects a busy channel , it freezes the backoff counter and the backoff process is resumed once the channel is idle for a duration of difs . to avoid channel capture ,
a node must wait a random backoff time between two consecutive new packet transmissions , even if the channel is sensed idle for a duration of difs .
once the packet is received successfully , the receiver will return an ack after a short interframe space ( sifs ) .
note that sifs is shorter than an idle slot so that there is no collision caused by a data packet and an ack .
802.11 uses the truncated exponential backoff technique to set its @xmath145 .
for example , in 802.11b , the initial @xmath145 is @xmath147 . each time
a collision occurs , @xmath145 doubles its size , up to a maximum of @xmath148 .
when the packet is successfully transmitted , @xmath145 is reset to @xmath149 .
the packet is dropped when it is retransmitted for _ six _ times and still not transmitted successfully .
[ fig:802_11 ] shows some parameters of 802.11b used in our paper .
[ cols= " < , < " , ] we illustrate the above calculations in scenario 1 ( see fig . [
fig : scen1 ] ) . from eq .
( [ eq : tau_gamma ] ) and ( [ eq : gamma_tau ] ) , @xmath150 and @xmath151 .
thus , @xmath152 , @xmath153 , @xmath154 and @xmath155 .
let @xmath156 be the value of eq .
( [ eq : ei ] ) .
[ fig : ex_mt ] shows @xmath156 and its @xmath157-upper constrained curve when @xmath158 .
we have @xmath159 and @xmath160 , as calculated by algorithm 1 .
-upper constrained curve of a node s @xmath156,title="fig:",scaledwidth=60.0% ] + furthermore , from eq .
( [ eq : i_ac ] ) and ( [ eq : s_sc ] ) we have @xmath161 where @xmath162 for any @xmath163 . fig .
[ fig : ex_sc ] plots @xmath29 and @xmath164 when @xmath165 .
+ before calculating the bounds , we first derive the stability condition . since @xmath166 , the channel capacity @xmath167 packet per slot . because @xmath168 is actually the percentage of the node s successful transmission time , by proposition [ prop : stab ] ,
the _ stability condition of 802.11 _ is @xmath169 we can immediately calculate backlog bounds by applying eq .
( [ eq : a_ac])-([eq : s_sc ] ) into theorem [ theo : backlog_delay ] .
the only technical issue is to select proper @xmath134 and @xmath135 a obtain tight bounds .
clearly , @xmath42 depends on @xmath134 and @xmath170 , and @xmath51 depends on @xmath135 and @xmath171 . according to eq .
( [ eq : stab_1 ] ) , we have @xmath172 and @xmath173 . in addition , @xmath170 ( @xmath171 ) should be set as large as possible because @xmath42 ( @xmath51 ) decreases with @xmath170 ( @xmath171 ) .
considering the above conditions , we have @xmath174\nonumber\\ & & \textnormal{subject to}\nonumber\\ & & r_a > \rho_a(\theta_1 ) , r_i > \rho_i(\theta_2)\
\textnormal{and}\ r_a + r_i = 1.\end{aligned}\ ] ] in general , we do not have an analytical solution of @xmath175 $ ] and we resort to numerical methods and use algorithm 2 to get a near - optimal solution ( see appendix a-2 ) . as noticed recently by researchers of network calculus ,
the delay bound in theorem [ theo : backlog_delay ] often returns trivial results . in our model in section [ sec : model ] , it is easy to see that @xmath176 .
we propose the following way to estimate delay bound .
little s law states that the average number of customers in a queueing system is equal to the average arrival rate of customers to that system , times the average time spent in that system@xcite .
let the average arrival rate is @xmath177 . assume the system can reach _ steady state _
when @xmath178 .
then we have the average backlog is @xmath179 and the average delay of each packet is greater than or equal to @xmath180 ( by its definition in eq .
( [ eq : delay ] ) , @xmath15 can be less than the delay of the bottom - of - line packet at @xmath1 ) .
therefore , by little s law , we have @xmath181 finally , we apply markov s inequality to the above equation and we have @xmath182 besides , according to eq .
( [ eq : eb ] ) , @xmath183 . and
we can use eq .
( [ eq : my_backlog_bound ] ) to bound @xmath184 .
note that eq .
( [ eq : my_delay_bound ] ) is derived when @xmath185 . in practice
, we can use this result to estimate delay bound when @xmath1 is sufficiently large .
in this section , we use ns-2 simulations to verify our derived backlog and delay bounds for poisson and constant bit rate ( cbr ) traffic arrivals .
we carry out all experiments for scenario 1 ( fig .
[ fig : scen1 ] ) .
each simulation duration is 100 seconds ( _ s _ ) which is long enough to let a node transmit thousands of packets .
each data point ( e.g. @xmath186 ) is calculated over 100 independent simulations .
let @xmath177 be the average traffic rate ( packets / slot ) .
in this case , we have @xmath187 ( see definition [ def : er ] ) . for poisson traffic , we have @xmath188 where @xmath189 is the probability of @xmath190 packets arriving within @xmath191 $ ] . from the above equation ,
poisson traffic is @xmath192-upper constrained and we can obtain its arrival curve by eq .
( [ eq : a_ac ] ) . from eq .
( [ eq : stab_802_11 ] ) , backlogs are stable when @xmath193 ( packet per slot ) . in fig .
[ fig : plot_10_stable_poisson ] , we plot the average backlog @xmath194 $ ] at @xmath195 and @xmath196 , @xmath197 and @xmath198 in ns-2 simulations .
we observe that there is sudden jump when @xmath199 , indicating the critical point of stability is indeed around @xmath197 .
this figure also indicates the accuracy of the 802.11 model in eq .
( [ eq : tau_gamma ] ) and eq .
( [ eq : gamma_tau ] ) .
( @xmath200s ) when @xmath201,title="fig:",scaledwidth=60.0% ] + * experiment 1*(scenario 1 with low poisson traffic load ) we set @xmath202 to simulate low traffic load .
[ fig : plot_10_poisson_low](a ) shows @xmath186 in ns-2 simulations and fig .
[ fig : plot_10_poisson_low](b ) shows the upper bound of @xmath186 calculated by eq .
( [ eq : my_backlog_bound ] ) .
note that stochastic network calculus gives very loose upper bounds
. there may be two reasons .
one is that we use the worst - case analysis in deriving the weak stochastic service curve of 802.11 .
the other is that there are many relaxations in proving the theorems of stochastic network calculus@xcite .
for example , relaxations are used in deriving @xmath42 in theorem [ theo : ac_theta ] @xcite , and this theorem is popularly used in deriving arrival curves and service curves ( see section [ sec : model ] ) .
the first reason may not be the key reason because we will see in experiment 2 the bound is even looser when we increase arrival rate and make the channel near saturated .
the second reason seems to be the key reason .
we will see in experiment 3 that backlog bounds improve substantially for cbr traffic where we are able to derive the arrival curve by hand without using theorem [ theo : ac_theta ] .
this indicates that refinements are needed in stochastic network calculus so as to tighten the bounds .
moreover , we found that backlog bounds are sensitive to adjusting parameters ( i.e. , @xmath134 , @xmath135 , @xmath170 and @xmath171 ) .
so it is necessary to use algorithm 2 to minimize the bounds .
we also conduct simulations to verify delay bounds at @xmath195 . since the backlog bounds are too loose , in order to avoid trivial validation , we use @xmath203 in ns-2 simulations to validate eq .
( [ eq : little ] ) and eq .
( [ eq : my_delay_bound ] ) ( assume that @xmath200s is sufficiently large so that we can apply these equations ) . actually , we get @xmath204 by eq .
( [ eq : little ] ) , which tightly bounds @xmath205 in ns-2 simulations . fig .
[ fig : plot_10_poisson_low](c ) shows @xmath206 in ns-2 simulations and fig .
[ fig : plot_10_poisson_low](d ) shows the upper bound of @xmath206 calculated by eq .
( [ eq : my_delay_bound ] ) . clearly , @xmath206 is upper - bounded by eq .
( [ eq : my_delay_bound ] ) .
( a ) @xmath186 ( b ) upper bound of @xmath186 ( c ) @xmath206 ( d ) upper bound of @xmath206,title="fig:",scaledwidth=60.0% ] + * experiment 2*(scenario 1 with high poisson traffic load ) we set @xmath207 to simulate high traffic load . fig .
[ fig : plot_10_poisson_high](a ) shows @xmath186 in ns-2 simulations and fig .
[ fig : plot_10_poisson_high](b ) shows the upper bound of @xmath186 calculated by eq .
( [ eq : my_backlog_bound ] ) . in this case
, @xmath134 ( @xmath135 ) is much smaller than that of experiment 1 so as to satisfy the constraint in eq .
( [ eq : my_backlog_bound ] ) , resulting in looser @xmath42 and @xmath51 . therefore , stochastic network calculus gives further loose backlog bounds .
( a ) @xmath186 ( b ) upper bound of @xmath186 ( c ) @xmath206 ( d ) upper bound of @xmath206,title="fig:",scaledwidth=60.0% ] + we also conduct simulations to verify delay bounds at @xmath195 . since the backlog bounds are too loose , in order to avoid trivial validation , we use @xmath203 in ns-2 simulations to validate eq .
( [ eq : little ] ) and eq .
( [ eq : my_delay_bound ] ) ( assume that @xmath200s is sufficiently large so that we can apply these equations ) . actually , we get @xmath208 by eq .
( [ eq : little ] ) , which tightly bounds @xmath209 in ns-2 simulations . fig .
[ fig : plot_10_poisson_high](c ) shows @xmath206 in ns-2 simulations and fig .
[ fig : plot_10_poisson_high](d ) shows the upper bound of @xmath206 calculated by eq .
( [ eq : my_delay_bound ] ) . clearly , @xmath206 is upper - bounded by eq .
( [ eq : my_delay_bound ] ) .
let @xmath177 be the average traffic rate ( packets / slot ) .
in this case , we have @xmath187 .
it is easy to see that @xmath210 < 1 $ ] for all @xmath1 because packets arrive one by one in a constant time interval .
thus , we have @xmath73 where @xmath211 again , from eq .
( [ eq : stab_802_11 ] ) , the stability condition is @xmath212 ( packet per slot ) . in fig . [
fig : plot_10_stable_cbr ] , we plot the average backlog @xmath194 $ ] at @xmath195 for @xmath213 , @xmath197 and @xmath198 in ns-2 simulations .
we observe that there is sudden jump when @xmath214 , indicating the critical point of stability is indeed around @xmath197 .
this figure also indicates the accuracy of the 802.11 model in eq .
( [ eq : tau_gamma])and([eq : gamma_tau ] ) .
( @xmath200s ) when @xmath201,title="fig:",scaledwidth=60.0% ] + * experiment 3*(scenario 1 with low cbr traffic load ) we set @xmath202 to simulate low traffic load .
[ fig : plot_10_cbr_low](a ) shows @xmath186 in ns-2 simulations and fig .
[ fig : plot_10_cbr_low](b ) shows upper bound of @xmath186 calculated by eq .
( [ eq : my_backlog_bound ] ) .
the backlog bounds are much tighter in cbr traffic than those in poisson traffic ( see experiment 1 ) .
the main reason is that we can derive a tight @xmath42 by hand instead of by theorem [ theo : ac_theta ] .
( a ) @xmath186 ( b ) upper bound of @xmath186 ( c ) @xmath206 ( d ) upper bound of @xmath206,title="fig:",scaledwidth=60.0% ] + we also conduct simulations to verify delay bounds at @xmath195 . since the backlog bounds are still loose , in order to avoid trivial validation , we use @xmath203 in ns-2 simulations to validate eq .
( [ eq : little ] ) and eq .
( [ eq : my_delay_bound ] ) ( assume that @xmath200s is sufficiently large so that we can apply these equations ) . actually , we get @xmath215 by eq .
( [ eq : little ] ) , which tightly bounds @xmath216 in ns-2 simulations . fig .
[ fig : plot_10_cbr_low](c ) shows @xmath206 in ns-2 simulations and fig .
[ fig : plot_10_cbr_low](d ) shows the upper bound of @xmath206 calculated by eq .
( [ eq : my_delay_bound ] ) . clearly , @xmath206 is upper - bounded by eq .
( [ eq : my_delay_bound ] ) . *
experiment 4*(scenario 1 with high cbr traffic load ) we set @xmath207 to simulate high traffic load . fig .
[ fig : plot_10_cbr_high](a ) shows @xmath186 in ns-2 simulations and fig .
[ fig : plot_10_cbr_high](b ) shows the upper bound of @xmath186 calculated by eq .
( [ eq : my_backlog_bound ] ) .
the backlog bounds are much tighter in cbr traffic than those in poisson traffic ( see experiment 2 ) because @xmath42 is tight here .
( a ) @xmath186 ( b ) upper bound of @xmath186 ( c ) @xmath206 ( d ) upper bound of @xmath206,title="fig:",scaledwidth=60.0% ] + we also conduct simulations to verify delay bounds at @xmath195 . since the backlog bounds are still loose , in order to avoid trivial validation , we use @xmath203 in ns-2 simulations to validate eq .
( [ eq : little ] ) and eq .
( [ eq : my_delay_bound ] ) ( assume that @xmath200s is sufficiently large so that we can apply these equations ) . actually , we get @xmath217 by eq .
( [ eq : little ] ) , which is close to ( although does not bound ) @xmath209 in ns-2 simulations .
[ fig : plot_10_cbr_high](c ) shows @xmath206 in ns-2 simulations and fig . [ fig : plot_10_cbr_high](d ) shows the upper bound of @xmath206 calculated by eq .
( [ eq : my_delay_bound ] ) .
clearly , @xmath206 is upper - bounded by eq .
( [ eq : my_delay_bound ] ) . to sum up
, the current version of stochastic network calculus often derives loose bounds when compared with simulations , especially in the case of high traffic load .
therefore , stochastic network calculus may not be effective in practice .
in this section , we first present relate work on stochastic network calculus and then on the performance analysis of 802.11 .
the increasing demand on transmitting multimedia and other real time applications over the internet has motivated the study of quality of service guarantees . towards it ,
stochastic network calculus , the probabilistic version of the deterministic network calculus @xcite@xcite@xcite@xcite , has been recognized by researchers as a promising step . during its development ,
traffic - amount - centric ( t.a.c ) stochastic arrival curve is proposed in @xcite , virtual - backlog - centric ( v.b.c ) stochastic arrival curve is proposed in @xcite and maximum - backlog - centric ( m.b.c ) stochastic arrival curve is proposed in @xcite .
weak stochastic service curve is proposed in @xcite@xcite and stochastic service curve is proposed in @xcite . in @xcite , jiang showed that only the combination of m.b.c stochastic arrival curve and stochastic service curve has all five basic properties required by a network calculus ( i.e. , superposition , concatenation , output characterization , per - flow service , service guarantees ) and the other combinations only have parts of these properties .
jiang also proposed the concept of stochastic strict server to facilitate calculation of stochastic service curve .
moreover , he presented independent case analysis to obtain tighter performance bounds for the case that flows and servers are independent .
however , there are a few bugs in his results recently found by researchers of network calculus , such as the trivial delay bound in theorem 3.5@xcite and theorem 5.1 .
therefore , we adopt v.b.c stochastic arrival curve and weak stochastic service curve in our study since we only consider backlog and delay bounds ( i.e. , service guarantee ) , ignoring the other properties .
there have been some applications of stochastic network calculus . in @xcite ,
jiang et al . analyzed a dynamic priority measurement - based admission control ( mbac ) scheme based on stochastic network calculus . in @xcite ,
liu et al . applied stochastic network calculus to studying the conformance deterioration problem in networks with service level agreements . in @xcite ,
based on stochastic network calculus , x. yu et al .
developed several upper bounds on the queue length distribution of generalized processor sharing ( gps ) scheduling discipline with long range dependent ( lrd ) traffic .
they also extended the gps results to a packet - based gps ( pgps ) system .
finally , agharebparast et al . modeled the behavior of a single wireless link using stochastic network calculus @xcite
however , little effort has been made on applying stochastic network calculus to multi - access communication systems such as 802.11 .
existing work on the performance of 802.11 has focused primarily on its throughput and capacity . in @xcite ,
bianchi proposed a markov chain throughput model of 802.11 . in @xcite , kumar
proposed a probability throughput model which is simpler than bianchi s model . in our paper , we adopt kumar s model to derive the service curve of 802.11
. there are also some work on queueing analysis of 802.11 . in @xcite , zhai et al . assumed poisson traffic arrival and proposed an m / g/1 queueing model of 802.11 . more generally , tickoo proposed a g / g/1 queueing model of 802.11@xcite@xcite . to our best knowledge ,
we are the first to model the queueing process of 802.11 based on stochastic network calculus .
in this paper , we presented a stochastic network calculus model of 802.11 . from stochastic network calculus ,
we first derived the general stability condition of a wireless node .
then we derived the stochastic service curve and the specific stability condition of an 802.11 node based on an existing model of 802.11 .
thus , we obtained the backlog and delay bounds of the node by using the corresponding theorem of stochastic network calculus .
finally , we carried out ns-2 simulations to verify these bounds .
there are some open problems for future work .
first , we derived the service curve based on an existing 802.11 model .
thus , the accuracy of the service curve depends on the accuracy of the model
. an open question may be whether we can derive the service curve of 802.11 without using any existing models .
second , we assumed the worst - case condition ( i.e. , saturate condition ) in our analysis .
can we remove this conservative assumption ? besides , under the worst - case assumption , we can assume flows and servers are independent and perform independent case analysis obtaining tighter backlog and delay bounds .
this is also one of our future work .
third , we observe that the derived bounds are loose when compared with ns-2 simulations , calling for further improvements in the current version of stochastic network calculus .
y. jiang , p. emstad , a. nevin , v. nicola , and m. fidler , `` measurement - based admission control for a flow - aware network , '' procs . , _
eurongi 1st conference on next generation internet networks - traffic engineering _ ,
2005 .
f. agharebparast and v.c.m .
leung , `` link - layer modeling of a wireless channel using stochastic network calculus , '' _
canadian conference on electrical and computer engineering _ , vol .
4 , pp . 1923 - 1926 , 2004 . 1
. let @xmath220 .
obviously , @xmath156 is an increasing function of @xmath1 with @xmath221 .
we define axes @xmath1 and axes @xmath222 ( vertical to @xmath1 ) on a plane , and plot @xmath156 on it .
we define the slope of @xmath156 , @xmath223 .
we calculate @xmath224 for @xmath225 until it converges at some @xmath226 , i.e. , @xmath227 where @xmath140 is a small number , e.g. @xmath228 .
we draw a straight line @xmath229 with the slope @xmath230 crossing the point @xmath231
. obviously , the line crosses the point @xmath232 . the maximum displacement between @xmath156 and @xmath229 ( in the direction of @xmath222 ) , @xmath233 .
we shift @xmath229 by @xmath234 in the direction of @xmath222 and get @xmath235 .
clearly , @xmath235 upperbounds @xmath156 . in other words
, we have @xmath236 and @xmath237 . in each iteration
, we generate a sample of @xmath134 , @xmath135 , @xmath170 and @xmath171 . if they satisfy the condition of eq .
( [ eq : my_backlog_bound ] ) , we calculate @xmath238 $ ] for the current and past iterations until it converges .
sample generations can use the interpolation or monte carlo method over valid ranges of the variables . | stochastic network calculus provides an elegant way to characterize traffic and service processes .
however , little effort has been made on applying it to multi - access communication systems such as 802.11 . in this paper , we take the first step to apply it to the backlog and delay analysis of an 802.11 wireless local network . in particular , we address the following questions : in applying stochastic network calculus , under what situations can we derive stable backlog and delay bounds ? how to derive the backlog and delay bounds of an 802.11 wireless node ? and how tight are these bounds when compared with simulations ? to answer these questions , we first derive the general stability condition of a wireless node ( not restricted to 802.11 ) . from this , we give the specific stability condition of an 802.11 wireless node .
then we derive the backlog and delay bounds of an 802.11 node based on an existing model of 802.11 .
we observe that the derived bounds are loose when compared with ns-2 simulations , indicating that improvements are needed in the current version of stochastic network calculus . |
the thomson problem has drawn considerable interest since the mid-1900s@xcite having found use in modeling fullerenes,@xcite drug encapsulants,@xcite spherical viruses,@xcite and crystalline order on curved surfaces.@xcite in the past few decades several computational algorithms have yielded precise numerical solutions for many-@xmath0 electron systems.@xcite the thomson problem has emerged as a benchmark for global optimization algorithms@xcite though its general solution remains unknown.@xcite numerical solutions of the thomson problem are those for which the total coulomb repulsion energy , @xmath4 is a minimum for each @xmath0-charge system with @xmath5 and @xmath6 constrained to the surface of a unit sphere . the initial condition of some minimization algorithms is the random distribution of @xmath0 point charges on the unit sphere .
this sets a relatively distant upper energy limit , @xmath7,@xcite from which numerous iterations progress toward a global minimum for each @xmath0-charge system . the distribution of numerical solutions has been fit with empirical functions including @xmath8@xcite and @xmath9.@xcite the quadratic term is ascribed to energy stored within a continuous charged shell of unit radius having total charge , @xmath0 . with this interpretation , the half - integer terms may correspond to self - energies of @xmath0 uniformly charged disks that are removed to yield the final minimized energy of discrete charges .
if the @xmath10 term is associated with the random distribution of point charges , the half - integer terms may be related to correlation energies of surface coulomb equilibrium states.@xcite these ascriptions of energy terms to physical entities have guided the development of fairly useful minimization algorithms . here
, the discrete derivative of @xmath10 is shown to correspond to the introduction of a single point charge , @xmath11 , at the origin of a given @xmath0-charge solution , and the discrete derivative of the remaining half - integer term(s ) accounts for energy needed to displace @xmath11 to the unit sphere .
this yields each subsequent @xmath1-charge solution of the thomson problem . in this manner ,
the upper limit of each minimized @xmath1 energy solution is given by @xmath12 .
this represents a well - defined charge configuration with a significantly lower energy than random charge distributions and may be useful as initial conditions in relatively more efficient energy minimization algorithms .
notably , the distribution of energy solutions of the thomson problem is `` systematic''@xcite and not random . here
, having ignored the linear term in the discrete derivative , the remaining `` systematic '' distribution of energy associated with the displacement of @xmath11 to the surface of the unit sphere demonstrably exhibits features uniquely correspondent with numerous shell - filling features throughout the periodic table of natural atomic systems.@xcite hence , introduction of @xmath11 to the thomson problem represents an intriguingly fresh perspective of this extensively useful mathematical model and the classical underpinnings of atomic structure .
numerical solutions of the thomson problem for up to 500 charges used in this communication have been obtained from a continually updated database maintained by syracuse university.@xcite
two upper energy bounds of some interest are those associated with a continuous charge shell@xcite consisting of infinite - many @xmath0 charges such that @xmath13 and the random distribution of @xmath0 discrete point charges across the surface of the unit sphere @xmath14 plots of eqs .
[ eq : lafave - shell ] and [ eq : lafave - random ] are shown in fig .
[ fig : lafave - uplus ] together with a few numerical solutions ( open circles ) of the thomson problem for illustration . for a continuous charge shell followed by @xmath15 , the energy associated with a random distribution of @xmath0 point charges .
significantly lower , @xmath16 , the energy of a given @xmath0-charge solution of the thomson problem with one charge at its origin is readily obtained by @xmath12 , where @xmath17 are solutions of the thomson problem.,width=283 ] the minimized global potential energy solutions for up to @xmath0=65 point charges were previously fit using,@xcite @xmath18 in which @xmath19 , and for up to @xmath20 using@xcite @xmath21 these smooth , continuous fit functions are generally unrepresentative of the absolute minimum energy configurations of discrete charges due to a variety of issues . among them
, discrete charges can not be infinitesimally subdivided so all points , @xmath22 , for non - integer @xmath0 have no physical significance without imposing additional arguments .
if these intermediate values of @xmath22 should have physical usefulness , for instance if an effective fractional charge on the unit sphere surface is admitted as a discrete charge approaches or leaves the surface , @xmath22 should have local minima at integer values of @xmath0 .
though these fit functions , eqs .
[ eq : lafave - erber ] and [ eq : lafave - glasser ] have no local minima at integer values of @xmath0 , they are coarsely instructive .
it is potentially more fruitful to design a fit function in accordance with the physical nature of electrostatic charge configurations . in particular , knowledge of @xmath22 at integer values of @xmath0 ,
given the discrete nature of point charges is paramount . consider the discrete derivative of @xmath22 at integer values of @xmath0 , @xmath23 of eq .
[ eq : lafave - erber ] , which yields , after binomial expansion of the half - integer term , @xmath24 for @xmath25 . for comparison , the derivative of eq .
[ eq : lafave - glasser ] , with binomial expansion of the half - integer terms , @xmath26\end{aligned}\ ] ] for @xmath25 .
the first set of parentheses in eqs .
[ eq : lafave - erberprime ] and [ eq : lafave - glasserprime ] is the discrete derivative of @xmath10 , the energy of a continuous charge shell , eq .
[ eq : lafave - shell ] .
to understand the physical charge distribution represented by the discrete energy differences , eqs .
[ eq : lafave - erberprime ] and [ eq : lafave - glasserprime ] , consider an @xmath0-charge solution , such as that of @xmath27 $ ] shown in fig .
[ fig : lafave - symmetrydiagram ] for which the total minimized electrostatic energy is @xmath17 as shown in fig .
[ fig : lafave - uplus ] . the discrete energy changes needed to obtain the solution @xmath2 in fig .
[ fig : lafave - uplus ] may be obtained by introducing an @xmath1th point charge , @xmath11 , to the origin denoted @xmath28 $ ] in fig .
[ fig : lafave - symmetrydiagram ] . its contribution to the total energy , given equal interaction with all @xmath0 charges on the unit sphere , is @xmath0 . the total energy , @xmath16 , ( fig . [ fig : lafave - uplus ] )
is @xmath12 as illustrated in fig .
[ fig : lafave - symmetrydiagram ] .
thus , the first term of each discrete derivative , @xmath29 , is due to the appearance of @xmath11 at the origin .
the remaining terms are due to the displacement of @xmath11 to the unit sphere surface and subsequent ( or in situ ) global minimization of energy , @xmath2 in fig .
[ fig : lafave - uplus ] , to yield the @xmath1 configuration ( fig .
[ fig : lafave - symmetrydiagram ] ) . in the strictest spirit of the thomson problem , introduction of @xmath11 to the _ origin _ of the unit sphere ( fig .
[ fig : lafave - symmetrydiagram]b ) increases the total energy of charges on the unit sphere by @xmath30 .
however , since the goal is to incorporate @xmath11 in the final @xmath1-charge configuration on the unit sphere , inclusion of the energy associated with @xmath11 , that is @xmath30 , while at the origin is necessary .
$ ] , with @xmath11 at the origin of @xmath27 $ ] linearly increases the energy by @xmath0 . consequently , @xmath16 is readily accessible .
displacement of @xmath11 to the unit sphere surface and global minimization of energy yields the new @xmath31 $ ] solution with @xmath32.,width=283 ] the example @xmath28 $ ] solution shown in fig .
[ fig : lafave - symmetrydiagram ] , consists of four charges at the vertices of a regular tetrahedron centered about @xmath11 at the origin .
symmetrically , the @xmath28 $ ] configuration is identical to the @xmath27 $ ] solution of the thomson problem . in general , @xmath33 .
therefore , the energy difference @xmath34 associated with the addition of a charge in the thomson problem may be expressed , @xmath35 as shown in fig .
[ fig : lafave - symmetrydiagram ] .
consequently , @xmath36 accounts for the half - integer terms in eqs .
[ eq : lafave - erberprime ] and [ eq : lafave - glasserprime ] and the @xmath37 term associated with the discrete derivative of @xmath10 . ,
fit to @xmath0=500 numerical solutions of the thomson problem , is shown on the first @xmath0=100 numerical solutions of the energy difference @xmath36.,width=283 ] to check the validity of the discrete transformations from each @xmath0-charge solution of the thomson problem to its adjacent @xmath1-charge solution , @xmath36 in eq .
[ eq : lafave - un ] is fit with the remaining terms of the discrete derivatives of eqs .
[ eq : lafave - erberprime ] and [ eq : lafave - glasserprime ] .
the distribution of energy differences for up to 500-charge numerical solutions of the thomson problem ( fig .
[ fig : lafave - fit ] ) is well - fit with the first three terms of eq .
[ eq : lafave - erberprime ] for which @xmath38 , in very good agreement with previous work in which @xmath39 for up to @xmath40.@xcite truncating eq .
[ eq : lafave - erberprime ] to only the first term , including the @xmath37 term from the discrete derivative of the quadratic term in eq .
[ eq : lafave - erber ] , @xmath41 provides a reasonably good fit with @xmath42 .
fitting with half - integer terms in eq .
[ eq : lafave - glasserprime ] yields large errors associated with @xmath43 and larger values of @xmath44 .
these half - integer terms are , however , useful for better fits to smaller @xmath0-charge solutions.@xcite since fitting smooth functions to a discrete , `` systematic '' data set is an insufficient means of characterizing solutions of the thomson problem , eq .
[ eq : lafave - sqrt ] is a convenient fit function for @xmath36 for up to 500 charges as shown for up to 100 charges in fig .
[ fig : lafave - fit ] .
applied to a continuous charge shell , introduction of @xmath11 ( one of an infinite number of equal charge elements ) to the origin raises the energy by @xmath0 , @xmath45 here , the transformation yields a lower energy than @xmath46 , @xmath47 however , the energy difference of @xmath48 accounts for the @xmath37 term in the finite difference of @xmath10 . applied to purely random charge distributions of energy given by eq .
[ eq : lafave - random ] , the @xmath11 transformation raises the energy of each @xmath0-charge configuration by @xmath0 , @xmath49 its difference from the energy of a random @xmath1-charge distribution is zero since @xmath50 consequently , the @xmath11 transformation _ generates _ energies of purely random charge distributions .
however , as there is no energy difference between @xmath51 randomly distributed charges on a unit sphere and @xmath0 charges on a unit sphere with @xmath11 at the origin , these two configurations are energetically indistinguishable .
the @xmath11 transformation uniquely distinguishes between continuous and discrete charge systems .
applied to the thomson problem , the @xmath11 transformation raises the energy of each @xmath0-charge solution by @xmath0 and represents an energetically unfavorable charge configuration to the @xmath51 solution with all charges on the unit sphere as made evident by fig .
[ fig : lafave - fit ] . in more detail , using eq .
[ eq : lafave - erber ] , inclusion of @xmath11 to each @xmath0-charge solution imposes the following inequality on the coarse fit parameter , @xmath52 , @xmath53^{-1}.\end{aligned}\ ] ] for nonzero positive integer values of @xmath0 , @xmath52 must be greater than @xmath54 ( @xmath55 ) . since the coarse fit parameter @xmath56 is greater than this threshold , the @xmath11 transformation yields an upper limit to all energy solutions of the thomson problem .
coarsely - defined fit functions @xmath57 presume the possibility of fractional charges between integer values of @xmath0 . indeed , each @xmath0-charge solution represents a global electrostatic potential energy _
minimum _ that is not represented by these fit functions .
consequently , the underlying physical nature of discrete charge systems is in conflict with the mathematics used to study them . as a new point charge
is added to a given @xmath0-charge configuration the @xmath0 charges respond to its arrival until an @xmath58 $ ] configuration is obtained .
a fit function may be designed to reflect local minima at integer values of @xmath0 .
consequently , the fit function may `` hop '' from one energy minimum to the next through discrete energy differences to satisfy the condition of local minima at integer values of @xmath0 .
a few possible fit functions are shown schematically in fig .
[ fig : lafave - functions ] .
the saw - tooth fit function sketched in fig .
[ fig : lafave - functions]a reflects abrupt introduction of @xmath11 at the origin of each @xmath0-charge configuration followed by a gradual relaxation of the function to the neighboring @xmath58 $ ] configuration
. this function may be useful if the strict definition of charges on the _ surface _ of the unit sphere is to be maintained .
however , if the number of charges within the _ volume _ of the unit sphere is of interest , then this function is inappropriate as the @xmath28 $ ] configuration contains @xmath51 charges and should be plotted to reflect this number .
is introduced abruptly to each @xmath0-charge solution , and the energy is gradually minimized to @xmath2 .
b ) @xmath11 forms gradually at the origin , and the total energy is minimized to @xmath2 .
c ) energy minima are obtained with a smooth , periodic function having minima at integer values of @xmath0 .
d ) energy solutions are obtained with a piecewise - continuous function with minima at integer @xmath0 values.,width=283 ] the sawtooth fit function sketched in fig .
[ fig : lafave - functions]b rises linearly from each @xmath0-charge energy solution to a new upper limit of each subsequent @xmath51 energy solution .
if this linear rise is considered as a gradual `` charging up '' of @xmath11 at the origin of each @xmath27 $ ] , then the slope of the function is @xmath59 .
the generally shorter drop from @xmath16 , ( cf .
[ fig : lafave - uplus ] ) plotted appropriately at each @xmath60 , to @xmath2 maintains the total number of charges in the system ( including the spherical volume ) as @xmath11 is displaced from the origin to the surface . during this transformation ,
the spatial symmetry of the system changes from that of @xmath27 $ ] to @xmath31 $ ] .
these discrete symmetry transformations are responsible for correlations made previously@xcite to electron shell - filling throughout the periodic table of elements as discussed in the next section . though not treated here , continuously smooth functions having periodic ( @xmath25 ) local minima , fig .
[ fig : lafave - functions]c , or sharply defined ( infinite derivative ) behavior at integer values of @xmath0 as shown in fig .
[ fig : lafave - functions]d may also be useful to obtain solutions of the thomson problem .
perhaps such a fit function may include a variety of polynomial series exhibiting oscillatory behavior .
the discrete derivative prohibits electrostatic energy `` states '' between integer values of @xmath0 . instead of beginning with an @xmath0-charge configuration we may begin with an @xmath1-charge configuration in fig . [
fig : lafave - symmetrydiagram ] .
the energy @xmath36 represents an increase for the @xmath1 system by an amount approximately proportional to @xmath61 .
the resulting @xmath16 may then be associated with the @xmath0-charge configuration in the thomson problem ( using a strict `` surface - only '' interpretation ) .
when plotted , fig .
[ fig : lafave-50 ] ( open circles ) , the distribution yields a pattern of energy differences , @xmath62 ( note the explicit association of energy with the @xmath0-charge configuration ) , consistent with electron shell - filling throughout the periodic table of elements as previously reported.@xcite in particular , as one progresses through the periodic table ( increasing atomic number ) the outermost electron in respective atomic structures is characterized by a wavefunction ( or `` orbital '' ) having different spatial symmetries .
for instance , the spherical @xmath63-shell associated with beryllium ( @xmath64=4 ) is followed by boron ( @xmath64=5 ) , a system with an outermost electron having a dumbbell - shaped @xmath65-shell .
these spatial symmetry differences among wavefunctions of neighboring atomic systems are repeatedly coincident with noticeable disparities found in @xmath62 intrinsic to the thomson problem .
moreover , a relatively lower energy is obtained at the `` close '' of each electron shell followed by a disparity to a relatively higher energy which `` opens '' each subsequent shell .
for example , a lower energy appears at @xmath0=4 followed by an abrupt disparity to a higher energy for @xmath0=5 .
( open circles ) exhibit disparities uniquely correspondent to shell - filling behavior in natural atomic structure .
these disparities are enunciated when the thomson problem is treated within a dielectric sphere ( filled circles ) .
these data are given for a dielectric sphere of 100 nm radius and dielectric constant , @xmath66.,width=283 ] the thomson problem treated within a dielectric sphere,@xcite using an appropriate model for discrete charges in the presence of dielectrics@xcite these disparities become more pronounced as shown in fig .
[ fig : lafave-50 ] ( solid circles ) .
the shift , @xmath67 , in numerical solutions of the thomson problem is unnecessary here .
note , for example , a dielectric sphere containing a single electron stores nonzero energy referenced to the empty , or electrically neutral , dielectric sphere .
enunciation of these features when using a dielectric sphere is partly due to the use of a `` thomson radius '' as a variational parameter rather than a static unit sphere .
both frameworks may be traced to the classical `` plum - pudding '' model of the atom@xcite in which electrons reside within a uniform positively - charged sphere .
a neutral dielectric sphere is used in place of the positively - charged sphere , and the thomson problem is a natural consequence of the removal of this positively - charged sphere in which electrons fly apart to form a spherical distribution.@xcite many additional correspondences with atomic shell - filling have been reported.@xcite
in place of coarsely - defined fit functions to numerical solutions of the thomson problem , the discrete derivative of these fit functions has been demonstrated to correspond with the well - defined configuration of @xmath0 charges on a unit sphere surrounding an @xmath1th charge , @xmath11 , at its origin followed by relaxation of this volumetric problem to each subsequent @xmath1-charge solution of the thomson problem .
this approach provides useful new information about the discrete charge system not obtained by the insufficient mathematical description of coarse fit functions previously used to model solutions of the thomson problem .
the existence of possibly better fit functions based on the underlying physical properties of the charge system described may lead to better approximations and potentially exact solutions of the thomson problem which appears to be intrinsically shared with atomic structure .
discrete transformations between neighboring @xmath27 $ ] and @xmath31 $ ] configurations through an intermediate @xmath28 $ ] configuration utilizing @xmath11 at the origin include a trivial linear increase in energy followed by a non - trivial minimization of energy . the latter transformation from a well - defined configuration to the unknown @xmath31 $ ] configuration is the embodiment of the thomson problem .
further developments toward exact solutions of this transformation are merited , substantiated by its many correspondences with electron shell - filling in atomic structure , and its usefulness to a wide range of other discrete charge modeling applications .
identification of these discrete transformations in the thomson problem may be the first step toward the development of new and more efficient energy minimization algorithms . | a significantly lower upper limit to minimum energy solutions of the electrostatic thomson problem is reported .
a point charge is introduced to the origin of each @xmath0-charge solution .
this raises the total energy by @xmath0 as an upper limit to each @xmath1-charge solution .
minimization of energy to @xmath2 is well fit with @xmath3 for up to @xmath0=500 .
the energy distribution due to this displacement exhibits correspondences with shell - filling behavior in atomic systems .
this work may aid development of more efficient and innovative numerical search algorithms to obtain @xmath0-charge configurations having global energy minima and yield new insights to atomic structure .
`` t. lafave jr .
_ j. electrostatics _ * 72*(1 ) 39 - 43 ( 2014 ) '' .
doi : 10.1016/j.elstat.2013.11.007 |
the duffing oscillator is generally used to describe nonlinear dynamics in oscillating systems @xcite . the corresponding duffing equation models a damping and driven oscillator with more complicated behavior than simple harmonic motion , which can be used to exhibit chaos in dynamics and hysteresis in resonance @xcite . on the other hand ,
the motion of trapped ions is highly controllable and can be employed to transfer quantum information when cooled down to ground state @xcite .
since it is effectively approximated to be harmonic , the ion motion in a quadruple electromagnetic trap @xcite can be regarded as a good mechanical oscillator , which may exhibit nonlinearity when driven to the nonlinear field .
for example , duffing nonlinear dynamics has been investigated in a single ion confined in the linear ion trap @xcite .
the trap nonlinearity introduces instability in the motion of the ion , which should be avoided in most times , but can also be used in resonance rejection and parameter detection in mass spectrometry @xcite .
recent research also showed the feasibility of phonon lasers based on the nonlinearity of a single trapped ion under laser irradiation @xcite .
we focus in this work on the nonlinearity in a home - built surface - electrode trap ( set ) .
the set , with capability to localize and transport trapped ions in different potential wells , is a promising setup for large - scale quantum information processing @xcite . in comparison to conventional linear paul traps , however , the reduced size and asymmetry in set lead to stronger high - order multipole potentials @xcite , which affect the stability of the ion trapping . to solve the problem we have to understand the source and the strength of the nonlinearity .
due to complexity resulted from the high - order multipole potentials , the nonlinear effect in the set can not be simply described by the duffing oscillator as for the linear trap , but an inhomogeneous - coupled duffing oscillator involving quadratic and cubic nonlinearities .
we observed the nonlinearity experimentally in our set , and by the method of multiple scales we derived an inhomogeneous - coupled duffing oscillator to fit the experimental values , which shows that both the nonlinearity and axial - radial coupling exist in the case of the frequency resonance ( i.e. , around the regime of driving detuning being zero ) .
moreover , we show in the non - coupling case different nonlinear effects in different dimensions , which is due to different asymmetry in fabrication of the set .
the @xmath0 ion is cooled by the laser beam at 397 nm on the @xmath1 transition , helped by another laser light at 866 nm as repumping . here , the trapped ion is above the electrodes by 0.8 mm ( see the bottom - right inset).,width=8 ] our home - built set is a 500 @xmath2m - scaled planar trap with five electrodes for radial confinement , and fabricated by printed circuit board technology @xcite . as shown in fig .
1 , the five electrodes consists of a central electrode , two radio - frequency ( rf ) electrodes and two outer segmented dc electrodes , where the rf electrodes , the central electrode and the gaps in between are of the same width of 500 @xmath3 m .
each outer segmented electrode consists of five component electrodes , i.e. , a middle electrode , two control electrodes and two end electrodes .
the widths of the control electrodes and end electrodes in the segment are 1.5 mm and the middle electrode is 1 mm wide .
the gap in the segmented electrode is of 500 @xmath3 m width .
when the set works , the trapped @xmath0 ion stays above the electrodes by @xmath4 mm , and the pseudopotential trapping depth is below @xmath5 ev with rf amplitude @xmath6(0-peak)@xmath7 v and rf frequency @xmath8 mhz .
the voltage on the four end electrodes is @xmath9 v but zero on other electrodes . the experimental setup is plotted in fig .
2 , where the ultraviolet radiation at @xmath10 nm excites @xmath11 transition by a grating stabilized laser diode with power up to @xmath12 mw and linewidth less than 2 mhz .
another grating stabilized laser diode at @xmath13 nm with power up to 100 mw and linewidth less than 5 mhz excites the @xmath14 transition .
the frequencies of the both laser diodes have been calibrated to the wavelength meter ( highfiness , ws-7 ) .
typical laser powers at the trap center are 50 @xmath3w for 397 nm in red detuning ( -80 mhz ) and 500 @xmath3w for 866 nm in carrier transition . in our set ,
the single @xmath0 ion is laser cooled and stably confined , which is monitored by photon scattering collected by both an electron - multiplying charge - coupled - device ( emccd ) camera and a photomultiplier tube ( pmt ) . outside the vacuum chamber
, the electrical connections immediately encounter a `` filter box '' , which provides low - pass filtering of the voltages applied to the electrode . an additional drive force is electrically connected to one of the middle electrodes behind the filter box , which provides an excitation to drive the ion away from equilibrium . due to the design of our set system ,
the motion of the ion is detected only in the @xmath15 plane by the emccd . as a result ,
what we study throughout the work is the oscillation along the axial direction ( @xmath16-axis ) and the radial direction ( @xmath17-axis ) , whose harmonic frequencies are , respectively , @xmath18 khz and @xmath19 khz
. moreover , since the harmonic frequency in @xmath20-axis is @xmath21 khz , much bigger than in other axes , the ion can be regarded as a very tight confinement in y direction .
we have suppressed the micro - motion by the rf - photon cross correlation compensation @xcite , which yield cooling of the ion down to the temperature below 10 mk .
to study the nonlinear mechanical response , we drive the ion to the nonlinear regime by a small oscillating voltage , i.e. , @xmath22 v , applied to one of the middle electrodes .
we slowly increase the driving frequency with the scan step @xmath23 khz , from @xmath24 khz to @xmath25 khz ( the positive scan ) , the ion oscillates first along the @xmath16-axis and then turns to the @xmath17-axis for oscillation .
the particularly interesting observation is the simultaneous responses , i.e. , a rectangle trajectory , in both @xmath17- and @xmath16- axes when the sweep is close to the harmonic resonator frequency in either of the axes .
similar behavior is also found in the negative scan .
we measure the ion oscillation by taking time - averaged images from the emccd . in fig .
3 , seventeen such images for different drive frequencies are presented for positive and negative scans , respectively . for the ion originally oscillating in the @xmath16-axis , we slowly scan the drive frequency @xmath26 across the harmonic resonance at @xmath27 . when the detuning @xmath28 approaches zero , the rectangle trajectory appears , implying a coupled motion between @xmath17- and @xmath16-axes due to axial - radial coupling ( explained later ) . for a more clarified observation
, we scan with smaller steps around the regime @xmath29 , as shown in fig . 4 which gives us an accurate range from the appearance of the rectangle to the disappearance .
to understand the observation above , we have to consider the multipole potential in the set , which is given by @xcite @xmath30 where the subscript @xmath31 labels different electrodes and the subscript @xmath32 is for the spherical harmonics @xmath33 . both @xmath34 and the related parameters @xmath35 are defined in littich .
@xmath36 is the weight factor for different electrodes . in this treatment ,
the initial equilibrium position of the single trapped ion is defined as the origin of the coordinates .
the five - wire set generally consists of quadrupole and hexapole potentials @xcite . considering the defect in our set
, we also involve octopole potential in our treatment . following the definition in @xcite
, we have the subscripts @xmath37 where @xmath32 from 5 to 9 , from 10 to 16 and from 17 to 25 correspond , respectively , to the quadrupole , hexapole and octopole potentials . for different potentials @xmath38 applied , respectively , to n electrodes , where @xmath39 is the dc voltage on the electrode @xmath31 and @xmath40 represents the rf voltage @xmath41 on the electrode @xmath31 driven at frequency @xmath42 , we rewrite eq . ( 1 ) for the dc potential @xmath43 and the rf potential @xmath44 as @xmath45where we have used @xmath46 and @xmath47 .
moreover , we have the 1d motional equation for the trapped ion @xcite , @xmath48with @xmath49 and @xmath16 . combining eq .
( 2 ) with eq . ( 3 ) , we obtain the equation of motion in the @xmath16 direction , @xmath50where @xmath17 , @xmath20 and @xmath16 represent , respectively , the displacement of the ion from the equilibrium position in the three dimensions , @xmath2 is the linear damping parameter originated from the recoil due to photon absorption , @xmath51 is driving amplitude .
the detailed expressions of the nonlinear coefficients @xmath52 @xmath53 can be found in appendix i. compared to the duffing oscillator in @xcite , eq .
( 4 ) is a complicated duffing oscillator , containing additional coupled - motion terms . using the method of multiple scales @xcite
, we obtain the steady - state solution to eq .
( 4 ) as @xmath54where the nonlinear coefficients @xmath55 and @xmath56 are relevant to the coupled motion along @xmath17- and @xmath20-axes .
@xmath57 , @xmath58 and @xmath59 are the response amplitudes , respectively , in @xmath17 , @xmath20 and @xmath16 directions .
@xmath60 and @xmath61 represent the harmonic frequencies in @xmath17- and @xmath20-axes . for more clarification
, we define a parameter @xmath62 as @xmath63where @xmath64 originates from the cubic nonlinearity , @xmath65 represents the nonlinear coefficient that comes from quadric nonlinearity , and @xmath66 corresponds to the nonlinear dispersion relevant to the coupled motion . substituting @xmath62 into eq .
( 5 ) , we obtain @xmath67
in our home - built set , since the harmonic frequency in y - axis is much bigger than in other axes , the ion is confined very tightly in y direction , which leads to a reasonable assumption @xmath681 . as a result
, the coupled term @xmath69 is reduced to @xmath70 with @xmath71 .
moreover , @xmath64 and @xmath72 in eq .
( 6 ) are nothing to do with the coupled motion and their sum @xmath73 can be measured experimentally by @xmath74with the maximal amplitude @xmath75 and the maximal detuning @xmath76 in the non - coupling case . as a result , eq .
( 7 ) is reduced to a steady - state solution to the amplitude of the response @xmath59 with respect to the drive detuning @xmath77 for the known driving force amplitude @xmath78 , @xmath79 khz .
( b ) the measured values correspond to the situation in fig .
calculation ( light blue solid line ) by eq .
( 9 ) with the coupling term fits the measured values with positive slow scan ( red stars ) around zero detuning .
the scan step is @xmath80 khz , and the blue stars are for the measured values of the vibrational amplitude @xmath57 in @xmath17-axis . as a comparison ,
the calculation without the coupling term is plotted ( the black solid curve).,width=8 ] in our experiment , @xmath81 is measured via ion response in the linear regime , @xmath82 hz@xmath83 m is obtained by observing the ion displacement versus the middle electrode voltage @xcite , @xmath84 hz @xmath83/m@xmath83 is measured using the observed dependence of @xmath75 on the maximal detuning @xmath85 .
we evaluate @xmath86 hz using the relation @xmath87 .
the comparison in the non - coupling case between the measured and calculated values of @xmath59 and @xmath77 is made in fig .
5(a ) , where eq . ( 9 ) without the coupling term @xmath88 ( the black solid curve ) can fit most experimental values for both the positive and negative scans ( red stars and green crosses , respectively ) .
in this situation , the vibrational amplitude @xmath57 in @xmath17-axis is negligible .
some experimental values around @xmath89 , which are not fitted well by the solid curve , are actually relevant to the case of coupled motion .
to be more clarified , we scan the region around @xmath89 with smaller step than in fig
the fitting by considering the coupling term @xmath88 in our calculation can fully cover the measured data , as shown in fig .
in such a case , we find that the vibrational amplitude @xmath57 in @xmath17-axis is visible , which is excited by the energy transfer from @xmath16-axis due to motional coupling .
this energy transferred from @xmath16-axis to @xmath17-axis is nearly constant in the adiabatic operation so that we obtain @xmath90 hz@xmath83/m@xmath83 .
5(b ) also shows that the motional coupling stops when @xmath91 approaches @xmath92 khz .
we see that @xmath59 goes up to a maximum with @xmath57 dropping to zero , implying that the system returns to the non - coupling case . therefore , the vibrational trajectories imaged in figs
. 3 and 4 can be fully understood by the complicated duffing oscillator with and without the term for coupled motion .
-axis with respect to the driving detuning , where the scan step is 1 khz and the measurement is made in the no - coupling case .
calculation ( black solid curve ) fits most of the measured values with positive scan ( red stars ) and negative scan ( green crosses ) .
, width=8 ] moreover , we also checked nonlinear effects in different directions in our set by applying the drive on the @xmath17-axis and repeating the experimental steps as above for @xmath16-axis .
it is physically evident that the behavior can be described by a slight modification of eq .
( 4 ) by exchanging @xmath16 and @xmath17 , and replacing @xmath78 and @xmath93 by @xmath94 and @xmath95 . as an example
, we only present the non - coupling case in fig .
6 , where the measured data is fitted well by the steady solution to the complicated duffing oscillator with different nonlinear coefficients and different damping rates from in fig .
5(a ) . in comparison to the shape of the curve in fig .
5(a ) , the oscillation in such a case reaches the maximum amplitude before @xmath89 , i.e , the red detuning , corresponding to @xmath96 in eq .
this implies negative coefficients of quadratic and cubic terms in eq .
( 4 ) originated from the different asymmetry from in @xmath97-axis .
in conclusion , we have experimentally investigated the complicated oscillations in our home - built set , which are related to the high - order multipole potentials .
both the coupling and non - coupling cases , as well as the driving along different axes , are studied .
our observation can be fully understood by the nonlinear effects and the motional coupling from the solution of a complicated duffing oscillator . in comparison to the relevant study on a single ion oscillating in the linear trap @xcite , our home - made set owns higher - order multipole potentials , which cause more fruitful nonlinear effects and even the motional coupling between different directions .
although there are also axial - radial couplings observed in the ion trap , e.g. , with ion cloud in cloud , such a motional coupling is much more evident in the set , which , in addition to the couplings regarding @xmath98 and @xmath99 , is also reflected in the @xmath100 term in eq .
( 4 ) with the coefficient @xmath101 .
according to our calculation , the motional coupling in our observation is mainly influenced by the coefficients @xmath101 and @xmath102 , which implies the combined action from the quadrupole , hexapole and octopole potentials .
this is the reason that the rectangle trajectories have never been observed previously in linear ion traps .
moreover , recent investigation of the phonon laser based on the nonlinear oscillation of the trapped ion demonstrated the analogy to the fabry - perot laser with 100% reflecting mirrors @xcite .
in contrast , the set under our study seems an asymmetry fabry - perot cavity , which may yield two split beams of the phonon laser in perpendicular axes .
coherent transfer between the two split phonon beams would be useful in fundamental physics and practical application .
further work in this aspect is underway . with the trapped ions cooled down to the motional ground state
, we may have an excellent platform to demonstrate nonlinear behavior in a fully quantum mechanical regime and also carry out quantum logic gate operations .
therefore our work presents a way to exploring complicated nonlinearity using experimentally available techniques and it is also useful for suppressing detrimental effects from the nonlinearity in quantum information processing using trapped ions .
this work is supported by national fundamental research program of china under grant no .
2012cb922102 , and by national natural science foundation of china under grants no .
11274352 and no .
our calculation is based on eq . ( 4 ) , in which the nonlinear coefficients originate from the high - order multipole potentials . by setting @xmath103 and @xmath104 to be the electric quantity and the mass of single calcium ion , we have the nonlinear coefficients as @xmath105 } { m^{2}r_{0}^{8}\omega ^{2}},\]]@xmath106 } { m^{2}r_{0}^{8}\omega ^{2}},\]]@xmath107 } { m^{2}r_{0}^{8}\omega ^{2 } } \notag \\ & & + \frac{3e\left [ 24er_{0}^{2}m_{12}m_{13}(u_{12}^{\ast } ) ( u_{13}^{\ast } ) + 7mr_{0}^{4}\omega ^{2}m_{20}(v_{20}^{\ast } ) \right ] } { m^{2}r_{0}^{8}% \omega ^{2}},\end{aligned}\]]@xmath108 } { m^{2}r_{0}^{8}\omega ^{2 } } \notag \\ & & + \frac{3e\left [ 24er_{0}^{2}m_{13}m_{14}(u_{13}^{\ast } ) ( u_{14}^{\ast } ) + 7mr_{0}^{4}\omega ^{2}m_{22}(v_{22}^{\ast } ) \right ] } { m^{2}r_{0}^{8}% \omega ^{2}},\end{aligned}\]]@xmath109 } { m^{2}r_{0}^{8}\omega ^{2 } } \notag \\ & & + \frac{e\left [ -60mr_{0}^{4}(v_{21}^{\ast } ) \omega ^{2}m_{21}-14mr_{0}^{4}(v_{23}^{\ast } ) \omega ^{2}m_{23}-240er_{0}^{2}(u_{7}^{\ast } ) \left ( u_{21}^{\ast } \right ) m_{7}m_{21}-56er_{0}^{2}(u_{7}^{\ast } ) \left ( u_{23}^{\ast } \right ) m_{7}m_{23}\right ] } { m^{2}r_{0}^{8}\omega ^{2}},\end{aligned}\ ] ] @xmath110 } { m^{2}r_{0}^{8}\omega ^{2 } } \notag \\ & & + \frac{e\left [ -60mr_{0}^{4}(v_{21}^{\ast } ) \omega ^{2}m_{21}+14mr_{0}^{4}(v_{23}^{\ast } ) \omega ^{2}m_{23}-240er_{0}^{2}(u_{7}^{\ast } ) \left ( u_{21}^{\ast } \right ) m_{7}m_{21}+56er_{0}^{2}(u_{7}^{\ast } ) \left ( u_{23}^{\ast } \right ) m_{7}m_{23}\right ] } { m^{2}r_{0}^{8}\omega ^{2}},\end{aligned}\ ] ] @xmath111 } { m^{2}r_{0}^{8}\omega ^{2 } } \notag \\ & & + \frac{e\left [ 64er_{0}^{2}(u_{12}^{\ast } ) \left ( u_{14}^{\ast } \right ) m_{12}m_{14}+56er_{0}^{2}\left ( u_{7}^{\ast } \right ) \left ( u_{19}^{\ast } \right ) m_{7}m_{19}+21er_{0}^{2}m_{8}m_{20}(u_{8}^{\ast } ) ( u_{20}^{\ast } ) % \right ] } { m^{2}r_{0}^{8}\omega ^{2}},\end{aligned}\ ] ] @xmath112 } { m^{2}r_{0}^{8}\omega ^{2}},\ ] ] @xmath113 , \ ] ] with @xmath114 /m^{2}r_{0}^{8}\omega ^{2}$ ] , and the scaling factor @xmath115 ( see definition in @xcite ) .
eq . ( 5 ) is obtained by the standard steps of the multiple scale method . starting from eq .
( 4 ) , we assume that the driving frequency is a perturbative expansion of harmonic oscillator frequency @xcite , i.e. , @xmath116 , with a small dimensionless parameter @xmath117 . following the method in nayfeh1984 , we rewrite eq .
( 4 ) by setting @xmath118 , @xmath119 , @xmath120 , the damping term as @xmath121 and the driving term as @xmath122 . introducing a new parameter @xmath123 @xmath124
, we rewrite @xmath125 as , @xmath126then we compare the coefficients of @xmath127 , @xmath128 and @xmath129 , which yields , @xmath130@xmath131@xmath132,\end{aligned}\]]where @xmath133 , @xmath134 .
we may solve @xmath135 and @xmath136/3\omega _
{ 0z}^{2}$ ] from eqs .
( 20 ) and ( 21 ) by eliminating the secular term .
similarly , from equations of the oscillations in x - axis and y - axis , we may solve the variables @xmath137 , @xmath138 , @xmath139 and @xmath140 as @xmath141@xmath142@xmath143/(3\omega _
{ 0x}^{2}),\]]@xmath144/(3\omega _ { 0y}^{2}),\]]where @xmath55 and @xmath56 correspond to the quadric nonlinearity of the ion motion equation in @xmath17 and @xmath20 directions , respectively . @xmath145 and @xmath61 represent the harmonic frequencies in x direction and y direction .
@xmath146 , @xmath147 and @xmath148 , where @xmath149 and @xmath150 are real functions of @xmath151 , @xmath152 and @xmath150 represent the phases of different dimensions .
@xmath153 , @xmath154 and @xmath155 are conjugate terms of a , b and c. substituting @xmath156 and @xmath157 into eq .
( 22 ) , we obtain an equation regarding the secular term , from which , in combination of eqs .
( 23 - 26 ) with the expressions of @xmath158 , @xmath159 and @xmath160 , we obtain @xmath161@xmath162with @xmath163 .
we assume the steady - state motion corresponding to @xmath164 .
so we have @xmath165which is actually eq . ( 5 )
. k. vahala , m. herrmann , s. knnz , v. batteiger , g. saathoff , t. w. hnsch and th .
udem , nat . phys . * 5 * , 682 ( 2009 ) ; s. knnz , m. herrmann , v. batteiger , g. saathoff , t. w. hnsch , k. vahala and th .
udem , phys .
* 105 * , 013004 ( 2010 ) . | the oscillation coupling and different nonlinear effects are observed in a single trapped @xmath0 ion confined in our home - built surface - electrode trap ( set ) .
the coupling and the nonlinearity are originated from the high - order multipole potentials due to different layouts and the fabrication asymmetry of the set .
we solve a complicated duffing equation with coupled oscillation terms by the multiple scale method , which fits the experimental values very well .
our investigation in the set helps for exploring nonlinearity using currently available techniques and for suppressing instability of qubits in quantum information processing with trapped ions . |
perhaps over half of all galaxies lie within groups containing 3 20 members @xcite ; yet , due to the difficulty of discerning them from the field , groups of galaxies are , as a whole , not as well studied as larger galaxy systems .
compact groups of galaxies ( cgs ) , however , defined by their small number of members ( @xmath0 10 ) , their compactness ( typical intra - group separations of a galaxy diameter or less ) , and their relative isolation ( intra - group separations @xmath3 group - field separations ) are readily identifiable .
studies of cgs mainly concentrate on two issues : ( 1 ) what is the origin and relative importance of cgs in the universe ? ( 2 ) is there a relation between the global properties of these systems and the formation and evolution of their member galaxies ?
the possibility that these two issues are connected makes cgs particularly interesting ( see the review by @xcite ) .
the first example of a cg was found over one hundred years ago by @xcite .
the best known catalog is that of the hickson compact groups ( hcgs ; @xcite ) , a sample comprising 100 groups selected from the red ( e ) prints of the palomar observatory sky survey ( poss ) .
other catalogs now available include an initial dposs cg catalog @xcite , the southern cg catalog @xcite , and those extracted from the 3d uzc galaxy catalog @xcite and from the cfa2 @xcite and las campanas @xcite redshift surveys . due to their high densities ( equivalent to those at the centers of rich clusters ) and low velocity dispersions ( roughly 200 km s@xmath4 ) ,
cgs represent an environment where interactions , tidally triggered activity , and galaxy mergers are expected to be more prevalent than in most other environments .
studies of interacting galaxy pairs , both in the infrared and ultraviolet , suggest that interactions can trigger an inflow of gas to the galactic nucleus , resulting in either starburst or agn activity .
although individual cgs are known to contain starbursts and agn @xcite , samples of cg galaxies as a whole do not appear to show rates of either activity enhanced beyond that of the field @xcite .
some calculations have predicted extremely short dynamical lifetimes for cgs , @xmath5 1 gyr , leading to speculation that many or most cgs may in fact be chance alignments instead of real structures @xcite .
however , evidence of interactions and the discovery of diffuse intra - group x - ray gas within perhaps as many as 75% of the hcgs @xcite attest to their physical reality .
n - body simulations have pointed out two possible classes of solution to the short dynamical time problem .
the first is that there is ongoing formation of cgs , and the longevity of the group is due to secondary infall . in this case
, cgs must be continually forming in moderately dense environments like those of loose groups , and , in fact , many are seen to be embedded in such systems @xcite . in the second class of solution ,
the longevity of cgs is due to either their specific initial conditions or a massive halo encompassing the entire group @xcite .
@xcite have shown that cgs with an appropriate arrangement of luminous and dark matter can persist for @xmath6 25 gyr ; that massive common dark matter halos around cgs might indeed exist is not inconsistent with with the analysis of the dynamics of satellite galaxies within individual cgs @xcite .
thus , cgs have sparked intense interest , both in their formation and evolutionary histories , and in the interaction phenomena associated with these dense environments .
our difficulties in understanding the existence of cgs and their dynamics may be a consequence of the small samples studied so far .
further progress will require a larger , deeper , and more uniform sample of cgs with which to study their nature , evolution , and origin .
the sloan digital sky survey @xcite , which will eventually cover up to one quarter of the sky with uniform photometry in five filters , makes for an obvious cg hunting ground .
we have thus embarked on a project to extract an objectively defined catalog of cgs from the sdss @xcite , starting with runs 752 and 756 of the sdss early data release ( edr ) @xcite .
we present this initial catalog as follows . in [ data ]
we describe the region of the sky used for this preliminary search . in [ cat_construct ] , we describe our cg catalog construction techniques . in
[ catalog ] we present the catalog and an atlas of corresponding postage - stamp images for the individual groups . in [ comparison_other ]
we compare the properties of this catalog to those of previous cg catalogs . in
[ morph - env ] we compare the properties of sdss cg member galaxies to the properties of sdss field galaxies . in [ conclusions ] we conclude and describe our future plans .
throughout the paper , a flat @xmath7 cosmological model with @xmath8 , @xmath9 , and @xmath10 km s@xmath4 mpc@xmath4 is assumed .
the sdss has as its goal to image up to 10,000 sq deg ( @xmath11 steradians ) of the northern galactic cap in five different filters ( @xmath12 , @xmath13 , @xmath14 , @xmath15 , @xmath16 ) and to perform followup spectroscopy of @xmath17 galaxies and @xmath18 quasars selected from the imaging survey
. the survey will be complete ( s : n @xmath6 5:1 ) to limiting @xmath12 @xmath13 @xmath14 @xmath15 @xmath16 magnitudes of roughly 22.3 , 23.3 , 23.1 , 22.3 and 20.8 respectively .
an additional three stripes in the southern galactic cap will be observed repeatedly for variability studies and will reach a co - added depth of @xmath19 . while the sdss was originally planned to be on the @xmath20 system of @xcite
, the delivered filters do not perfectly conform to this system .
the photometric calibration system for sdss ( called @xmath21 ) had only been finalized as of the recent data release 1 ( dr1 ) @xcite , so the preliminary magnitudes as presented in this paper , which are based upon a pre - dr1 calibration , will be denoted as @xmath22 , @xmath23 , @xmath24 , @xmath25 , and @xmath26 ( for more details see @xcite ) . the system used in this paper should differ absolutely from the final ( @xmath27 ) sdss photometric system by only a few percent .
[ see @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , and @xcite for further details about the photometric and spectroscopic surveys . ] the base galaxy catalog for the present study was generated from the spring equatorial scan subset of the sdss edr data @xcite . instead of using the star - galaxy classifications from the standard sdss imaging pipeline ( photo ) , however , we chose to use the more robust classifications measured by the code of @xcite .
galaxies were separated from stars using a bayesian method which combines the automated classifications from the photometric pipelines with a correction for variations in seeing and assigns a probability for each object being a galaxy @xcite .
the accurate astrometry @xcite combined with color and morphological information derived from the sdss digital images @xcite allows for robust star - galaxy separation to a limiting magnitude of @xmath28 ( with 1.75 arcsecond or better seeing ) .
very few objects have an uncertain star - galaxy probability ( near 50% ) ; thus we have simply cut at 50% .
this method suffers less at the faint end from stellar contamination than does the standard sdss edr star - galaxy separation .
our earliest cg catalogs using pre - edr versions of the sdss outputs had stellar contamination ( a star identified as a galaxy ) in as many as 25% of the selected cgs . using the @xcite catalog and cutting at a galaxy likelihood of 50%
, our catalogs now have at worst @xmath29 stellar contamination .
this star - galaxy separation breaks down for small , faint galaxies , and routines used for this initial study do not measure bright or large galaxies well .
the combination limits the present study to galaxies within the @xmath24 magnitude range of 14.0 to 21.0 ( which limits the brightest cg member to @xmath24 between 14.0 and 18.0 , see [ cat_construct ] ) , excluding much of the hcg catalog but still providing some overlap .
the data presented here are from runs 752 and 756 ( a @xmath30 wide stripe from @xmath31 to @xmath32 centered on @xmath33 = @xmath34 ) which were observed on the nights of 1999 march 20 and 22 .
galaxies from the entire @xmath6 230 sq deg region were used for checking the isolation criteria described in [ cat_construct ] , however , in order to obtain the least stellar contamination of the galaxy sample , only cgs observed in frames with seeing @xmath35 arcseconds were retained for the catalog , yielding an effective area on the sky for our base sample of 153 sq deg ( see fig . [ sky_dist ] ) .
finally , we note that the @xcite galaxy catalog was based upon an earlier processing ( `` rerun '' ) of the imaging data reductions than is presented in the final sdss edr . as a result ,
@xmath36 magnitudes may differ slightly from the values obtained by a direct query of the sdss edr database .
similarly , although each object s run , camera column , and field identification numbers will remain the same , the object identification number within the field will likely differ from the value provided by the sdss edr database .
hickson s @xcite criteria for extracting compact groups from the poss red ( e ) prints include the following : * @xmath37 ( population ) , * @xmath38 ( isolation ) , and * @xmath39 mag arcsec@xmath40 ( compactness ) , where * @xmath41 is the total number of galaxies within 3 magnitudes of the brightest group member , * @xmath42 is the total magnitude of these galaxies per square arcsec averaged over the smallest circle containing their geometric centers ( note that using a mean surface brightness yields , to first order , a distance - independent measure of compactness ) , * @xmath43 is the angular diameter of this smallest circle , and * @xmath44 is the angular diameter of the largest concentric circle that contains no other ( external ) galaxies within this magnitude range or brighter .
searching the poss prints , which cover 67% of the celestial sphere , by eye , hickson found 100 cgs ; the median redshift of this sample is @xmath45 .
to select our sdss cgs we use computer code embodying a slightly modified hickson criteria to extract a cg catalog from this galaxy catalog : * @xmath46 , * @xmath47 , * @xmath48 , and * @xmath49 mag arcsec@xmath40 ( in sdss @xmath24 band ) , where @xmath24 is the sdss @xmath14-band _ model _ magnitude , which tends to be more robust than the sdss @xmath14-band _ petrosian _ magnitude for galaxies fainter than @xmath50 ( see @xcite for definitions of the various sdss magnitudes ) .
three changes were made from the hickson criteria .
the first merely added an upper limit of @xmath51 to the number of members in a cg .
this additional limit was added in order to constrain the run - time of the group - finding algorithm , which is basically an @xmath52 process .
since the most populous group in our catalog contains only @xmath53 members , this was ( intentionally ) not a very restrictive constraint .
the second modification simply requires that @xmath44 is strictly greater than @xmath54 instead of greater than or equal to ; this change simplified the coding of the algorithm which identifies isolated groups and finds the smallest enclosing circle .
the third , the change from @xmath55 to @xmath56 mag arcsec@xmath40 , resulted from tests suggested by iovino ( private communication , see also @xcite ) to reduce the rate of false cg detection .
following the approach described by @xcite and @xcite , we estimated the number of non - real cgs produced through random alignments as a function of @xmath57 , and compared it with the number of cgs found in the original sdss catalog . using a subset of our galaxy catalog
, we generated a randomized galaxy catalog by assigning to each galaxy an ra and dec drawn at random from within the area of the subset ; a value for the seeing at this random position was taken to be that for the nearest galaxy to this position in the real catalog .
the compact group finding code was then run on this randomized galaxy catalog to generate a list of false cgs produced by projection effects , and this fake cg catalog was compared to the portion of our real cg catalog from the same region . in fig
( [ fig1],a ) shows the histogram of surface brightness for cgs in each catalog .
the ratio of random to real cgs is @xmath58 4% when restricting the search to cgs with @xmath59 24.0 , while the ratio increases to 55% within 24.0@xmath0 @xmath60 25.0 and 80% within 25.0@xmath0 @xmath60 26.0 . from this plot
it is clear that cutting at @xmath59 24 leaves a much more clean ( and compact ) catalog . in fig ( [ fig1],b )
we show a histogram of the difference in magnitude between the dimmest and brightest group member ( @xmath61 ) where the maximum allowed by the hickson criteria is 3 .
in contrast with iovino et al .
( 2003 ) , for our sample we found no strong trend in the contamination rate as a function @xmath61 .
there may be a small offset between the peaks , but a large gain in detection efficiency was unlikely to come from modifying this selection parameter in our algorithm .
we therefore restricted the surface brightness limit to be @xmath62 24 mag arcsec@xmath40 , and left @xmath61 unchanged , in order to reduce the contamination rate of our sample to reasonable values .
in addition , we examined the isolation criterion , which required that the next closest galaxy bright enough to be in the group or brighter be at least three group radii away .
allam & tucker ( 2000 ) found that one third of their compact groups are part of larger structures , also de carvalho et al .
( 1997 ) and ribeiro et al .
( 1998 ) found that a large fraction of hcgs are part of larger structures , and only a small number of groups can be defined as truly isolated . to overcome this problem , iovino et al .
( 2003 ) adopt a flexible isolation criterion , such that no galaxy of magnitude brighter than mag of the faintest cg member + 0.5 mag is within the isolation radius .
our algorithm , which builds groups based on adding the closest galaxies which are in the correct magnitude range to be members , did not allow any flexibility in lowering the isolation criteria .
however we did note that the majority of the closest non - member galaxies ( which were eligible for membership ) for our groups were within @xmath630.5 magnitudes of the dimmest group member ( 73% ) and were between three and four group radii of the center of the group ( 51% ) .
( numbers for the random catalog were similar , at 73% within @xmath630.5 magnitudes and 40% nearby , thus relaxing isolation will add random alignments in nearly equal proportion to real groups . ) relaxing the isolation criteria for dim galaxies should be explored in our future efforts , as it seems likely the strict isolation criteria used excluded some compact groups .
using our modified hickson criteria , we found a total of 232 candidate cgs in 153 sq deg . in order to identify any potential weaknesses of our selection technique and improve future versions ,
each of these candidate cgs was inspected by eye , and much of the sdss @xmath14 band imaging data ( the corrected frames ) for these candidate cgs were re - analyzed using gaia @xcite and sextractor @xcite to double - check galaxy identification . of these 232 candidate cgs , 175 survived inspection to be included in the final cg catalog .
of the 57 which were discarded , 55 were found to be false detections due to poor deblending of saturated bright stars or angularly large galaxies and/or poor magnitude determination of faint diffuse objects in the original sdss image processing .
another 2 candidate groups were found to contain a qso which was mis - classified by photo and sextractor as a galaxy .
these 2 groups are listed as tri since they contain only 3 galaxies .
our final catalog of sdss cgs , extracted from 153 sq deg of sky from runs 752 run 756 of the @xcite galaxy catalogs in the sdss edr , contains a total of 744 galaxies in 175 cgs .
of the 744 cg member galaxies , 158 have spectroscopic redshifts ( 135 from the sdss itself ) . despite this low number , 131 ( 75% ) of sdss cgs contain at least one group member with a spectroscopic redshift .
the properties of the cgs are described in table [ tab_cg ] ; the properties of the cg member galaxies are described in table [ tab_cgmem ] . in table
[ tab_cg ] we list the general properties of all 175 sdsscgs ( plus two triplet systems ) : * a running identification number . *
@xmath64 , the ra ( j2000.0 ) of the center of the smallest circle containing all group members within 3 magnitudes of the brightest group member . *
@xmath65 , the dec ( j2000.0 ) of the center of the smallest circle containing all group members within 3 magnitudes of the brightest group member .
* @xmath43 , the angular diameter of this smallest circle ( in arcsec ) . *
@xmath42 , the total @xmath24 magnitude of the @xmath41 [ column ( 6 ) ] group members averaged over the smallest circle containing their geometric centers .
* @xmath41 , the total number of galaxies in the group within 3 magnitudes of the brightest group member . * @xmath66 , the difference in the @xmath24 magnitude between the brightest group member and the faintest . *
@xmath67 , the ratio of @xmath44 , the angular diameter of the largest concentric circle that contains no other ( external ) galaxies within 3 magnitudes of the brightest group member , to @xmath43 , the angular diameter of the smallest circle containing all group members within this magnitude range [ column ( 4 ) ] .
* @xmath68 , the photometric redshift of the brightest group member as determined by the method described in [ redshift_dist ] . *
@xmath69 , the spectroscopic redshift for the group based upon the median of the spectroscopic redshifts of its member galaxies ( see table [ tab_cgmem ] ) . if available , a group member s sdss spectroscopic redshift was used ; otherwise , if available , its spectroscopic redshift from ned ( the nasa extragalactic database ) , was used .
* @xmath70 , the number of group member spectroscopic redshifts used in determining @xmath69 . * comments * * a faint background galaxy is visible which could be a member of the group * * two faint background galaxies are visible which could be members of the group * * faint background galaxies are visible which could be members of the group * * galaxies a and b appear to be merging * * galaxies a and b appear to be interacting * * galaxy a shows a tidal tail * * galaxies a , b , and c appear to be embedded within a common halo * * galaxy a is classified as a qso in ned in table [ tab_cgmem ] we list the general properties of the individual member galaxies in each of the 175 sdsscgs ( and the 2 triplet systems ) : * name of the group member , composed of the cg identification number ( from column ( 1 ) of table [ tab_cg ] ) and an identification letter for galaxy ( where `` a '' is the brightest group member , `` b '' the second brightest , ... ) .
* @xmath71 , the ra ( j2000.0 ) of the galaxy . *
@xmath33 , the dec ( j2000.0 ) of the galaxy . * the sdss @xmath14-band model magnitude @xmath72 ( non - de - reddened ) . * the estimated rms error in the sdss @xmath14-band model magnitude @xmath73 . * the reddening in @xmath72 , @xmath74 , as estimated from the @xcite reddening maps .
* @xmath75 , where @xmath22 and @xmath23 are , respectively , the ( non - de - reddened ) sdss @xmath12-band and @xmath13-band model magnitudes . *
@xmath76 , the rms error in @xmath75 [ column ( 8) ] estimated by adding the rms errors in @xmath22 and @xmath23 in quadrature . *
@xmath77 , where @xmath23 and @xmath24 are , respectively , the ( non - de - reddened ) sdss @xmath13-band and @xmath14-band model magnitudes . *
@xmath78 , the rms error in @xmath77 [ column ( 10 ) ] estimated by adding the rms errors in @xmath23 and @xmath24 in quadrature . *
@xmath79 , where @xmath24 and @xmath25 are , respectively , the ( non - de - reddened ) sdss @xmath14-band and @xmath15-band model magnitudes . *
@xmath80 , the rms error in @xmath79 [ column ( 12 ) ] estimated by adding the rms errors in @xmath24 and @xmath25 in quadrature . *
@xmath81 , where @xmath25 and @xmath26 are , respectively , the ( non - de - reddened ) sdss @xmath15-band and @xmath16-band model magnitudes . *
@xmath82 , the rms error in @xmath81 [ column ( 14 ) ] estimated by adding the rms errors in @xmath25 and @xmath26 in quadrature .
* @xmath83 , the spectroscopic redshift for the galaxy as measured by the sdss , if available . of the 744 cg members ,
135 have sdss spectroscopic redshifts .
* @xmath84 , the spectroscopic redshift for the galaxy from the nasa extragalactic database ( ned ) , if available . of the 744 cg members ,
49 have spectroscopic redshifts from ned . of these 49
, 23 have no corresponding spectroscopic redshift from the sdss . * the source name of the ned - derived spectroscopic redshift [ column ( 16 ) ] . for each of the sdsscgs ( and the two triplets )
we have prepared an atlas image ( see fig [ fig : cg ] ) .
these atlas images have the following properties : * the size of each image is 3@xmath85 the group radius ; * the images were obtained from the @xmath14-band sdss corrected frames ; * each image is labeled by the name of the group , which has the format `` sdss0 cg@xmath86 '' , where `` sdss0 '' refers to the fact that this catalog is based on sdss commissioning data ( `` data release 0 '' ) , and where @xmath86 refers to the running i d number of the group ; * each group member is identified by a `` @xmath85 '' mark and labeled by its i d letter ( a , b , c , d , ... ) ; * the center of the group is marked by a square box ; and * a circle of diameter @xmath43 ( as defined in [ cat_construct ] ) is also drawn .
as noted at the beginning of this section , only 158 of the 744 cg member galaxies have spectroscopic redshifts either from the sdss or from the literature ( see fig .
[ fig5 ] ) .
our magnitude cut for cg galaxies is @xmath87 ( [ cat_construct ] ) , whereas the magnitude limit for the main galaxy sample of the sdss spectroscopic program is @xmath88 [ table 29 of @xcite ] .
further complicating matters is an sdss single plate target proximity limit of 55 arcsec to avoid fiber collisions @xcite , which in general would allow only a single galaxy from a compact group to be targeted on a spectroscopic plate .
therefore , it is not surprising that so few of our cg galaxies have spectroscopically determined redshifts .
nonetheless , these 158 galaxies are spread fairly uniformly over the sdss cg sample ; so , a full 75% of our cgs ( 131 out of 175 ) contain at least one group member with a spectroscopically determined redshift .
we were , however , able to obtain a photometric redshift for every sdss cg ( see table [ tab_cg ] ) .
we did this by taking the photometric redshift of the brightest group member as the photometric redshift of the cg as a whole .
these brightest group member photometric redshifts were determined by a variation on the polynomial method of @xcite . instead of using polynomials ,
however , pad approximants @xcite were used .
pad approximants , which have seen applications in statistical mechanics , critical phenomena , and circuit design @xcite , are basically ratios of polynomials . as a training sample , @xmath58 500
main sample galaxies @xcite and @xmath58 500 luminous red galaxies ( lrgs ) @xcite were taken from the sdss spectroscopic sample ; the low - redshift training set sample was supplemented with an additional @xmath58 500 galaxies from a special spectroscopic plate which attempts to sample fully from the bright end of the luminosity function at @xmath89 @xcite .
details on this method can be found elsewhere @xcite . in calculating the photometric redshifts of the cgs
, we ignored the non - first - ranked group members since using them tended to result in systematic errors of @xmath90 or more .
this is likely due to the fact that 85% of the non - first - ranked galaxies in our sample are fainter than @xmath91 , where the training set for our photometric redshift relation is dominated by lrgs , a type of galaxy which probably does not describe these fainter cg galaxies very well as a whole .
in contrast , only 16% of brightest group members are fainter than @xmath91 and none fainter than @xmath92 ; furthermore , since lrgs have luminosities and spectral energy distributions similar to brightest cluster members , the lrg training set is probably better suited to our brightest group members than for the fainter group members .
how well does a brightest group member photometric redshift , @xmath93 , track the group s spectroscopic redshift , @xmath94 ?
in figure [ fig : xxx1]a , we plot these two parameters against each other for those 131 sdss cgs which contain at least one member with a spectroscopic redshift .
the relationship is quite good .
the rms of the residual of the unweighted least squares fit of brightest group member photometric redshift to the group spectroscopic redshift is @xmath95 ( fig .
[ fig : xxx1]b ) , which means that the typical @xmath96 rms error in the photometric redshift of an individual group is about 0.037 .
further , if we plot the histograms of @xmath97 and @xmath94 , we find that , overall , the two distributions look quite similar ( fig .
[ fig : xxx2 ] ) . based on the 131 sdss cgs with spectroscopic redshifts
, we can estimate that the median redshift of the full sample of 175 sdss cgs to be @xmath0@xmath98@xmath99@xmath100 .
this estimate , however , may be biased : 25% of the sdss cgs in the full sample of have no spectroscopic redshifts and this subset may be systematically more ( or less ) distant than the rest of the sdss cgs . in figure
[ fig : xxx3 ] we plot the distribution of photometric redshifts for the 131 sdss cgs with spectroscopic redshifts and the distribution of photometric redshifts for the 44 sdss cgs without spectroscopic redshifts .
indeed , the set of 44 sdss cgs without spectroscopic redshifts appears to be , on average , more distant than the set of 131 galaxies with spectroscopic redshifts .
even so , it is a smaller set , and , if we combine both sets of cgs , we find the median photometric redshift for the full sample of 175 sdss cgs to be @xmath0@xmath101@xmath99@xmath102 .
this value is only marginally larger than our estimate for @xmath0@xmath98@xmath99@xmath103 . therefore ,
unless otherwise noted , we will use the spectroscopic redshifts throughout the remainder of the paper .
following @xcite s and @xcite s studies of the local environments surrounding hcgs and scgs , we have undertaken a similar study of the local environments surrounding sdss cgs . for each sdss cg we searched a circular region @xmath104 the group s radius @xmath43 in the sdss edr data base for galaxies in the interval @xmath105 , where @xmath106 and @xmath107 are the sdss @xmath14-band model magnitudes of the brightest and the faintest group members , respectively , for that particular group , as listed in table [ tab_cgmem ] . the surface density of the sdss cg ( @xmath108 )
was obtained by dividing the number of these galaxies within 1 group radius by @xmath109 .
the surface density of galaxies in that sdss cg s local environment @xmath110 was calclated by dividing the number of these galaxies in an annulus @xmath111 the group s radius by the area of this annulus .
the distribution of the ratio @xmath112 is shown in figure [ fig : pho1 ] .
note that the surface densities of sdss cgs are factors of @xmath113 greater than those of their local environments ( on average , they are about a factor of 40 greater ) , similar to what @xcite found for hcgs and what @xcite found for scgs . to test how the local environment of sdss cgs compares with that of the field , we considered the local environments of galaxies from the sdss isolated galaxy catalog @xcite .
for each sdss cg , we selected isolated galaxies of comparable brightness to the brightest group member ( @xmath114 ) and observed under comparable seeing conditions .
we then calculated the surface density of galaxies surrounding of each of these isolated galaxies in an annulus @xmath115 the group s radius , just like we did to calculate @xmath116 for the sdss cg in question .
we then calculated the mean value of the field @xmath110 and its scatter @xmath117 .
figure [ fig : pho2 ] shows the histogram of the quantity @xmath118 .
we find that the environments of sdsscgs are , on average , similar to the environments of field galaxies , although there is considerable scatter .
this result is comparable to those found by @xcite for hcgs and by @xcite for scgs .
it is instructive to compare the properties of sdsscgs with those of cgs from other catalogs . in table
[ tab : cgstat ] we summarize the properties of the current sdsscg catalog and those of six other cg catalogs the initial dposs compact group ( pcg ) catalog by @xcite , the southern compact group ( scg ) catalog by @xcite , the uzc compact group ( uzc - cg ) catalog by @xcite , the las campanas compact group ( lccg ) catalog by @xcite , the redshift survey compact group ( rscg ) catalog by @xcite , and the hickson compact group ( hcg ) catalog by @xcite . of these seven cg catalogs ,
four are based upon galaxy sky positions and photometry alone ( the sdsscg , pcg , scg , and hcg catalogs ) , and three also make use of galaxy redshift information ( the uzc - cg , lccg , and rscg catalogs ) . looking at this table and its companion figures ( figs .
[ statcgsn ] , [ statcgsz ] , & [ statcgsthetag ] ) , it is apparent that sdsscgs are quite similar to most other cg catalogs in terms of their typical group membership ( @xmath41 ) and linear group diameter ( @xmath119 ; fig .
[ statcgsthetag ] ) . with the possible exception of the pcg catalog , however , the sdsscg catalog is on average much deeper than any of the other cg catalogs .
( although we do not have redshift information for the scg catalog , the mean and median angular diameters of its groups , coupled with a galaxy magnitude limit of @xmath120 , indicate that it has a depth comparable to that of the rscg catalog . ) with such a mean depth , the sdsscg catalog is a first step toward the study of the evolutionary properties of cgs .
furthermore , at these depths , the probability of finding cgs acting as strong lenses becomes noticeably higher @xcite ; perhaps a cg from a future sdsscg catalog will be the first to be discovered with giant arcs .
it is also curious to see that , although we used a compactness criterion to detect sdsscgs that was substantially tighter than the original @xcite value , the mean surface brightnesses ( @xmath42 ) in @xmath24 for sdsscgs and hcgs are nearly identical .
recall that the hcg catalog was extracted from the first palomar sky survey ( poss - i ) via a search - by - eye .
@xcite expected , and @xcite confirmed , that the hcg catalog was incomplete approaching the limits of the hickson criteria . here , we appear to be seeing an independent confirmation of the hcg catalog s incompleteness for `` less compact '' cgs as clearly seen in fig [ statcgsmugall ] . finally , although a detailed calculation of the space density of cgs is beyond the scope of this paper , we can perform a rough , `` back - of - the - envelope '' estimate by assuming that a given cg sample is complete out to its median redshift and calculating @xmath121 where @xmath122 is the estimated cg space density , @xmath41 is the total number of cg s in the sample , @xmath123 is the sky area covered by the sample in square degrees , and @xmath124 is the co - moving distance at the median redshift of the survey .
we use a euclidean measure of the volume since we will assume a zero - curvature universe in which @xmath125 and @xmath126 ( see , for example , @xcite ) .
in particular , we wish to compare the space density of sdss cgs with that of the pcg catalog , which is presently the only other deep ( @xmath127 ) , wide area survey for cgs , and with that of the hcg catalog , which remains the benchmark for all cg catalogs . for the sdss cgs ,
we take @xmath128 , @xmath129 sq deg , and @xmath130 ( @xmath131 mpc ) , yielding an estimated space density of @xmath132 mpc@xmath133 . using the above equation
, we can also estimate the space density of the pcg catalog .
for the pcgs , we take @xmath134 , @xmath135 sq deg , and @xmath136 ( @xmath137 mpc ) , yielding an estimated space density of @xmath138 mpc@xmath133 . clearly , the space density of sdss cgs is about a factor of ten greater than that for pcgs . the pcg catalog , however , has a much tighter mean group surface brightness constraint .
if we consider just the 14 sdss cgs that satisfy the pcg surface brightness criterion , we find a space density for `` pcg - like '' sdss cgs of @xmath139 mpc@xmath133 almost identical to the value obtained for the original pcg sample .
this is a good consistency check between these two modern cg catalogs .
comparing the space density of cgs with that of hcgs is a bit harder .
since the hcg catalog is known to be incomplete , we can not merely use equation [ eqn : ncg ] . using simulations
, @xcite estimated that the space density for hcgs is @xmath140 mpc@xmath133 , or about one - third the space density of sdss cgs .
worse yet , sdss cgs have a more stringent surface brightness criterion .
we find that 63% ( 58 out of 92 ) of hcgs meet the sdss cg surface brightness constraint .
thus , the space density of sdss cg - like hcgs should be about @xmath141 mpc@xmath133 , or only about one - fifth to one - fourth the space density of sdss cgs .
four possibilities come readily to mind to explain this discrepancy .
( 1 ) the hcg catalog may still be substantially incomplete even for surface brightnesses of @xmath142 .
if so , a more complete sample of hcg - like groups would have a higher space density .
( 2 ) @xcite s estimate of the space density of hcgs is itself uncertain and is very sensitive to the assumptions regarding the completeness of the hcg sample @xcite .
therefore , the discrepancy may not be very statistically significant .
( 3 ) cosmic variance may play an effect .
out to their median redshifts , the hcg and the sdss cg volumes are both roughly equivalent to boxes only 100@xmath143 mpc on a side .
( 4 ) although much care was taken to avoid contamination by chance projections ( [ cat_construct ] ) , the sdss cg catalog itself probably contains some small fraction of spurious groups , thus artificially increasing our estimated space density of sdss cgs .
we suspect that some combination of these four possibilities will eventually explain the present discrepancy .
to search for environmental effects on the morphologies of cg galaxies , we have used the sub - sample of 158 sdsscg galaxies which have known spectroscopic redshifts . in order to compare this sample fairly
, we have constructed a field sample by selecting for each of these 158 sdsscg galaxies the ten galaxies with the nearest redshifts to it from the sdss edr @xcite .
since these ten galaxies will tend to be randomly spread in ra and dec over the sdss edr survey area , they will reasonably sample the field with little contamination from clusters .
( recall that only about 10% of galaxies lie within rich clusters ) .
we have thus collected a field sample of 1580 galaxies which has a redshift distribution essentially identical to that of the original 158 sdsscg galaxies . that we met this goal
is attested to by the fact that the means and medians of the two redshift distributions are identical ( table [ tab : meanmediancompare ] ) .
furthermore , a one - dimensional kolmogorov - smirnov ( ks ) test @xcite indicates that these two redshift distributions are drawn from the same parent distribution with an extremely high probability ( table [ tab:1dks ] ) .
a galaxy s color is a function of current star formation rate , and , as such , its color can be used as a rough indication of morphological type .
red galaxies tend to be ellipticals and s0 s ; blue galaxies tend to be spirals and irregulars .
the sdss data set contains not one but five filters worth of photometry , which can define a set of four non - redundant colors : @xmath144 , @xmath145 , @xmath146 , and @xmath147 .
we will use this plethora of color information to compare our two samples . for our comparisons
, we construct color - color and color - magnitude diagrams for these two samples . to do this , though , we must first correct the galaxy magnitudes and colors for interstellar absorption and for cosmological effects . to correct for interstellar absorption ,
the following reddening corrections @xcite were subtracted from the galaxy magnitudes : @xmath148 where the values for @xmath149 were obtained from the @xcite reddening maps .
( note that we correct only for interstellar absorption due to the milky way galaxy and do not attempt to correct for internal interstellar absorption associated with the individual galaxies themselves . )
we also apply a k - correction to convert the de - reddened colors and magnitudes to their rest - frame ( i.e. , redshift @xmath150 ) equivalents .
our k - corrections were estimated using the publicly available kcorrect ( v1.10 ) package of @xcite .
this code fits a galaxy s ( redshifted ) broad - band magnitudes to a suite of template galaxy spectral energy distributions in order to reconstruct the galaxy s broad - band magnitudes at another redshift ( e.g. , at a redshift of @xmath150 ) .
finally , to calculate absolute @xmath24 magnitudes , we must assume a cosmological model .
we take the current standard a flat model with @xmath151 , @xmath9 , and @xmath152 km s@xmath4 mpc@xmath4 using the analytical relation by @xcite to calculate luminosity distance @xmath153 .
figure [ gr_vs_z ] plots the de - reddened and k - corrected @xmath154 color against spectroscopic redshift for our two samples . the e / s0 ridgeline the locus of ellipticals and s0 galaxies at @xmath155 is nearly flat for the redshift range plotted , indicating that the k - corrections did a good job .
the slight positive slope of the ridgeline between @xmath150 and @xmath156 may be the result of a slight color evolution in early - type galaxies over this redshift range . figures [ mag_vs_color ] and [ color_vs_color ] display the color - magnitude and the color - color diagrams for the two samples , respectively .
visually , it is hard to distinguish much difference between the sdsscg and field galaxy samples in these plots , with the possible exception that the sdsscg galaxies appear on average slightly redder than the field galaxies in @xmath157 and in @xmath154 . if , however , we look at the statistics of the absolute magnitude and rest - frame color distributions , differences between the two samples become more apparent .
if we look at the means and medians of these distributions for the two samples ( table [ tab : meanmediancompare ] ) , it is clear that the sdsscgs are on average redder than the field galaxies in @xmath157 and in @xmath154 at about the @xmath158 level .
furthermore , there is some evidence at about the @xmath96 level that sdsscgs are on average less luminous than field galaxies ( in the @xmath24-band ) .
we have also run two sets of ks tests @xcite for these two samples : ( 1 ) a set of one - dimensional ks tests on the distributions of @xmath16 , @xmath159 , @xmath157 , @xmath154 , @xmath160 , and @xmath161 ( table [ tab:1dks ] ) ; and ( 2 ) a set of two - dimensional ks tests on the plots shown in figures [ gr_vs_z ] [ color_vs_color ] ( table [ tab:2dks ] ) .
these two sets of ks tests provide strong evidence that the rest - frame colors of cg galaxies do indeed differ from those of field galaxies at least for @xmath157 , @xmath154 , and even @xmath160 .
the distribution of @xmath161 rest - frame colors , however , does not appear to differ significantly between the two galaxy samples .
this is not surprising .
the intrinsic rest - frame @xmath161 color of an elliptical is only about 0.25 mag redder than that of an irregular ; in comparison , the @xmath157 color of an elliptical is about 1.3 mag redder than that of an irregular @xcite .
thus , the @xmath161 colors are not a very sensitive measure of galaxy morphology . from the evidence of their @xmath157 , @xmath154 , and @xmath160 rest - frame colors ,
we conclude that sdsscgs contain a relatively higher fraction of elliptical galaxies than does the field .
why does the relative fraction of elliptical galaxies in sdss cgs appear to be larger than in the field ?
a likely culprit is the cumulative effect of interactions and mergers over the course of a typical cg lifetime .
n - body simulations like those pioneered by @xcite indicate that the end - product of merging spirals can be an elliptical galaxy .
there is evidence for interactions and/or mergers in sdss cgs . while culling the original set of 232 candidate sdss cgs to the final set of 175 ( [ cat_construct ] ) , the member galaxies were visually inspected for tidal tails , bridges , and obvious other signs of interaction activity .
those sdss cgs showing evidence of interactions and/or mergers are indicated in the `` comments '' column ( column 12 ) of table [ tab_cg ] .
evidence of some sort of interaction or tidal tail is seen in 55 sdss cgs ( 31% of the final catalog of 175 sdss cgs ) ; full - blown merger events appear to be occurring in 26 sdss cgs ( 14% ) .
a luminous halo enveloping the all or part of the group perhaps the remnant of a completely disrupted galaxy appears in 4 of the sdss cgs ( 2% ) .
@xcite estimated that roughly 7% of the galaxies in hcgs are in the process of merging .
his conclusion was based on roughly consistent frequencies of ( a ) optical signatures of merging , ( b ) warm far - infrared colors , and ( c ) sinusoidal rotation curves .
our initial results , described above , indicate comparable or even greater levels of merger activity in the sdss cg sample .
we have presented a new catalog of cgs , one extracted via an objective algorithm from 153 sq deg of runs 752 and 756 in the sdss edr @xcite .
the algorithm , using a modified form of hickson s @xcite original criteria , detected 175 cgs down to a limiting galaxy magnitude of @xmath162 .
we have estimated that the median redshift of this current version of the sdss cg catalog is @xmath163 , substantially deeper than previous cg catalogs .
this catalog will be useful for conducting studies of the general characteristics of cgs , their environments , and their component galaxies .
our initial results show that sdss cgs are , on average , about a factor of 40 denser than their local surroundings ; that their general physical properties are similar to those of other , less deep cg catalogs ; that the fraction of early - type galaxies is higher in sdss cgs than in the field ; and that there is strong visual evidence of interactions and mergers within a significant fraction of sdss cgs .
our future goals are three - fold : further analyses of the properties of the current version of the sdss cg catalog with its 175 cgs , the enlargement of the sample of sdss cgs as more sdss data become available , and improvement of cg selection techniques .
clearly , the next step is to apply our algorithm to the current sdss data release . at a detection rate of slightly better than 1 cg per square degree
, we expect that the final sdss cg catalog , based upon the a completed sdss covering up to one - quarter of the sky , will contain on the order of 5,000 10,000 cgs .
funding for the creation and distribution of the sdss archive has been provided by the alfred p. sloan foundation , the participating institutions , the national aeronautics and space administration , the national science foundation , the u.s .
department of energy , the japanese monbukagakusho , and the max planck society .
the sdss web site is http://www.sdss.org/. the sdss is managed by the astrophysical research consortium ( arc ) for the participating institutions .
the participating institutions are the university of chicago , fermilab , the institute for advanced study , the japan participation group , the johns hopkins university , los alamos national laboratory , the max - planck - institute for astronomy ( mpia ) , the max - planck - institute for astrophysics ( mpa ) , new mexico state university , university of pittsburgh , princeton university , the united states naval observatory , and the university of washington .
this research has made use of the nasa / ipac extragalactic database ( ned ) , which is operated by the jet propulsion laboratory , at the california institute of technology , under contract with the national aeronautics and space administration .
lupton , r. , gunn , j. e. , ivezi , z. , knapp , g. r. , kent , s. , & yasuda , n. 2001 , asp conf .
238 , astronomical data analysis software and systems x , eds . f. r. harnden , jr . , f. a. primini , h. e. payne , ( san francisco : asp ) , p. 269
@l@c@c@c@c@c@c@c@c@c@c@c@c@c@c@r@r@r@ 1a & 09:40:49.77 & -00:51:47.07 & 16.707 & 0.007 & 0.108 & 2.219 & 0.093 & 1.134 & 0.010 & 0.498 & 0.008 & 0.349 & 0.008 & & & + 1b & 09:40:49.80 & -00:51:26.92 & 18.123 & 0.012 & 0.108 & 2.047 & 0.231 & 1.129 & 0.022 & 0.480 & 0.015 & 0.344 & 0.024 & & & + 1c & 09:40:47.40 & -00:51:39.69 & 18.487 & 0.024 & 0.107 & 1.187 & 0.138 & 0.633 & 0.034 & 0.328 & 0.032 & 0.107 & 0.073 & & & + 1d & 09:40:48.99 & -00:51:41.70 & 19.655 & 0.029 & 0.108 & 0.847 & 0.133 & 0.537 & 0.043 & 0.393 & 0.041 & 0.162 & 0.091 & & & + & & & & & & & & & & & & & & & & + 2a & 09:41:04.64 & -00:28:13.50 & 16.895 & 0.008 & 0.158 & 2.072 & 0.103 & 1.155 & 0.010 & 0.461 & 0.008 & 0.382 & 0.010 & & & + 2b & 09:41:06.09 & -00:28:24.50 & 17.818 & 0.012 & 0.158 & 1.488 & 0.104 & 1.009 & 0.016 & 0.545 & 0.013 & 0.403 & 0.018 & 0.1478 & & + 2c & 09:41:05.90 & -00:28:28.81 & 18.309 & 0.020 & 0.157 & 1.006 & 0.106 & 0.650 & 0.026 & 0.459 & 0.025 & 0.058 & 0.055 & & & + 2d & 09:41:05.39 & -00:28:16.93 & 18.968 & 0.017 & 0.158 & 1.835 & 0.287 & 1.096 & 0.027 & 0.470 & 0.020 & 0.327 & 0.035 & & & + & & & & & & & & & & & & & & & & + 3a & 09:49:58.96 & -00:52:49.22 & 16.255 & 0.012 & 0.178 & 0.915 & 0.042 & 0.785 & 0.016 & 0.250 & 0.014 & 0.440 & 0.021 & & & + 3b & 09:49:58.48 & -00:52:59.32 & 16.520 & 0.008 & 0.178 & 2.066 & 0.051 & 0.984 & 0.009 & 0.444 & 0.008 & 0.352 & 0.007 & 0.0916 & 0.0916 & 2masxij0949584 - 0052599 + 3c & 09:49:58.24 & -00:52:54.89 & 18.877 & 0.016 & 0.178 & 1.864 & 0.220 & 1.017 & 0.027 & 0.375 & 0.020 & 0.179 & 0.041 & & & + 3d & 09:49:57.36 & -00:53:05.14 & 19.069 & 0.015 & 0.179 & 1.953 & 0.197 & 0.831 & 0.024 & 0.414 & 0.019 & 0.398 & 0.031 & & & + & & & & & & & & & & & & & & & & + 4a & 09:56:09.83 & 00:19:33.02 & 16.557 & 0.012 & 0.083 & 1.276 & 0.021 & 0.273 & 0.012 & 0.207 & 0.012 & 0.036 & 0.021 & 0.0354 & & + 4b & 09:56:09.62 & 00:19:29.87 & 17.937 & 0.013 & 0.083 & 0.333 & 0.013 & -0.267 & 0.014 & 0.078 & 0.017 & 0.213 & 0.045 & & & + 4c & 09:56:08.68 & 00:19:38.33 & 18.189 & 0.013 & 0.083 & 1.050 & 0.034 & 0.186 & 0.014 & 0.207 & 0.016 & 0.179 & 0.040 & & 0.0352 & 2dfgrsn351z064 + 4d & 09:56:08.80 & 00:19:33.51 & 19.086 & 0.021 & 0.083 & 1.074 & 0.097 & 0.506 & 0.026 & 0.360 & 0.026 & 0.218 & 0.066 & & & + & & & & & & & & & & & & & & & & + 5a & 09:57:32.96 & 00:40:05.41 & 17.393 & 0.011 & 0.098 & 0.942 & 0.021 & 0.274 & 0.011 & 0.323 & 0.012 & -0.483 & 0.033 & 0.0870 & & + 5b & 09:57:32.85 & 00:40:04.25 & 17.452 & 0.011 & 0.098 & 1.362 & 0.054 & 0.851 & 0.014 & 0.657 & 0.012 & 0.499 & 0.011 & & & + 5c & 09:57:32.32 & 00:39:58.29 & 18.234 & 0.010 & 0.097 & 2.008 & 0.095 & 0.971 & 0.013 & 0.419 & 0.012 & 0.349 & 0.013 & & & + 5d & 09:57:32.93 & 00:40:09.34 & 18.269 & 0.017 & 0.099 & 0.743 & 0.047 & 0.285 & 0.020 & 0.226 & 0.024 & 0.011 & 0.058 & & & + & & & & & & & & & & & & & & & & + 6a & 10:00:25.22 & 00:14:57.67 & 17.835 & 0.011 & 0.074 & 2.338 & 0.232 & 1.389 & 0.017 & 0.533 & 0.012 & 0.446 & 0.017 & 0.2195 & & + 6b & 10:00:24.84 & 00:15:10.64 & 17.866 & 0.014 & 0.075 & 1.950 & 0.225 & 1.268 & 0.022 & 0.530 & 0.016 & 0.434 & 0.027 & & & + 6c & 10:00:25.36 & 00:14:53.01 & 18.747 & 0.015 & 0.074 & 1.491 & 0.184 & 1.423 & 0.027 & 0.506 & 0.018 & 0.422 & 0.029 & & & + 6d & 10:00:24.27 & 00:15:02.94 & 18.750 & 0.017 & 0.074 & 2.410 & 0.446 & 1.221 & 0.029 & 0.523 & 0.021 & 0.468 & 0.035 & & & + 6e & 10:00:24.81 & 00:15:02.81 & 20.822 & 0.043 & 0.074 & 2.564 & 0.814 & 1.082 & 0.076 & 0.627 & 0.053 & 0.563 & 0.087 & & & + & & & & & & & & & & & & & & & & + 7a & 10:01:24.49 & 01:07:23.66 & 16.737 & 0.007 & 0.056 & 1.961 & 0.059 & 0.879 & 0.008 & 0.421 & 0.008 & 0.418 & 0.008 & & & + 7b & 10:01:24.35 & 01:07:31.56 & 18.509 & 0.012 & 0.056 & 1.659 & 0.110 & 0.844 & 0.015 & 0.404 & 0.014 & 0.444 & 0.022 & & & + 7c & 10:01:25.82 & 01:07:40.52 & 18.576 & 0.018 & 0.056 & 1.583 & 0.201 & 0.894 & 0.027 & 0.266 & 0.024 & 0.239 & 0.050 & & & + 7d & 10:01:25.44 & 01:07:32.10 & 19.562 & 0.026 & 0.056 & 1.702 & 0.362 & 0.840 & 0.041 & 0.414 & 0.036 & 0.302 & 0.077 & & & + & & & & & & & & & & & & & & & & + 8a & 10:01:26.92 & 00:20:39.08 & 17.539 & 0.011 & 0.074 & 2.398 & 0.233 & 1.433 & 0.016 & 0.564 & 0.011 & 0.396 & 0.015 & & & + 8b & 10:01:27.34 & 00:20:54.56 & 17.626 & 0.011 & 0.074 & 2.545 & 0.311 & 1.402 & 0.018 & 0.569 & 0.013 & 0.326 & 0.020 & 0.2316 & & + 8c & 10:01:26.63 & 00:20:55.17 & 19.420 & 0.024 & 0.074 & 1.599 & 0.326 & 1.171 & 0.042 & 0.581 & 0.029 & 0.280 & 0.063 & & & + 8d & 10:01:26.36 & 00:20:39.56 & 19.791 & 0.029 & 0.075 & 1.621 & 0.439 & 1.210 & 0.055 & 0.576 & 0.036 & 0.294 & 0.079 & & & + 8e & 10:01:26.39 & 00:20:32.57 & 19.834 & 0.029 & 0.075 & 2.204 & 0.703 & 1.213 & 0.054 & 0.665 & 0.035 & 0.359 & 0.067 & & & + & & & & & & & & & & & & & & & & + 9a & 10:04:07.74 & 00:17:46.52 & 16.308 & 0.009 & 0.085 & 1.994 & 0.048 & 0.949 & 0.010 & 0.485 & 0.009 & 0.397 & 0.008 & & & + 9b & 10:04:08.60 & 00:17:56.55 & 17.357 & 0.010 & 0.085 & 2.143 & 0.125 & 1.149 & 0.013 & 0.515 & 0.011 & 0.411 & 0.014 & 0.1398 & & + 9c & 10:04:07.27 & 00:17:43.80 & 18.523 & 0.015 & 0.085 & 1.549 & 0.113 & 0.732 & 0.019 & 0.366 & 0.019 & 0.266 & 0.041 & & & + 9d & 10:04:06.54 & 00:18:14.45 & 19.290 & 0.020 & 0.085 & 1.994 & 0.597 & 1.644 & 0.048 & 0.628 & 0.024 & 0.336 & 0.047 & & & + & & & & & & & & & & & & & & & & + 10a & 10:06:02.12 & 01:12:04.22 & 16.871 & 0.008 & 0.086 & 1.781 & 0.075 & 1.070 & 0.010 & 0.517 & 0.009 & 0.396 & 0.010 & & & + 10b & 10:06:01.79 & 01:11:58.97 & 17.910 & 0.009 & 0.086 & 1.730 & 0.106 & 1.205 & 0.013 & 0.506 & 0.010 & 0.396 & 0.013 & & & + 10c & 10:06:01.21 & 01:11:48.95 & 19.641 & 0.026 & 0.086 & 1.133 & 0.283 & 1.123 & 0.047 & 0.484 & 0.035 & 0.386 & 0.067 & & & + 10d & 10:06:00.13 & 01:11:55.46 & 19.813 & 0.032 & 0.086 & 1.285 & 0.281 & 0.703 & 0.046 & 0.450 & 0.044 & 0.382 & 0.087 & & & + & & & & & & & & & & & & & & & & + 11a & 10:08:19.97 & 00:12:44.59 & 16.378 & 0.008 & 0.079 & 2.046 & 0.045 & 1.020 & 0.009 & 0.468 & 0.008 & 0.388 & 0.007 & 0.0935 & & + 11b & 10:08:20.72 & 00:12:58.35 & 18.592 & 0.015 & 0.079 & 1.881 & 0.182 & 0.938 & 0.021 & 0.423 & 0.018 & 0.333 & 0.036 & & & + 11c & 10:08:19.49 & 00:12:54.47 & 18.788 & 0.017 & 0.079 & 1.832 & 0.211 & 0.922 & 0.025 & 0.378 & 0.021 & 0.345 & 0.046 & & & + 11d & 10:08:20.62 & 00:12:54.61 & 19.217 & 0.026 & 0.079 & 1.393 & 0.211 & 0.798 & 0.037 & 0.334 & 0.034 & 0.504 & 0.071 & & & + & & & & & & & & & & & & & & & & + 12a & 10:08:21.77 & 01:11:26.90 & 17.126 & 0.012 & 0.102 & 1.942 & 0.244 & 1.406 & 0.021 & 0.510 & 0.014 & 0.307 & 0.024 & & & + 12b & 10:08:20.74 & 01:11:10.25 & 18.483 & 0.013 & 0.102 & 1.888 & 0.213 & 1.194 & 0.020 & 0.575 & 0.015 & 0.384 & 0.023 & & & + 12c & 10:08:21.75 & 01:11:24.29 & 19.162 & 0.027 & 0.102 & 2.136 & 0.622 & 1.021 & 0.046 & 0.775 & 0.033 & 0.504 & 0.049 & & & + 12d & 10:08:21.23 & 01:11:00.41 & 19.374 & 0.018 & 0.102 & 2.713 & 0.777 & 1.364 & 0.034 & 0.588 & 0.022 & 0.378 & 0.038 & & & + & & & & & & & & & & & & & & & & + 13a & 10:08:36.40 & -00:04:24.23 & 17.039 & 0.010 & 0.079 & 2.055 & 0.094 & 1.239 & 0.012 & 0.523 & 0.010 & 0.397 & 0.010 & 0.1844 & & + 13b & 10:08:37.33 & -00:04:36.72 & 19.009 & 0.019 & 0.079 & 2.671 & 0.648 & 1.237 & 0.034 & 0.575 & 0.022 & 0.387 & 0.037 & & & + 13c & 10:08:37.07 & -00:04:17.30 & 19.316 & 0.025 & 0.079 & 1.929 & 0.435 & 1.077 & 0.044 & 0.554 & 0.031 & 0.326 & 0.059 & & & + 13d & 10:08:37.51 & -00:04:33.93 & 19.865 & 0.027 & 0.079 & 1.445 & 0.307 & 1.093 & 0.048 & 0.499 & 0.034 & 0.381 & 0.064 & & & + & & & & & & & & & & & & & & & & + 14a & 10:10:56.62 & 00:15:22.76 & 17.148 & 0.010 & 0.084 & 1.962 & 0.114 & 1.271 & 0.013 & 0.502 & 0.011 & 0.384 & 0.014 & 0.1862 & & + 14b & 10:10:56.20 & 00:15:12.60 & 19.715 & 0.038 & 0.083 & 5.545 & 0.579 & 1.567 & 0.096 & 0.739 & 0.046 & 0.469 & 0.085 & & & + 14c & 10:10:56.72 & 00:14:58.26 & 19.893 & 0.025 & 0.083 & 1.894 & 0.441 & 1.183 & 0.043 & 0.397 & 0.031 & 0.403 & 0.068 & & & + 14d & 10:10:56.32 & 00:15:03.84 & 20.021 & 0.031 & 0.083 & 2.091 & 0.821 & 1.424 & 0.066 & 0.500 & 0.039 & 0.407 & 0.084 & & & + & & & & & & & & & & & & & & & & + 15a & 10:13:39.11 & -00:08:36.04 & 17.657 & 0.010 & 0.095 & 1.902 & 0.067 & 0.900 & 0.011 & 0.380 & 0.011 & 0.343 & 0.013 & & & + 15b & 10:13:38.34 & -00:08:47.50 & 18.177 & 0.011 & 0.095 & 1.917 & 0.102 & 0.964 & 0.014 & 0.461 & 0.012 & 0.378 & 0.016 & & & + 15c & 10:13:39.48 & -00:08:44.58 & 19.660 & 0.033 & 0.094 & 1.900 & 0.432 & 0.783 & 0.049 & 0.428 & 0.043 & 0.220 & 0.100 & & & + 15d & 10:13:38.93 & -00:08:42.97 & 19.755 & 0.041 & 0.095 & 2.587 & 1.952 & 1.849 & 0.133 & 0.898 & 0.047 & 0.573 & 0.063 & & & + & & & & & & & & & & & & & & & & + 16a & 10:15:38.96 & -00:52:01.09 & 17.791 & 0.011 & 0.126 & 2.167 & 0.159 & 1.172 & 0.016 & 0.497 & 0.012 & 0.365 & 0.015 & & 0.1778 & 2dfgrsn289z111 + 16b & 10:15:39.69 & -00:51:51.88 & 19.064 & 0.020 & 0.126 & 1.472 & 0.180 & 0.824 & 0.032 & 0.394 & 0.026 & 0.265 & 0.053 & & & + 16c & 10:15:40.44 & -00:51:54.89 & 19.819 & 0.036 & 0.127 & 0.652 & 0.206 & 1.024 & 0.070 & 0.403 & 0.049 & 0.469 & 0.087 & & & + 16d & 10:15:39.97 & -00:51:44.38 & 20.021 & 0.028 & 0.126 & 1.489 & 0.367 & 1.118 & 0.058 & 0.447 & 0.037 & 0.417 & 0.066 & & & + 16e & 10:15:38.85 & -00:51:56.95 & 20.319 & 0.046 & 0.126 & 2.661 & 1.184 & 0.849 & 0.083 & 0.468 & 0.063 & 0.118 & 0.147 & & & + & & & & & & & & & & & & & & & & + 17a & 10:21:11.62 & 01:04:49.51 & 17.180 & 0.011 & 0.145 & 1.701 & 0.136 & 1.089 & 0.016 & 0.456 & 0.014 & 0.328 & 0.023 & & & + 17b & 10:21:11.33 & 01:04:52.45 & 17.263 & 0.009 & 0.145 & 2.195 & 0.178 & 1.343 & 0.013 & 0.525 & 0.010 & 0.480 & 0.012 & & & + 17c & 10:21:10.76 & 01:04:46.19 & 17.714 & 0.013 & 0.144 & 1.704 & 0.189 & 1.174 & 0.020 & 0.477 & 0.016 & 0.506 & 0.024 & & & + 17d & 10:21:10.40 & 01:05:06.73 & 19.145 & 0.021 & 0.146 & 2.155 & 0.550 & 1.229 & 0.037 & 0.460 & 0.027 & 0.429 & 0.049 & & & + 17e & 10:21:11.95 & 01:05:08.04 & 19.367 & 0.020 & 0.147 & 2.962 & 1.008 & 1.268 & 0.036 & 0.543 & 0.025 & 0.413 & 0.044 & & & + 17f & 10:21:12.71 & 01:05:06.70 & 19.801 & 0.061 & 0.147 & 1.720 & 0.934 & 0.899 & 0.099 & 0.360 & 0.089 & 0.173 & 0.228 & & & + & & & & & & & & & & & & & & & & + 18a & 10:25:37.41 & -00:33:52.66 & 17.245 & 0.010 & 0.185 & 2.159 & 0.124 & 1.208 & 0.012 & 0.534 & 0.010 & 0.433 & 0.010 & 0.1693 & 0.1693 & 2masxij1025374 - 003353 + 18b & 10:25:37.87 & -00:33:44.26 & 18.010 & 0.012 & 0.186 & 2.483 & 0.266 & 1.237 & 0.016 & 0.527 & 0.013 & 0.375 & 0.016 & & & + 18c & 10:25:38.10 & -00:33:31.49 & 19.716 & 0.023 & 0.188 & 1.489 & 0.288 & 1.144 & 0.038 & 0.532 & 0.028 & 0.348 & 0.047 & & & + 18d & 10:25:38.11 & -00:33:48.01 & 19.858 & 0.037 & 0.186 & 0.625 & 0.895 & 2.610 & 0.229 & 0.875 & 0.044 & 0.581 & 0.053 & & & + 18e & 10:25:37.48 & -00:33:32.96 & 20.242 & 0.052 & 0.187 & 1.528 & 0.545 & 0.795 & 0.078 & 0.350 & 0.073 & 0.537 & 0.127 & & & + & & & & & & & & & & & & & & & & + 19a & 10:25:45.66 & 01:10:27.21 & 17.217 & 0.010 & 0.154 & 1.453 & 0.057 & 0.804 & 0.012 & 0.448 & 0.011 & 0.382 & 0.014 & & & + 19b & 10:25:44.63 & 01:10:28.51 & 17.437 & 0.010 & 0.155 & 1.608 & 0.068 & 0.787 & 0.011 & 0.380 & 0.011 & 0.356 & 0.017 & & & + 19c & 10:25:45.95 & 01:10:46.99 & 19.268 & 0.018 & 0.153 & 1.504 & 0.194 & 0.888 & 0.027 & 0.377 & 0.024 & 0.335 & 0.049 & & & + 19d & 10:25:44.91 & 01:10:41.53 & 20.133 & 0.046 & 0.154 & 0.657 & 0.171 & 0.446 & 0.059 & 0.269 & 0.068 & 0.410 & 0.145 & & & + & & & & & & & & & & & & & & & & + 20a & 10:26:01.52 & -00:29:31.05 & 17.723 & 0.011 & 0.212 & 2.261 & 0.204 & 1.266 & 0.016 & 0.520 & 0.012 & 0.459 & 0.014 & 0.1715 & & + 20b & 10:26:01.33 & -00:29:38.96 & 18.233 & 0.018 & 0.212 & 1.503 & 0.206 & 1.130 & 0.028 & 0.520 & 0.022 & 0.466 & 0.032 & & & + 20c & 10:26:00.79 & -00:29:31.32 & 19.209 & 0.026 & 0.212 & 2.885 & 1.111 & 1.111 & 0.043 & 0.522 & 0.032 & 0.441 & 0.053 & & & + 20d & 10:26:00.92 & -00:29:20.40 & 19.794 & 0.035 & 0.213 & 0.584 & 0.192 & 1.027 & 0.059 & 0.481 & 0.046 & 0.543 & 0.073 & & & + 20e & 10:26:02.29 & -00:29:37.03 & 20.077 & 0.052 & 0.211 & 1.421 & 1.208 & 1.778 & 0.157 & 0.935 & 0.060 & 0.508 & 0.078 & & & + & & & & & & & & & & & & & & & & + 21a & 10:26:13.85 & -00:02:04.47 & 16.924 & 0.012 & 0.156 & 2.254 & 0.134 & 0.965 & 0.014 & 0.449 & 0.013 & 0.420 & 0.016 & 0.1034 & 0.1034 & 2masxij1026137 - 000201 + 21b & 10:26:13.36 & -00:02:10.72 & 17.364 & 0.009 & 0.157 & 1.937 & 0.043 & 0.933 & 0.010 & 0.434 & 0.009 & 0.360 & 0.007 & & & + 21c & 10:26:13.82 & -00:02:00.89 & 18.474 & 0.012 & 0.157 & 1.659 & 0.080 & 0.984 & 0.015 & 0.475 & 0.013 & 0.322 & 0.016 & 0.1048 & & + 21d & 10:26:13.15 & -00:02:32.91 & 19.000 & 0.021 & 0.156 & 1.169 & 0.105 & 0.573 & 0.027 & 0.400 & 0.027 & 0.333 & 0.054 & & & + & & & & & & & & & & & & & & & & + 22a & 10:26:36.45 & 00:44:08.19 & 16.372 & 0.008 & 0.144 & 2.056 & 0.037 & 1.009 & 0.008 & 0.502 & 0.008 & 0.376 & 0.005 & 0.1056 & & + 22b & 10:26:37.44 & 00:43:32.08 & 18.078 & 0.010 & 0.145 & 1.867 & 0.077 & 0.972 & 0.012 & 0.460 & 0.011 & 0.397 & 0.012 & & & + 22c & 10:26:38.25 & 00:43:56.11 & 18.638 & 0.012 & 0.145 & 2.304 & 0.266 & 1.256 & 0.019 & 0.542 & 0.014 & 0.361 & 0.021 & & & + 22d & 10:26:38.10 & 00:43:43.06 & 19.169 & 0.029 & 0.145 & 1.150 & 0.154 & 0.696 & 0.038 & 0.321 & 0.040 & 0.236 & 0.080 & & & + & & & & & & & & & & & & & & & & + 23a & 10:28:33.74 & -00:01:21.63 & 17.672 & 0.013 & 0.189 & 1.372 & 0.097 & 1.196 & 0.018 & 0.508 & 0.014 & 0.463 & 0.018 & & & + 23b & 10:28:34.35 & -00:01:09.32 & 19.072 & 0.033 & 0.188 & 1.069 & 0.330 & 1.282 & 0.064 & 0.496 & 0.042 & 0.340 & 0.085 & & & + 23c & 10:28:33.76 & -00:01:13.26 & 20.376 & 0.041 & 0.188 & 1.331 & 0.538 & 1.306 & 0.081 & 0.565 & 0.051 & 0.395 & 0.092 & & & + 23d & 10:28:34.45 & -00:01:03.64 & 20.592 & 0.043 & 0.188 & 1.001 & 0.286 & 0.846 & 0.065 & 0.241 & 0.062 & 0.758 & 0.101 & & & + & & & & & & & & & & & & & & & & + 24a & 10:34:00.40 & 00:24:39.26 & 17.233 & 0.010 & 0.188 & 2.039 & 0.094 & 1.149 & 0.012 & 0.560 & 0.011 & 0.367 & 0.012 & 0.1499 & & + 24b & 10:33:59.92 & 00:24:42.20 & 18.809 & 0.015 & 0.188 & 2.171 & 0.261 & 1.205 & 0.021 & 0.511 & 0.017 & 0.392 & 0.028 & & & + 24c & 10:34:00.63 & 00:24:22.97 & 18.825 & 0.031 & 0.189 & 0.897 & 0.314 & 1.464 & 0.066 & 0.630 & 0.039 & 0.433 & 0.076 & & & + 24d & 10:34:00.17 & 00:24:33.95 & 19.241 & 0.022 & 0.188 & 1.456 & 0.336 & 1.445 & 0.043 & 0.761 & 0.026 & 0.528 & 0.040 & & & + & & & & & & & & & & & & & & & & + 25a & 10:36:26.28 & 01:10:40.47 & 17.555 & 0.009 & 0.127 & 1.639 & 0.046 & 0.652 & 0.010 & 0.377 & 0.010 & 0.350 & 0.013 & & & + 25b & 10:36:27.52 & 01:10:31.69 & 18.901 & 0.012 & 0.127 & 2.672 & 0.334 & 1.279 & 0.017 & 0.745 & 0.013 & 0.426 & 0.015 & & & + 25c & 10:36:27.15 & 01:10:31.60 & 19.387 & 0.024 & 0.127 & 3.288 & 1.302 & 1.081 & 0.040 & 0.539 & 0.031 & 0.054 & 0.075 & & & + 25d & 10:36:27.35 & 01:10:31.51 & 20.059 & 0.054 & 0.127 & -0.136 & 0.164 & 1.020 & 0.094 & -4.214 & 1.930 & 4.541 & 1.941 & & & + & & & & & & & & & & & & & & & & + 26a & 10:37:22.77 & -00:30:48.46 & 17.048 & 0.011 & 0.192 & 1.349 & 0.043 & 0.764 & 0.013 & 0.509 & 0.012 & 0.370 & 0.012 & 0.1140 & & + 26b & 10:37:23.85 & -00:30:54.92 & 19.402 & 0.046 & 0.191 & 0.933 & 0.236 & 0.608 & 0.062 & 0.368 & 0.063 & 0.344 & 0.127 & & & + 26c & 10:37:22.89 & -00:31:02.72 & 19.441 & 0.027 & 0.192 & 1.005 & 0.124 & 0.504 & 0.033 & 0.327 & 0.035 & 0.192 & 0.079 & & & + 26d & 10:37:23.98 & -00:31:03.41 & 19.834 & 0.032 & 0.191 & 0.906 & 0.117 & 0.249 & 0.039 & 0.319 & 0.045 & 0.233 & 0.102 & & & + & & & & & & & & & & & & & & & & + 27a & 10:41:01.81 & -00:57:08.74 & 16.285 & 0.008 & 0.144 & 1.981 & 0.035 & 1.000 & 0.008 & 0.452 & 0.008 & 0.349 & 0.005 & 0.0869 & 0.0869 & 2masx j10405936 + 0057318 + 27b & 10:40:59.13 & -00:57:28.15 & 16.590 & 0.008 & 0.144 & 1.965 & 0.035 & 0.924 & 0.008 & 0.414 & 0.008 & 0.332 & 0.006 & & & + 27c & 10:40:59.70 & -00:57:40.01 & 17.576 & 0.011 & 0.145 & 1.550 & 0.060 & 0.862 & 0.014 & 0.421 & 0.012 & 0.344 & 0.015 & & & + 27d & 10:40:59.71 & -00:57:32.76 & 18.215 & 0.011 & 0.144 & 1.945 & 0.092 & 0.818 & 0.014 & 0.396 & 0.013 & 0.278 & 0.018 & & & + 27e & 10:40:58.28 & -00:57:32.80 & 18.754 & 0.023 & 0.144 & 1.594 & 0.214 & 0.725 & 0.035 & 0.291 & 0.032 & 0.209 & 0.073 & & & + & & & & & & & & & & & & & & & & + 28a & 10:49:19.43 & 00:14:06.16 & 17.209 & 0.008 & 0.112 & 2.072 & 0.067 & 1.130 & 0.009 & 0.450 & 0.008 & 0.424 & 0.009 & 0.1254 & & + 28b & 10:49:19.03 & 00:14:01.92 & 17.335 & 0.015 & 0.112 & 1.116 & 0.099 & 1.054 & 0.021 & 0.364 & 0.019 & 0.368 & 0.040 & & & + 28c & 10:49:20.13 & 00:14:17.62 & 18.001 & 0.011 & 0.112 & 1.656 & 0.090 & 0.990 & 0.014 & 0.410 & 0.012 & 0.409 & 0.021 & & & + 28d & 10:49:20.68 & 00:14:19.70 & 18.015 & 0.009 & 0.112 & 1.092 & 0.032 & 0.748 & 0.011 & 0.404 & 0.011 & 0.337 & 0.017 & & & + & & & & & & & & & & & & & & & & + 29a & 10:50:36.80 & 00:22:16.41 & 17.263 & 0.009 & 0.121 & 1.956 & 0.081 & 1.105 & 0.011 & 0.482 & 0.010 & 0.370 & 0.012 & 0.1496 & & + 29b & 10:50:36.05 & 00:21:56.74 & 18.679 & 0.015 & 0.121 & 2.267 & 0.294 & 1.031 & 0.022 & 0.535 & 0.018 & 0.426 & 0.033 & & & + 29c & 10:50:36.31 & 00:22:12.57 & 19.445 & 0.028 & 0.121 & 1.497 & 0.268 & 0.865 & 0.039 & 0.515 & 0.035 & 0.600 & 0.062 & & & + 29d & 10:50:36.59 & 00:21:51.39 & 19.786 & 0.027 & 0.121 & 1.337 & 0.230 & 0.910 & 0.038 & 0.473 & 0.034 & 0.453 & 0.070 & & & + 29e & 10:50:37.12 & 00:22:08.42 & 20.099 & 0.037 & 0.121 & 0.766 & 0.301 & 1.329 & 0.072 & 0.454 & 0.049 & 0.454 & 0.108 & & & + 29f & 10:50:35.49 & 00:22:12.67 & 20.184 & 0.058 & 0.120 & 2.023 & 1.791 & 1.799 & 0.161 & 1.201 & 0.063 & 0.589 & 0.077 & & & + & & & & & & & & & & & & & & & & + 30a & 11:01:01.80 & 01:05:47.36 & 16.696 & 0.008 & 0.079 & 2.366 & 0.146 & 1.188 & 0.010 & 0.501 & 0.009 & 0.394 & 0.010 & 0.1849 & & + 30b & 11:01:01.51 & 01:05:59.90 & 18.251 & 0.013 & 0.079 & 1.582 & 0.174 & 1.200 & 0.019 & 0.496 & 0.015 & 0.418 & 0.025 & & & + 30c & 11:01:03.15 & 01:06:01.17 & 19.026 & 0.019 & 0.078 & 1.250 & 0.131 & 0.618 & 0.024 & 0.361 & 0.026 & 0.233 & 0.057 & & & + 30d & 11:01:01.31 & 01:05:43.80 & 19.220 & 0.018 & 0.079 & 1.903 & 0.407 & 1.300 & 0.031 & 0.502 & 0.023 & 0.235 & 0.048 & & & + & & & & & & & & & & & & & & & & + 31a & 11:02:45.15 & 00:39:38.66 & 16.233 & 0.007 & 0.092 & 2.085 & 0.040 & 0.991 & 0.008 & 0.439 & 0.008 & 0.364 & 0.006 & 0.0964 & & + 31b & 11:02:45.11 & 00:39:22.73 & 17.213 & 0.011 & 0.092 & 1.501 & 0.057 & 0.763 & 0.013 & 0.414 & 0.013 & 0.192 & 0.020 & & 0.1233 & 2dfgrsn367z014 + 31c & 11:02:44.32 & 00:39:23.86 & 17.808 & 0.012 & 0.092 & 1.185 & 0.036 & 0.494 & 0.013 & 0.331 & 0.014 & 0.152 & 0.025 & & & + 31d & 11:02:43.67 & 00:39:18.05 & 18.360 & 0.012 & 0.092 & 1.740 & 0.098 & 0.967 & 0.015 & 0.377 & 0.014 & 0.313 & 0.021 & & & + & & & & & & & & & & & & & & & & + 32a & 11:03:55.55 & 00:37:54.97 & 16.649 & 0.015 & 0.106 & 1.074 & 0.042 & 0.537 & 0.016 & 0.239 & 0.018 & 0.167 & 0.032 & 0.0964 & & + 32b & 11:03:57.17 & 00:38:05.11 & 17.786 & 0.011 & 0.106 & 1.283 & 0.038 & 0.606 & 0.012 & 0.332 & 0.012 & 0.206 & 0.020 & & & + 32c & 11:03:54.90 & 00:37:59.71 & 18.703 & 0.018 & 0.106 & 1.366 & 0.131 & 0.892 & 0.025 & 0.427 & 0.024 & 0.273 & 0.043 & & & + 32d & 11:03:56.01 & 00:38:23.51 & 18.732 & 0.021 & 0.107 & 1.204 & 0.094 & 0.595 & 0.025 & 0.368 & 0.027 & 0.139 & 0.054 & & & + & & & & & & & & & & & & & & & & + 33a & 11:04:04.74 & 01:04:37.20 & 17.375 & 0.008 & 0.145 & 1.125 & 0.030 & 0.859 & 0.009 & 0.591 & 0.008 & 0.366 & 0.009 & 0.1204 & & + 33b & 11:04:04.76 & 01:04:49.26 & 18.115 & 0.015 & 0.146 & 1.852 & 0.211 & 1.000 & 0.021 & 0.567 & 0.018 & 0.494 & 0.027 & & & + 33c & 11:04:05.69 & 01:04:39.72 & 19.952 & 0.026 & 0.146 & 1.704 & 0.391 & 1.028 & 0.041 & 0.418 & 0.036 & 0.317 & 0.075 & & & + 33d & 11:04:05.91 & 01:04:44.92 & 20.087 & 0.041 & 0.147 & 0.857 & 0.272 & 0.887 & 0.063 & 0.408 & 0.058 & 0.073 & 0.156 & & & + & & & & & & & & & & & & & & & & + 34a & 11:09:40.39 & 00:45:24.08 & 17.950 & 0.015 & 0.114 & 4.919 & 2.853 & 1.712 & 0.030 & 0.671 & 0.017 & 0.366 & 0.023 & & & + 34b & 11:09:40.00 & 00:45:36.55 & 18.500 & 0.014 & 0.114 & 3.386 & 1.044 & 1.761 & 0.028 & 0.648 & 0.016 & 0.345 & 0.021 & 0.3319 & & + 34c & 11:09:40.74 & 00:45:37.70 & 19.717 & 0.032 & 0.114 & 1.122 & 0.418 & 1.584 & 0.074 & 0.660 & 0.041 & 0.273 & 0.072 & & & + 34d & 11:09:39.85 & 00:45:34.66 & 19.758 & 0.034 & 0.114 & 2.536 & 1.305 & 1.609 & 0.079 & 0.580 & 0.044 & 0.451 & 0.072 & & & + 34e & 11:09:39.21 & 00:45:42.02 & 20.543 & 0.044 & 0.114 & 2.236 & 1.011 & 1.526 & 0.098 & 0.672 & 0.055 & 0.443 & 0.086 & & & + 34f & 11:09:40.47 & 00:45:31.17 & 20.920 & 0.049 & 0.114 & 1.635 & 0.779 & 1.526 & 0.113 & 0.565 & 0.066 & 0.564 & 0.101 & & & + & & & & & & & & & & & & & & & & + 35a & 11:11:10.97 & 01:06:17.22 & 16.244 & 0.007 & 0.097 & 1.907 & 0.033 & 0.932 & 0.007 & 0.422 & 0.007 & 0.387 & 0.005 & 0.0909 & & + 35b & 11:11:11.75 & 01:06:24.75 & 17.616 & 0.010 & 0.097 & 1.826 & 0.103 & 0.857 & 0.013 & 0.372 & 0.012 & 0.320 & 0.020 & & & + 35c & 11:11:13.26 & 01:06:35.08 & 18.380 & 0.010 & 0.098 & 1.546 & 0.085 & 0.690 & 0.013 & 0.369 & 0.013 & 0.295 & 0.025 & & & + 35d & 11:11:11.97 & 01:06:35.42 & 18.928 & 0.023 & 0.097 & 1.525 & 0.253 & 0.942 & 0.033 & 0.520 & 0.028 & 0.326 & 0.051 & & & + & & & & & & & & & & & & & & & & + 36a & 11:13:36.54 & -00:28:49.86 & 16.659 & 0.010 & 0.102 & 1.857 & 0.079 & 1.027 & 0.012 & 0.485 & 0.011 & 0.415 & 0.011 & 0.1000 & 0.1000 & nedgroup + 36b & 11:13:36.53 & -00:29:00.92 & 18.213 & 0.010 & 0.102 & 2.477 & 0.198 & 1.327 & 0.013 & 0.677 & 0.011 & 0.384 & 0.009 & & & + 36c & 11:13:35.85 & -00:28:44.11 & 18.857 & 0.021 & 0.102 & 0.976 & 0.073 & 0.363 & 0.024 & 0.264 & 0.028 & 0.071 & 0.068 & & & + 36d & 11:13:36.45 & -00:28:39.81 & 19.081 & 0.019 & 0.102 & 3.033 & 1.157 & 1.481 & 0.037 & 0.568 & 0.023 & 0.379 & 0.036 & & & + 36e & 11:13:37.07 & -00:29:02.51 & 19.565 & 0.021 & 0.102 & 1.494 & 0.314 & 1.306 & 0.037 & 0.479 & 0.025 & 0.299 & 0.045 & & & + & & & & & & & & & & & & & & & & + 37a & 11:14:59.21 & -00:35:53.15 & 16.493 & 0.010 & 0.153 & 1.375 & 0.028 & 0.662 & 0.011 & 0.447 & 0.011 & 0.305 & 0.010 & & & + 37b & 11:14:58.80 & -00:36:27.48 & 18.017 & 0.012 & 0.153 & 2.483 & 0.402 & 1.545 & 0.021 & 0.617 & 0.014 & 0.416 & 0.017 & & & + 37c & 11:14:59.64 & -00:36:13.86 & 19.093 & 0.015 & 0.154 & 0.700 & 0.045 & 0.421 & 0.018 & 0.307 & 0.020 & 0.145 & 0.045 & & & + 37d & 11:14:58.12 & -00:35:50.51 & 19.309 & 0.019 & 0.151 & 1.694 & 0.354 & 1.409 & 0.036 & 0.633 & 0.022 & 0.450 & 0.034 & & & + & & & & & & & & & & & & & & & & + 38a & 11:15:37.85 & -00:01:54.60 & 17.548 & 0.009 & 0.160 & 0.875 & 0.014 & 0.403 & 0.009 & 0.322 & 0.010 & 0.131 & 0.014 & & & + 38b & 11:15:36.76 & -00:01:45.34 & 17.610 & 0.009 & 0.161 & 1.963 & 0.054 & 0.929 & 0.010 & 0.412 & 0.009 & 0.328 & 0.009 & & & + 38c & 11:15:37.73 & -00:01:46.08 & 18.396 & 0.016 & 0.161 & 1.118 & 0.071 & 0.619 & 0.020 & 0.424 & 0.019 & 0.162 & 0.041 & & & + 38d & 11:15:37.39 & -00:01:55.52 & 19.347 & 0.020 & 0.160 & 2.067 & 0.308 & 0.894 & 0.029 & 0.375 & 0.025 & 0.348 & 0.050 & & & + & & & & & & & & & & & & & & & & + 39a & 11:21:09.61 & 00:12:42.81 & 15.972 & 0.008 & 0.095 & 2.008 & 0.038 & 1.030 & 0.008 & 0.444 & 0.008 & 0.388 & 0.006 & & & + 39b & 11:21:07.97 & 00:12:42.51 & 17.760 & 0.011 & 0.096 & 1.954 & 0.106 & 0.948 & 0.013 & 0.411 & 0.012 & 0.336 & 0.020 & & & + 39c & 11:21:06.74 & 00:12:35.80 & 18.528 & 0.014 & 0.096 & 1.631 & 0.128 & 0.970 & 0.018 & 0.409 & 0.017 & 0.387 & 0.031 & & & + 39d & 11:21:09.14 & 00:12:36.86 & 18.751 & 0.017 & 0.095 & 1.273 & 0.157 & 1.248 & 0.026 & 0.374 & 0.021 & 0.551 & 0.036 & & & + & & & & & & & & & & & & & & & & + 40a & 11:22:02.20 & 00:24:55.44 & 17.792 & 0.013 & 0.077 & 1.019 & 0.044 & 0.724 & 0.015 & 0.477 & 0.014 & 0.261 & 0.026 & & & + 40b & 11:22:03.08 & 00:25:05.10 & 19.562 & 0.024 & 0.077 & 1.433 & 0.271 & 1.192 & 0.038 & 0.602 & 0.028 & 0.321 & 0.057 & & & + 40c & 11:22:03.04 & 00:24:57.01 & 19.995 & 0.032 & 0.077 & 0.744 & 0.137 & 0.694 & 0.042 & 0.184 & 0.046 & 0.329 & 0.124 & & & + 40d & 11:22:02.39 & 00:24:51.31 & 20.671 & 0.048 & 0.077 & 0.574 & 0.249 & 1.031 & 0.077 & 0.555 & 0.061 & 0.718 & 0.098 & & & + & & & & & & & & & & & & & & & & + 41a & 11:29:30.10 & -00:33:13.88 & 17.751 & 0.011 & 0.080 & 1.418 & 0.065 & 0.648 & 0.013 & 0.405 & 0.013 & 0.304 & 0.022 & 0.0754 & & + 41b & 11:29:30.55 & -00:33:02.99 & 20.051 & 0.037 & 0.079 & 2.462 & 1.463 & 1.725 & 0.100 & 0.576 & 0.047 & 0.406 & 0.080 & & & + 41c & 11:29:31.12 & -00:33:07.56 & 20.060 & 0.041 & 0.079 & 1.178 & 0.755 & 1.747 & 0.117 & 0.620 & 0.052 & 0.440 & 0.087 & & & + 41d & 11:29:31.41 & -00:33:10.16 & 20.229 & 0.041 & 0.079 & 1.752 & 0.797 & 1.327 & 0.083 & 0.598 & 0.052 & 0.492 & 0.083 & & & + & & & & & & & & & & & & & & & & + 42a & 11:31:39.63 & 00:12:54.16 & 16.180 & 0.011 & 0.069 & 1.613 & 0.045 & 0.927 & 0.012 & 0.403 & 0.011 & 0.357 & 0.012 & 0.1324 & & + 42b & 11:31:38.25 & 00:12:53.33 & 16.918 & 0.010 & 0.069 & 1.307 & 0.037 & 0.972 & 0.012 & 0.529 & 0.011 & 0.344 & 0.011 & & & + 42c & 11:31:38.80 & 00:13:08.41 & 16.949 & 0.011 & 0.069 & 2.077 & 0.112 & 1.115 & 0.013 & 0.582 & 0.012 & 0.475 & 0.013 & 0.1324 & & + 42d & 11:31:38.13 & 00:13:01.77 & 18.552 & 0.016 & 0.069 & 2.315 & 0.484 & 1.565 & 0.030 & 0.555 & 0.018 & 0.429 & 0.030 & & & + 42e & 11:31:37.28 & 00:12:48.98 & 18.580 & 0.017 & 0.069 & 1.411 & 0.140 & 1.004 & 0.023 & 0.455 & 0.020 & 0.310 & 0.039 & & & + 42f & 11:31:37.08 & 00:12:55.77 & 18.592 & 0.017 & 0.069 & 1.584 & 0.202 & 1.261 & 0.026 & 0.598 & 0.019 & 0.471 & 0.030 & & & + & & & & & & & & & & & & & & & & + 43a & 11:33:11.86 & -00:27:52.84 & 15.918 & 0.010 & 0.067 & 1.316 & 0.041 & 0.862 & 0.011 & 0.431 & 0.011 & 0.351 & 0.012 & 0.1050 & 0.1050 & masx j11331190 - 0027528 + 43b & 11:33:12.48 & -00:27:19.82 & 17.732 & 0.010 & 0.067 & 1.922 & 0.099 & 0.965 & 0.012 & 0.435 & 0.011 & 0.404 & 0.013 & & & + 43c & 11:33:10.93 & -00:26:57.89 & 17.787 & 0.010 & 0.067 & 1.963 & 0.098 & 1.048 & 0.012 & 0.459 & 0.011 & 0.352 & 0.012 & & & + 43d & 11:33:12.46 & -00:27:06.84 & 17.849 & 0.011 & 0.067 & 1.981 & 0.124 & 1.030 & 0.013 & 0.466 & 0.012 & 0.389 & 0.015 & & & + 43e & 11:33:09.85 & -00:27:32.23 & 17.997 & 0.010 & 0.067 & 2.317 & 0.341 & 1.480 & 0.019 & 0.548 & 0.012 & 0.416 & 0.018 & & & + & & & & & & & & & & & & & & & & + 44a & 11:43:41.31 & 00:38:50.88 & 17.966 & 0.018 & 0.076 & 1.476 & 0.235 & 1.389 & 0.031 & 0.439 & 0.022 & 0.416 & 0.034 & & & + 44b & 11:43:40.54 & 00:39:06.94 & 20.081 & 0.035 & 0.076 & 1.052 & 0.221 & 0.828 & 0.049 & 0.281 & 0.050 & 0.329 & 0.095 & & & + 44c & 11:43:40.28 & 00:38:59.99 & 20.315 & 0.037 & 0.076 & 2.985 & 1.138 & 1.323 & 0.069 & 0.507 & 0.048 & 0.440 & 0.075 & & & + 44d & 11:43:40.72 & 00:38:46.17 & 20.477 & 0.036 & 0.076 & 3.366 & 0.854 & 1.397 & 0.072 & 0.522 & 0.048 & 0.266 & 0.085 & & & + & & & & & & & & & & & & & & & & + 45a & 11:46:30.02 & -01:01:21.60 & 17.436 & 0.010 & 0.055 & 1.082 & 0.026 & 0.486 & 0.011 & 0.310 & 0.011 & 0.190 & 0.017 & 0.0799 & & + 45b & 11:46:28.65 & -01:01:15.89 & 18.165 & 0.011 & 0.054 & 1.639 & 0.100 & 0.907 & 0.016 & 0.517 & 0.013 & 0.348 & 0.020 & & & + 45c & 11:46:29.46 & -01:01:24.41 & 19.730 & 0.030 & 0.055 & 1.130 & 0.143 & 0.510 & 0.041 & 0.224 & 0.043 & 0.114 & 0.103 & & & + 45d & 11:46:28.36 & -01:01:29.38 & 20.427 & 0.056 & 0.055 & 0.772 & 0.196 & 0.459 & 0.076 & 0.119 & 0.086 & -0.071 & 0.263 & & & + & & & & & & & & & & & & & & & & + 46a & 11:48:36.47 & 00:46:52.74 & 16.094 & 0.008 & 0.069 & 2.087 & 0.042 & 1.058 & 0.009 & 0.456 & 0.008 & 0.401 & 0.005 & 0.1260 & & + 46b & 11:48:37.01 & 00:47:18.70 & 17.724 & 0.010 & 0.069 & 1.776 & 0.077 & 1.051 & 0.012 & 0.441 & 0.011 & 0.378 & 0.012 & & & + 46c & 11:48:37.19 & 00:46:52.69 & 18.464 & 0.012 & 0.069 & 2.034 & 0.124 & 1.098 & 0.015 & 0.402 & 0.013 & 0.344 & 0.016 & & & + 46d & 11:48:36.28 & 00:47:18.93 & 18.553 & 0.015 & 0.069 & 1.532 & 0.099 & 0.806 & 0.018 & 0.473 & 0.017 & 0.351 & 0.024 & & & + 46e & 11:48:38.21 & 00:46:56.29 & 18.787 & 0.020 & 0.069 & 1.734 & 0.243 & 0.960 & 0.030 & 0.406 & 0.027 & 0.340 & 0.046 & & & + & & & & & & & & & & & & & & & & + 47a & 11:51:11.03 & 00:17:01.23 & 17.901 & 0.024 & 0.066 & 1.217 & 0.127 & 0.734 & 0.030 & 0.300 & 0.030 & -0.001 & 0.086 & & & + 47b & 11:51:10.74 & 00:17:04.47 & 18.107 & 0.012 & 0.066 & 2.144 & 0.144 & 1.106 & 0.015 & 0.518 & 0.012 & 0.371 & 0.017 & & & + 47c & 11:51:10.81 & 00:16:52.46 & 20.096 & 0.039 & 0.066 & 1.039 & 0.247 & 0.821 & 0.057 & 0.307 & 0.053 & 0.432 & 0.121 & & & + 47d & 11:51:11.64 & 00:17:04.89 & 20.323 & 0.039 & 0.066 & 0.830 & 0.138 & 0.367 & 0.048 & 0.289 & 0.053 & 0.213 & 0.146 & & & + & & & & & & & & & & & & & & & & + 48a & 11:55:03.58 & -00:34:50.64 & 16.283 & 0.011 & 0.073 & 1.495 & 0.056 & 0.855 & 0.012 & 0.419 & 0.011 & 0.354 & 0.013 & & & + 48b & 11:55:04.81 & -00:35:22.95 & 16.344 & 0.009 & 0.072 & 1.536 & 0.046 & 0.872 & 0.010 & 0.406 & 0.010 & 0.335 & 0.010 & & & + 48c & 11:55:05.75 & -00:34:56.23 & 16.624 & 0.013 & 0.073 & 1.170 & 0.053 & 0.681 & 0.015 & 0.282 & 0.015 & 0.217 & 0.024 & 0.1353 & & + 48d & 11:55:05.05 & -00:35:07.61 & 17.460 & 0.011 & 0.073 & 1.142 & 0.043 & 0.708 & 0.013 & 0.401 & 0.013 & 0.343 & 0.016 & 0.1274 & & + 48e & 11:55:06.34 & -00:34:56.04 & 18.135 & 0.016 & 0.073 & 1.376 & 0.129 & 0.841 & 0.022 & 0.439 & 0.020 & 0.265 & 0.034 & & & + & & & & & & & & & & & & & & & & + 49a & 11:55:40.76 & -00:55:30.74 & 17.006 & 0.008 & 0.063 & 1.866 & 0.065 & 1.110 & 0.010 & 0.381 & 0.009 & 0.107 & 0.011 & 0.1067 & 0.1067 & 2masx j11554077 - 0055309 + 49b & 11:55:41.23 & -00:55:29.37 & 17.803 & 0.010 & 0.063 & 0.990 & 0.051 & 1.190 & 0.015 & 0.419 & 0.011 & 0.217 & 0.015 & & & + 49c & 11:55:39.76 & -00:55:24.74 & 17.954 & 0.011 & 0.063 & 1.434 & 0.066 & 0.762 & 0.015 & 0.388 & 0.013 & 0.265 & 0.019 & & & + 49d & 11:55:40.36 & -00:55:32.40 & 19.079 & 0.018 & 0.063 & 0.323 & 0.139 & 1.793 & 0.055 & 0.250 & 0.025 & -0.122 & 0.066 & & & + & & & & & & & & & & & & & & & & + 50a & 11:55:41.09 & -00:53:02.31 & 17.358 & 0.009 & 0.064 & 1.850 & 0.059 & 0.851 & 0.010 & 0.350 & 0.010 & 0.353 & 0.010 & 0.0761 & & + 50b & 11:55:41.80 & -00:53:11.95 & 17.759 & 0.009 & 0.064 & 2.085 & 0.090 & 0.959 & 0.011 & 0.449 & 0.010 & 0.378 & 0.010 & & & + 50c & 11:55:41.29 & -00:53:08.74 & 18.840 & 0.016 & 0.064 & 1.410 & 0.272 & 1.729 & 0.042 & 0.528 & 0.019 & 0.414 & 0.026 & 0.3390 & & + 50d & 11:55:41.87 & -00:53:15.12 & 19.246 & 0.025 & 0.064 & 1.468 & 0.184 & 0.613 & 0.036 & 0.169 & 0.037 & 0.001 & 0.094 & & & + & & & & & & & & & & & & & & & & + 51a & 11:56:37.38 & 01:07:00.84 & 16.280 & 0.011 & 0.044 & 1.040 & 0.038 & 0.430 & 0.013 & 0.178 & 0.014 & 0.158 & 0.028 & & & + 51b & 11:56:38.62 & 01:07:16.70 & 17.241 & 0.008 & 0.044 & 0.824 & 0.019 & 0.373 & 0.009 & 0.290 & 0.010 & 0.157 & 0.016 & 0.1573 & & + 51c & 11:56:35.52 & 01:07:20.09 & 17.719 & 0.008 & 0.044 & 1.996 & 0.085 & 0.966 & 0.010 & 0.381 & 0.009 & 0.395 & 0.011 & & & + 51d & 11:56:37.43 & 01:06:52.77 & 19.214 & 0.022 & 0.044 & 1.310 & 0.168 & 0.679 & 0.030 & 0.311 & 0.030 & 0.356 & 0.059 & & & + & & & & & & & & & & & & & & & & + 52a & 11:57:43.41 & -00:52:16.60 & 16.128 & 0.010 & 0.069 & 1.131 & 0.024 & 0.676 & 0.011 & 0.179 & 0.011 & 0.129 & 0.015 & & 0.1318&2masx j11574348 - 0052178 + 52b & 11:57:43.05 & -00:52:31.39 & 18.314 & 0.013 & 0.069 & 1.981 & 0.176 & 0.976 & 0.020 & 0.404 & 0.016 & 0.308 & 0.025 & & & + 52c & 11:57:44.16 & -00:52:43.44 & 18.601 & 0.013 & 0.069 & 1.115 & 0.057 & 0.680 & 0.017 & 0.362 & 0.015 & 0.252 & 0.026 & & & + 52d & 11:57:44.95 & -00:52:09.32 & 19.072 & 0.021 & 0.069 & 1.013 & 0.095 & 0.533 & 0.029 & 0.371 & 0.028 & 0.158 & 0.057 & & & + & & & & & & & & & & & & & & & & + 53a & 11:58:20.85 & -00:05:05.07 & 16.766 & 0.009 & 0.092 & 2.118 & 0.066 & 1.072 & 0.010 & 0.431 & 0.009 & 0.420 & 0.007 & 0.1058 & 0.1058 & 2masx j11582083 - 0005047 + 53b & 11:58:21.42 & -00:05:04.53 & 17.625 & 0.010 & 0.092 & 1.702 & 0.068 & 1.042 & 0.012 & 0.440 & 0.011 & 0.431 & 0.011 & & & + 53c & 11:58:19.18 & -00:05:13.86 & 18.645 & 0.018 & 0.093 & 1.663 & 0.306 & 1.486 & 0.035 & 0.499 & 0.021 & 0.421 & 0.033 & & & + 53d & 11:58:21.52 & -00:05:20.78 & 18.773 & 0.012 & 0.093 & 1.032 & 0.045 & 0.499 & 0.014 & 0.353 & 0.014 & 0.129 & 0.029 & & & + & & & & & & & & & & & & & & & & + 54a & 11:58:27.70 & 00:43:04.45 & 14.843 & 0.007 & 0.062 & 2.093 & 0.016 & 1.014 & 0.007 & 0.486 & 0.007 & 0.411 & 0.002 & & 0.0474 & ugc06959ned03 + 54b & 11:58:26.12 & 00:42:47.14 & 16.216 & 0.007 & 0.061 & 2.046 & 0.023 & 0.885 & 0.008 & 0.407 & 0.008 & 0.309 & 0.003 & 0.0784 & 0.0786 & ugc06959ned02 + 54c & 11:58:25.30 & 00:42:24.61 & 16.759 & 0.008 & 0.061 & 2.055 & 0.026 & 0.864 & 0.008 & 0.430 & 0.008 & 0.394 & 0.003 & & 0.0475 & ugc06959ned01 + 54d & 11:58:24.80 & 00:42:17.79 & 16.796 & 0.026 & 0.061 & 2.006 & 0.250 & 0.688 & 0.033 & 0.596 & 0.030 & 0.584 & 0.033 & & & + & & & & & & & & & & & & & & & & + 55a & 12:00:08.06 & 00:40:25.81 & 15.931 & 0.007 & 0.066 & 2.151 & 0.031 & 0.967 & 0.008 & 0.439 & 0.008 & 0.339 & 0.004 & 0.0854 & & + 55b & 12:00:10.35 & 00:40:21.92 & 18.276 & 0.011 & 0.066 & 1.775 & 0.100 & 1.090 & 0.014 & 0.437 & 0.013 & 0.319 & 0.015 & & & + 55c & 12:00:08.34 & 00:40:36.00 & 18.290 & 0.010 & 0.066 & 2.441 & 0.130 & 1.244 & 0.012 & 0.400 & 0.011 & 0.372 & 0.009 & & & + 55d & 12:00:08.38 & 00:40:37.71 & 18.796 & 0.013 & 0.066 & 1.567 & 0.097 & 1.137 & 0.018 & 0.426 & 0.015 & 0.203 & 0.020 & & & + & & & & & & & & & & & & & & & & + 56a & 12:00:09.90 & -00:35:49.93 & 16.416 & 0.008 & 0.075 & 1.924 & 0.066 & 1.204 & 0.010 & 0.491 & 0.009 & 0.399 & 0.006 & 0.1694 & 0.1694 & 2masx j12000991 - 0035499 + 56b & 12:00:10.57 & -00:35:36.77 & 17.222 & 0.009 & 0.075 & 1.798 & 0.079 & 1.197 & 0.011 & 0.529 & 0.010 & 0.436 & 0.008 & 0.1695 & 0.1695&2masx j12001057 - 0035369 + 56c & 12:00:08.79 & -00:36:03.46 & 19.128 & 0.017 & 0.075 & 2.275 & 0.412 & 1.177 & 0.027 & 0.447 & 0.020 & 0.368 & 0.031 & & 0.1485 & + 56d & 12:00:11.35 & -00:35:58.46 & 19.396 & 0.021 & 0.074 & 2.792 & 0.823 & 1.103 & 0.036 & 0.439 & 0.027 & 0.416 & 0.044 & & & + & & & & & & & & & & & & & & & & + 57a & 12:00:13.54 & -00:25:18.55 & 16.582 & 0.008 & 0.080 & 1.940 & 0.041 & 0.946 & 0.008 & 0.422 & 0.008 & 0.375 & 0.005 & 0.0766 & 0.0766 & 2masx j12001350 - 0025189 + 57b & 12:00:13.95 & -00:25:27.06 & 17.457 & 0.010 & 0.080 & 1.035 & 0.024 & 0.471 & 0.010 & 0.334 & 0.010 & 0.161 & 0.014 & & & + 57c & 12:00:14.01 & -00:25:29.36 & 18.470 & 0.011 & 0.080 & 2.382 & 0.218 & 1.192 & 0.016 & 0.557 & 0.012 & 0.365 & 0.012 & & & + 57d & 12:00:12.44 & -00:25:49.22 & 19.341 & 0.018 & 0.080 & 2.343 & 0.576 & 1.320 & 0.033 & 0.436 & 0.022 & 0.339 & 0.035 & & & + 57e & 12:00:12.40 & -00:25:36.41 & 19.576 & 0.023 & 0.080 & 2.302 & 0.776 & 1.361 & 0.045 & 0.480 & 0.028 & 0.453 & 0.043 & & & + & & & & & & & & & & & & & & & & + 58a & 12:00:56.89 & -00:07:58.92 & 17.095 & 0.012 & 0.059 & 1.521 & 0.098 & 1.077 & 0.017 & 0.451 & 0.014 & 0.391 & 0.019 & & & + 58b & 12:00:57.65 & -00:07:53.74 & 18.643 & 0.017 & 0.059 & 1.981 & 0.273 & 1.123 & 0.026 & 0.480 & 0.020 & 0.280 & 0.035 & & & + 58c & 12:00:57.56 & -00:08:00.54 & 19.616 & 0.023 & 0.059 & 2.196 & 0.557 & 1.282 & 0.042 & 0.584 & 0.027 & 0.386 & 0.043 & & & + 58d & 12:00:56.31 & -00:08:01.72 & 19.755 & 0.028 & 0.059 & 0.444 & 0.080 & 0.531 & 0.036 & 0.020 & 0.042 & 0.256 & 0.105 & & & + & & & & & & & & & & & & & & & & + 59a & 12:02:23.46 & 01:14:43.24 & 17.868 & 0.012 & 0.057 & 1.780 & 0.237 & 1.410 & 0.022 & 0.557 & 0.014 & 0.326 & 0.022 & & & + 59b & 12:02:23.06 & 01:14:42.55 & 18.246 & 0.016 & 0.057 & 2.306 & 0.557 & 1.432 & 0.032 & 0.465 & 0.020 & 0.554 & 0.030 & & & + 59c & 12:02:22.48 & 01:14:56.31 & 18.709 & 0.018 & 0.057 & 1.146 & 0.165 & 1.076 & 0.029 & 0.417 & 0.023 & 0.435 & 0.039 & & & + 59d & 12:02:22.04 & 01:14:52.91 & 20.293 & 0.041 & 0.057 & 1.962 & 0.928 & 1.200 & 0.079 & 0.551 & 0.053 & 0.579 & 0.084 & & & + 59e & 12:02:23.15 & 01:14:35.28 & 20.574 & 0.058 & 0.057 & 0.807 & 0.353 & 0.743 & 0.088 & 0.068 & 0.098 & 0.812 & 0.164 & & & + & & & & & & & & & & & & & & & & + 60a & 12:09:28.31 & -00:00:32.08 & 17.523 & 0.011 & 0.062 & 1.713 & 0.110 & 1.306 & 0.015 & 0.481 & 0.012 & 0.357 & 0.014 & & & + 60b & 12:09:27.70 & -00:00:30.05 & 19.720 & 0.032 & 0.062 & 1.316 & 0.348 & 1.148 & 0.057 & 0.407 & 0.042 & 0.429 & 0.075 & & & + 60c & 12:09:27.87 & -00:00:26.56 & 19.913 & 0.027 & 0.062 & 2.752 & 1.041 & 1.639 & 0.063 & 0.712 & 0.031 & 0.415 & 0.044 & & & + 60d & 12:09:28.77 & -00:00:25.48 & 20.211 & 0.030 & 0.062 & 2.534 & 0.937 & 1.751 & 0.077 & 0.580 & 0.035 & 0.356 & 0.058 & & & + & & & & & & & & & & & & & & & & + 61a & 12:09:52.42 & -00:33:58.28 & 16.817 & 0.009 & 0.056 & 2.056 & 0.081 & 1.299 & 0.010 & 0.516 & 0.009 & 0.391 & 0.006 & & & + 61b & 12:09:50.88 & -00:34:00.97 & 17.421 & 0.010 & 0.056 & 1.938 & 0.120 & 1.232 & 0.013 & 0.504 & 0.011 & 0.360 & 0.011 & 0.1869 & & + 61c & 12:09:51.27 & -00:34:14.78 & 18.858 & 0.018 & 0.056 & 1.591 & 0.248 & 1.190 & 0.030 & 0.447 & 0.022 & 0.340 & 0.035 & & & + 61d & 12:09:50.96 & -00:33:54.76 & 19.286 & 0.023 & 0.056 & 1.281 & 0.190 & 0.903 & 0.034 & 0.429 & 0.029 & 0.355 & 0.047 & & & + 61e & 12:09:51.66 & -00:34:03.86 & 19.571 & 0.138 & 0.056 & 0.531 & 1.310 & 1.736 & 0.367 & 0.043 & 0.212 & -0.018 & 0.606 & & & + & & & & & & & & & & & & & & & & + 62a & 12:14:15.32 & -00:26:57.87 & 17.367 & 0.011 & 0.056 & 2.274 & 0.269 & 1.502 & 0.018 & 0.565 & 0.012 & 0.284 & 0.015 & 0.2452 & 0.2452 & 2masx j12141532 - 0026584 + 62b & 12:14:15.48 & -00:26:56.69 & 19.271 & 0.017 & 0.056 & 2.224 & 0.450 & 1.541 & 0.032 & 0.577 & 0.019 & 0.329 & 0.024 & & 0.2452 & 2masx j12141532 - 0026584 + 62c & 12:14:15.94 & -00:26:57.50 & 19.352 & 0.023 & 0.056 & 2.224 & 0.676 & 1.381 & 0.044 & 0.378 & 0.029 & 0.640 & 0.039 & & & + 62d & 12:14:15.31 & -00:27:01.17 & 19.619 & 0.027 & 0.056 & 1.325 & 0.376 & 1.374 & 0.056 & 0.356 & 0.035 & 0.575 & 0.053 & & & + & & & & & & & & & & & & & & & & + 63a & 12:15:31.40 & 00:38:32.89 & 17.883 & 0.011 & 0.063 & 1.869 & 0.085 & 0.887 & 0.013 & 0.399 & 0.012 & 0.062 & 0.018 & & & + 63b & 12:15:31.16 & 00:38:30.78 & 18.730 & 0.017 & 0.063 & 1.008 & 0.062 & 0.528 & 0.021 & 0.258 & 0.022 & 0.077 & 0.045 & & & + 63c & 12:15:30.47 & 00:38:40.20 & 18.784 & 0.033 & 0.063 & 0.808 & 0.025 & 0.202 & 0.033 & 0.352 & 0.034 & -0.075 & 0.030 & 0.0752 & & + 63d & 12:15:31.80 & 00:38:24.72 & 20.716 & 0.053 & 0.063 & 1.080 & 0.355 & 0.816 & 0.081 & 0.124 & 0.085 & 0.408 & 0.162 & & & + & & & & & & & & & & & & & & & & + 64a & 12:17:04.76 & -00:55:07.83 & 17.330 & 0.009 & 0.066 & 2.343 & 0.154 & 1.315 & 0.014 & 0.516 & 0.010 & 0.335 & 0.010 & 0.1977 & & + 64b & 12:17:05.20 & -00:54:52.42 & 19.032 & 0.016 & 0.067 & 1.900 & 0.265 & 1.266 & 0.030 & 0.529 & 0.018 & 0.329 & 0.026 & & & + 64c & 12:17:04.89 & -00:54:39.69 & 19.424 & 0.020 & 0.067 & 2.404 & 0.556 & 1.246 & 0.040 & 0.538 & 0.024 & 0.410 & 0.034 & & & + 64d & 12:17:04.39 & -00:54:47.73 & 20.030 & 0.031 & 0.066 & 0.767 & 0.114 & 0.472 & 0.043 & 0.306 & 0.042 & 0.261 & 0.084 & & & + & & & & & & & & & & & & & & & & + 65a & 12:22:39.15 & -00:58:48.41 & 16.401 & 0.010 & 0.065 & 1.554 & 0.073 & 1.172 & 0.014 & 0.362 & 0.011 & -0.265 & 0.022 & & & + 65b & 12:22:39.33 & -00:58:51.13 & 16.730 & 0.009 & 0.065 & 2.238 & 0.148 & 1.300 & 0.014 & 0.574 & 0.010 & 0.495 & 0.009 & & 0.1753 & 2masx j12223930 - 0058503 + 65c & 12:22:41.23 & -00:58:58.78 & 19.280 & 0.020 & 0.064 & 1.886 & 0.325 & 1.171 & 0.038 & 0.485 & 0.024 & 0.334 & 0.039 & & & + 65d & 12:22:38.20 & -00:58:41.36 & 19.318 & 0.020 & 0.065 & 1.556 & 0.276 & 1.246 & 0.042 & 0.497 & 0.024 & 0.261 & 0.043 & & & + & & & & & & & & & & & & & & & & + 66a & 12:25:13.71 & -00:10:49.95 & 17.503 & 0.011 & 0.075 & 2.795 & 0.396 & 1.578 & 0.018 & 0.606 & 0.012 & 0.455 & 0.012 & 0.2873 & & + 66b & 12:25:13.31 & -00:10:52.61 & 18.648 & 0.021 & 0.075 & 1.630 & 0.406 & 1.605 & 0.047 & 0.528 & 0.025 & 0.370 & 0.041 & & & + 66c & 12:25:14.70 & -00:10:54.54 & 19.835 & 0.025 & 0.075 & 1.090 & 0.357 & 1.657 & 0.063 & 0.590 & 0.030 & 0.381 & 0.049 & & & + 66d & 12:25:14.34 & -00:10:41.29 & 20.315 & 0.182 & 0.075 & -0.109 & 0.609 & 1.142 & 0.329 & 0.086 & 0.277 & 0.413 & 0.610 & & & + & & & & & & & & & & & & & & & & + 67a & 12:31:39.53 & -00:34:14.04 & 16.411 & 0.010 & 0.057 & 2.068 & 0.114 & 1.257 & 0.012 & 0.460 & 0.010 & 0.385 & 0.010 & 0.2021 & & + 67b & 12:31:39.46 & -00:33:54.39 & 18.604 & 0.017 & 0.056 & 1.882 & 0.283 & 1.164 & 0.027 & 0.393 & 0.020 & 0.422 & 0.032 & & & + 67c & 12:31:39.63 & -00:33:52.93 & 19.205 & 0.017 & 0.056 & 1.072 & 0.116 & 1.031 & 0.025 & 0.456 & 0.020 & 0.294 & 0.031 & & & + 67d & 12:31:40.22 & -00:34:18.09 & 19.381 & 0.023 & 0.057 & 1.319 & 0.223 & 0.997 & 0.036 & 0.546 & 0.028 & 0.192 & 0.051 & & & + & & & & & & & & & & & & & & & & + 68a & 12:33:48.93 & -00:09:37.11 & 16.230 & 0.009 & 0.066 & 2.174 & 0.060 & 1.043 & 0.010 & 0.484 & 0.009 & 0.438 & 0.006 & & & + 68b & 12:33:48.22 & -00:09:29.09 & 17.654 & 0.010 & 0.066 & 1.916 & 0.066 & 1.017 & 0.011 & 0.436 & 0.010 & 0.468 & 0.009 & 0.1354 & & + 68c & 12:33:49.98 & -00:09:24.62 & 18.365 & 0.012 & 0.066 & 2.236 & 0.161 & 1.045 & 0.016 & 0.438 & 0.013 & 0.428 & 0.016 & & & + 68d & 12:33:49.88 & -00:09:17.50 & 18.894 & 0.018 & 0.066 & 1.724 & 0.202 & 0.944 & 0.027 & 0.392 & 0.022 & 0.371 & 0.040 & & & + 68e & 12:33:49.29 & -00:09:49.31 & 19.173 & 0.015 & 0.066 & 1.826 & 0.170 & 1.050 & 0.022 & 0.419 & 0.017 & 0.554 & 0.023 & & & + & & & & & & & & & & & & & & & & + 69a & 12:36:44.20 & -01:00:27.50 & 17.978 & 0.016 & 0.066 & 1.229 & 0.134 & 1.052 & 0.028 & 0.444 & 0.020 & 0.405 & 0.031 & & & + 69b & 12:36:44.17 & -01:00:21.03 & 18.734 & 0.014 & 0.066 & 2.008 & 0.263 & 1.248 & 0.025 & 0.502 & 0.017 & 0.367 & 0.024 & & & + 69c & 12:36:44.80 & -01:00:26.22 & 18.809 & 0.018 & 0.066 & 0.969 & 0.089 & 0.724 & 0.026 & 0.400 & 0.023 & 0.217 & 0.043 & & & + 69d & 12:36:43.43 & -01:00:19.81 & 20.160 & 0.074 & 0.066 & 1.083 & 0.664 & 0.944 & 0.147 & -0.066 & 0.138 & 0.478 & 0.290 & & & + & & & & & & & & & & & & & & & & + 70a & 12:36:51.87 & -00:35:15.68 & 16.601 & 0.075 & 0.064 & 0.896 & 0.060 & 0.654 & 0.076 & -0.078 & 0.077 & -0.298 & 0.075 & 0.0088 & & + 70b & 12:36:51.00 & -00:34:55.19 & 18.733 & 0.015 & 0.064 & 2.108 & 0.374 & 1.410 & 0.025 & 0.477 & 0.017 & 0.378 & 0.025 & & & + 70c & 12:36:50.68 & -00:34:55.45 & 18.735 & 0.016 & 0.064 & 2.467 & 0.596 & 1.464 & 0.029 & 0.538 & 0.018 & 0.366 & 0.026 & & & + 70d & 12:36:52.03 & -00:35:21.82 & 19.346 & 0.023 & 0.064 & 0.571 & 0.037 & -0.342 & 0.024 & -0.890 & 0.066 & -0.030 & 0.237 & & & + & & & & & & & & & & & & & & & & + 71a & 12:37:50.44 & -00:18:15.48 & 17.167 & 0.008 & 0.068 & 1.780 & 0.080 & 1.051 & 0.011 & 0.428 & 0.009 & 0.423 & 0.012 & 0.1407 & & + 71b & 12:37:50.18 & -00:17:46.80 & 18.347 & 0.013 & 0.068 & 1.554 & 0.116 & 0.859 & 0.018 & 0.428 & 0.016 & 0.318 & 0.028 & & & + 71c & 12:37:50.14 & -00:17:54.70 & 19.337 & 0.021 & 0.068 & 1.448 & 0.220 & 0.953 & 0.033 & 0.396 & 0.027 & 0.376 & 0.053 & & & + 71d & 12:37:49.54 & -00:17:56.18 & 19.674 & 0.028 & 0.068 & 1.905 & 0.462 & 0.953 & 0.045 & 0.414 & 0.037 & 0.388 & 0.072 & & & + & & & & & & & & & & & & & & & & + 72a & 12:43:09.48 & -00:47:01.40 & 16.715 & 0.008 & 0.066 & 2.124 & 0.095 & 1.069 & 0.010 & 0.429 & 0.008 & 0.444 & 0.009 & 0.1430 & & + 72b & 12:43:09.39 & -00:46:39.36 & 18.138 & 0.010 & 0.066 & 2.450 & 0.200 & 1.012 & 0.013 & 0.449 & 0.011 & 0.374 & 0.014 & & & + 72c & 12:43:08.65 & -00:46:49.33 & 18.831 & 0.014 & 0.066 & 2.003 & 0.247 & 1.022 & 0.021 & 0.418 & 0.016 & 0.423 & 0.025 & & & + 72d & 12:43:09.33 & -00:46:50.96 & 19.403 & 0.020 & 0.066 & 1.580 & 0.323 & 1.151 & 0.037 & 0.432 & 0.025 & 0.438 & 0.042 & & & + & & & & & & & & & & & & & & & & + 73a & 12:43:32.33 & -01:07:00.35 & 17.118 & 0.007 & 0.096 & 2.030 & 0.085 & 1.151 & 0.010 & 0.442 & 0.008 & 0.370 & 0.009 & 0.1661 & & + 73b & 12:43:31.59 & -01:07:04.77 & 18.332 & 0.012 & 0.096 & 2.840 & 0.498 & 1.292 & 0.023 & 0.509 & 0.014 & 0.289 & 0.022 & & & + 73c & 12:43:31.05 & -01:06:59.54 & 19.159 & 0.019 & 0.096 & 1.767 & 0.298 & 1.029 & 0.036 & 0.442 & 0.025 & 0.276 & 0.048 & & & + 73d & 12:43:30.88 & -01:06:55.00 & 19.227 & 0.017 & 0.096 & 2.424 & 0.451 & 1.078 & 0.030 & 0.422 & 0.021 & 0.174 & 0.042 & & & + & & & & & & & & & & & & & & & & + 74a & 12:44:46.21 & -01:01:09.40 & 15.984 & 0.008 & 0.091 & 2.063 & 0.070 & 1.079 & 0.010 & 0.412 & 0.008 & 0.340 & 0.008 & 0.1468 & & + 74b & 12:44:45.96 & -01:01:13.30 & 17.134 & 0.009 & 0.091 & 2.301 & 0.139 & 1.233 & 0.012 & 0.464 & 0.009 & 0.366 & 0.010 & & & + 74c & 12:44:45.73 & -01:01:29.57 & 17.745 & 0.010 & 0.091 & 1.978 & 0.107 & 1.006 & 0.013 & 0.422 & 0.011 & 0.386 & 0.013 & & & + 74d & 12:44:47.93 & -01:01:27.74 & 17.878 & 0.012 & 0.091 & 2.044 & 0.194 & 1.128 & 0.019 & 0.460 & 0.014 & 0.394 & 0.019 & & & + 74e & 12:44:44.51 & -01:01:46.80 & 18.266 & 0.011 & 0.091 & 2.032 & 0.154 & 1.068 & 0.016 & 0.459 & 0.012 & 0.461 & 0.015 & & & + 74f & 12:44:45.02 & -01:01:08.11 & 18.507 & 0.012 & 0.091 & 1.837 & 0.146 & 1.057 & 0.018 & 0.476 & 0.014 & 0.495 & 0.016 & & & + 74 g & 12:44:45.10 & -01:01:01.80 & 18.958 & 0.017 & 0.091 & 1.507 & 0.177 & 0.997 & 0.027 & 0.369 & 0.021 & 0.653 & 0.028 & & & + & & & & & & & & & & & & & & & & + 75a & 12:45:57.83 & -00:59:46.94 & 17.839 & 0.011 & 0.070 & 2.117 & 0.164 & 1.167 & 0.016 & 0.467 & 0.012 & 0.362 & 0.015 & & & + 75b & 12:45:58.25 & -00:59:57.16 & 19.039 & 0.015 & 0.070 & 2.015 & 0.267 & 1.121 & 0.026 & 0.401 & 0.019 & 0.377 & 0.030 & & & + 75c & 12:45:58.89 & -00:59:57.74 & 20.132 & 0.150 & 0.070 & -0.167 & 0.435 & 1.007 & 0.274 & -0.103 & 0.257 & -4.664 & 1.149 & & & + 75d & 12:45:57.77 & -00:59:55.63 & 20.301 & 0.046 & 0.070 & 0.208 & 0.137 & 0.666 & 0.071 & 0.215 & 0.069 & 0.221 & 0.151 & & & + & & & & & & & & & & & & & & & & + 76a & 12:50:58.83 & -00:44:58.91 & 17.414 & 0.010 & 0.068 & 2.599 & 0.449 & 1.609 & 0.018 & 0.565 & 0.011 & 0.343 & 0.014 & 0.2566 & & + 76b & 12:50:58.63 & -00:44:58.07 & 19.067 & 0.032 & 0.068 & 1.349 & 0.431 & 1.151 & 0.057 & 0.491 & 0.040 & 0.516 & 0.062 & & & + 76c & 12:50:57.81 & -00:45:01.13 & 19.105 & 0.029 & 0.068 & 2.109 & 0.623 & 1.004 & 0.045 & 0.522 & 0.035 & 0.518 & 0.049 & & & + 76d & 12:50:58.38 & -00:44:53.48 & 19.226 & 0.018 & 0.068 & 4.471 & 1.406 & 1.450 & 0.035 & 0.531 & 0.021 & 0.390 & 0.031 & & & + 76e & 12:50:58.82 & -00:44:44.08 & 19.300 & 0.021 & 0.068 & 2.516 & 0.946 & 1.390 & 0.042 & 0.492 & 0.026 & 0.325 & 0.044 & & & + & & & & & & & & & & & & & & & & + 77a & 12:53:22.35 & -00:25:42.73 & 15.315 & 0.008 & 0.070 & 1.496 & 0.020 & 0.775 & 0.008 & 0.372 & 0.008 & 0.345 & 0.005 & & 0.0471&cgcg015 - 030ned + 77b & 12:53:23.51 & -00:25:22.56 & 15.750 & 0.024 & 0.070 & 1.388 & 0.054 & 0.787 & 0.025 & -0.099 & 0.025 & 0.083 & 0.028 & & & + 77c & 12:53:23.85 & -00:25:23.49 & 16.814 & 0.013 & 0.070 & 2.069 & 0.082 & 0.506 & 0.014 & 0.678 & 0.013 & 0.411 & 0.011 & 0.0478 & 0.0478&cgcg015 - 030ned + 77d & 12:53:23.57 & -00:25:30.26 & 16.817 & 0.019 & 0.070 & 0.982 & 0.064 & 0.952 & 0.022 & -0.425 & 0.025 & -0.253 & 0.072 & 0.0471 & & + 77e & 12:53:24.16 & -00:25:12.80 & 17.403 & 0.025 & 0.070 & 0.862 & 0.069 & 0.657 & 0.029 & -0.584 & 0.039 & -1.199 & 0.308 & & & + & & & & & & & & & & & & & & & & + 78a & 12:54:37.84 & -00:11:04.46 & 15.256 & 0.006 & 0.074 & 1.957 & 0.022 & 0.900 & 0.007 & 0.417 & 0.007 & 0.368 & 0.004 & 0.0826 & 0.0826&2masx j12543782 - 0011048 + 78b & 12:54:35.06 & -00:10:55.59 & 17.091 & 0.007 & 0.075 & 1.884 & 0.046 & 0.933 & 0.008 & 0.435 & 0.008 & 0.374 & 0.007 & & 0.0819&2masx j12543503 - 0010558 + 78c & 12:54:35.55 & -00:10:58.61 & 17.827 & 0.011 & 0.075 & 1.643 & 0.098 & 0.896 & 0.016 & 0.376 & 0.013 & 0.254 & 0.024 & 0.0818 & & + 78d & 12:54:34.79 & -00:10:22.60 & 17.896 & 0.011 & 0.075 & 1.701 & 0.103 & 0.853 & 0.015 & 0.360 & 0.013 & 0.318 & 0.023 & & & + & & & & & & & & & & & & & & & & + 79a & 12:55:09.54 & -01:05:27.99 & 17.543 & 0.009 & 0.053 & 1.463 & 0.093 & 1.522 & 0.016 & 0.193 & 0.010 & 0.462 & 0.013 & 0.1229 & & + 79b & 12:55:09.20 & -01:05:30.48 & 17.724 & 0.011 & 0.053 & -2.367 & 0.359 & 4.651 & 0.352 & 0.311 & 0.013 & 0.415 & 0.018 & & & + 79c & 12:55:09.42 & -01:05:29.72 & 17.894 & 0.010 & 0.053 & 2.540 & 0.132 & 0.550 & 0.012 & 0.571 & 0.011 & 0.045 & 0.016 & & & + 79d & 12:55:09.33 & -01:05:12.19 & 19.385 & 0.025 & 0.053 & 1.177 & 0.146 & 0.596 & 0.035 & 0.407 & 0.032 & 0.349 & 0.054 & & & + 79e & 12:55:07.83 & -01:05:30.36 & 20.251 & 0.057 & 0.053 & -0.105 & 0.197 & 1.039 & 0.114 & 0.293 & 0.082 & 0.177 & 0.178 & & & + & & & & & & & & & & & & & & & & + 80a & 12:56:17.15 & -00:38:22.34 & 17.418 & 0.009 & 0.055 & 1.723 & 0.072 & 0.892 & 0.011 & 0.385 & 0.010 & 0.326 & 0.012 & 0.1349 & & + 80b & 12:56:16.52 & -00:38:23.96 & 19.447 & 0.024 & 0.055 & 2.206 & 1.089 & 1.718 & 0.066 & 0.645 & 0.029 & 0.353 & 0.044 & & & + 80c & 12:56:16.35 & -00:38:17.87 & 20.145 & 0.044 & 0.055 & 1.186 & 0.483 & 1.075 & 0.078 & 0.552 & 0.055 & 0.414 & 0.091 & & & + 80d & 12:56:17.27 & -00:38:27.61 & 20.173 & 0.039 & 0.055 & 3.398 & 1.075 & 2.179 & 0.165 & 0.668 & 0.047 & 0.611 & 0.060 & & & + & & & & & & & & & & & & & & & & + 81a & 12:57:59.93 & -01:05:25.87 & 17.091 & 0.011 & 0.061 & 2.023 & 0.166 & 1.249 & 0.018 & 0.462 & 0.012 & 0.362 & 0.016 & & & + 81b & 12:58:00.07 & -01:05:35.18 & 17.970 & 0.013 & 0.061 & 2.007 & 0.259 & 1.280 & 0.027 & 0.546 & 0.016 & 0.379 & 0.022 & & & + 81c & 12:58:00.27 & -01:05:38.65 & 19.322 & 0.021 & 0.061 & 1.893 & 0.506 & 1.472 & 0.056 & 0.592 & 0.025 & 0.328 & 0.040 & & & + 81d & 12:58:00.45 & -01:05:59.65 & 19.348 & 0.028 & 0.061 & 1.151 & 0.201 & 0.799 & 0.048 & 0.341 & 0.038 & 0.123 & 0.086 & & & + 81e & 12:57:58.75 & -01:05:43.90 & 19.661 & 0.026 & 0.061 & 1.434 & 0.277 & 0.951 & 0.049 & 0.374 & 0.034 & 0.394 & 0.061 & & & + & & & & & & & & & & & & & & & & + 82a & 13:05:41.28 & 00:23:53.46 & 17.247 & 0.011 & 0.056 & 2.284 & 0.260 & 1.410 & 0.017 & 0.609 & 0.012 & 0.352 & 0.017 & 0.2232 & & + 82b & 13:05:41.98 & 00:23:40.69 & 18.738 & 0.022 & 0.056 & 1.836 & 0.474 & 1.371 & 0.042 & 0.546 & 0.026 & 0.289 & 0.056 & & & + 82c & 13:05:40.89 & 00:23:58.66 & 18.802 & 0.016 & 0.056 & 1.417 & 0.139 & 0.853 & 0.022 & 0.514 & 0.019 & 0.172 & 0.042 & & & + 82d & 13:05:41.71 & 00:24:04.95 & 19.609 & 0.029 & 0.056 & 1.496 & 0.347 & 0.913 & 0.046 & 0.420 & 0.039 & 0.193 & 0.101 & & & + & & & & & & & & & & & & & & & & + 83a & 13:06:51.81 & -00:33:14.30 & 16.938 & 0.007 & 0.052 & 1.874 & 0.066 & 1.060 & 0.009 & 0.472 & 0.008 & 0.357 & 0.008 & 0.1283 & 0.1279 & 2masx j13065177 - 0033141 + 83b & 13:06:50.98 & -00:33:39.49 & 18.159 & 0.015 & 0.052 & 1.677 & 0.206 & 1.032 & 0.023 & 0.532 & 0.018 & 0.350 & 0.029 & & & + 83c & 13:06:50.45 & -00:33:40.76 & 19.524 & 0.042 & 0.052 & 1.899 & 0.778 & 0.925 & 0.069 & 0.536 & 0.055 & 0.446 & 0.091 & & & + 83d & 13:06:50.32 & -00:33:35.88 & 19.802 & 0.032 & 0.052 & 1.497 & 0.434 & 0.981 & 0.054 & 0.391 & 0.044 & 0.228 & 0.093 & & & + & & & & & & & & & & & & & & & & + 84a & 13:10:38.19 & -00:39:24.43 & 15.172 & 0.007 & 0.066 & 2.077 & 0.030 & 0.987 & 0.007 & 0.414 & 0.007 & 0.381 & 0.003 & 0.0864 & 0.0864&2masx j13103821 - 0039242 + 84b & 13:10:42.28 & -00:39:52.36 & 16.237 & 0.010 & 0.067 & 1.211 & 0.025 & 0.521 & 0.011 & 0.391 & 0.011 & 0.235 & 0.011 & 0.0762 & & + 84c & 13:10:42.62 & -00:40:00.17 & 16.579 & 0.008 & 0.067 & 1.857 & 0.051 & 1.036 & 0.009 & 0.455 & 0.008 & 0.383 & 0.006 & 0.0805 & & + 84d & 13:10:41.73 & -00:40:10.19 & 16.692 & 0.007 & 0.067 & 2.038 & 0.044 & 0.983 & 0.008 & 0.443 & 0.007 & 0.383 & 0.004 & & & + 84e & 13:10:37.20 & -00:39:55.54 & 18.087 & 0.011 & 0.066 & 1.897 & 0.130 & 0.893 & 0.014 & 0.358 & 0.012 & 0.308 & 0.019 & & & + & & & & & & & & & & & & & & & & + 85a & 13:12:05.54 & 00:32:19.40 & 15.300 & 0.007 & 0.085 & 0.996 & 0.021 & 0.598 & 0.008 & 0.144 & 0.008 & 0.128 & 0.016 & & & + 85b & 13:12:07.14 & 00:32:01.09 & 15.684 & 0.011 & 0.086 & 1.069 & 0.029 & 0.682 & 0.012 & 0.211 & 0.012 & 0.184 & 0.017 & 0.1214 & & + 85c & 13:12:03.50 & 00:32:49.55 & 16.873 & 0.008 & 0.085 & 1.780 & 0.049 & 0.944 & 0.009 & 0.413 & 0.009 & 0.352 & 0.008 & 0.1203 & & + 85d & 13:12:05.07 & 00:31:40.65 & 17.657 & 0.020 & 0.085 & 1.040 & 0.128 & 0.877 & 0.030 & 0.334 & 0.027 & 0.253 & 0.054 & & & + & & & & & & & & & & & & & & & & + 86a & 13:12:25.88 & -01:06:13.07 & 16.508 & 0.007 & 0.092 & 1.973 & 0.042 & 0.904 & 0.008 & 0.400 & 0.007 & 0.347 & 0.006 & & 0.1087&apmuks(bj ) b130952.04 - 00 ned02 + 86b & 13:12:26.40 & -01:06:20.04 & 16.943 & 0.008 & 0.093 & 1.838 & 0.063 & 0.946 & 0.010 & 0.421 & 0.009 & 0.447 & 0.009 & & 0.1095&pmuks(bj ) b130952.04 - 00 ned01 + 86c & 13:12:23.62 & -01:06:18.03 & 17.147 & 0.008 & 0.093 & 2.052 & 0.068 & 0.964 & 0.009 & 0.418 & 0.008 & 0.390 & 0.008 & 0.1091 & & + 86d & 13:12:23.63 & -01:06:05.78 & 18.516 & 0.018 & 0.092 & 1.021 & 0.082 & 0.534 & 0.025 & 0.291 & 0.024 & 0.163 & 0.051 & & & + & & & & & & & & & & & & & & & & + 87a & 13:13:04.19 & -01:13:31.40 & 16.066 & 0.009 & 0.069 & 1.388 & 0.044 & 0.794 & 0.011 & 0.352 & 0.010 & 0.088 & 0.016 & & & + 87b & 13:13:03.95 & -01:13:22.15 & 16.803 & 0.007 & 0.069 & 1.997 & 0.062 & 1.031 & 0.009 & 0.452 & 0.008 & 0.423 & 0.007 & & & + 87c & 13:13:00.75 & -01:13:24.01 & 17.859 & 0.010 & 0.072 & 1.932 & 0.148 & 1.095 & 0.016 & 0.478 & 0.012 & 0.406 & 0.017 & & & + 87d & 13:13:03.60 & -01:13:37.78 & 18.913 & 0.015 & 0.069 & 1.505 & 0.137 & 0.775 & 0.023 & 0.461 & 0.019 & 0.290 & 0.033 & & & + & & & & & & & & & & & & & & & & + 88a & 13:15:01.99 & -01:04:06.89 & 17.202 & 0.008 & 0.081 & 2.031 & 0.117 & 1.441 & 0.012 & 0.480 & 0.009 & 0.295 & 0.009 & & & + 88b & 13:15:02.57 & -01:04:04.10 & 17.688 & 0.010 & 0.081 & 1.633 & 0.206 & 1.900 & 0.025 & 0.490 & 0.011 & 0.293 & 0.015 & 0.2148 & & + 88c & 13:15:02.25 & -01:04:10.78 & 19.017 & 0.027 & 0.081 & 0.767 & 0.207 & 1.210 & 0.058 & 0.299 & 0.039 & 0.549 & 0.062 & & & + 88d & 13:15:02.30 & -01:04:00.91 & 19.043 & 0.016 & 0.081 & 0.786 & 0.352 & 2.479 & 0.086 & 0.327 & 0.021 & 0.235 & 0.039 & & & + & & & & & & & & & & & & & & & & + 89a & 13:18:16.20 & 00:19:39.78 & 16.237 & 0.006 & 0.082 & 1.984 & 0.033 & 0.928 & 0.006 & 0.405 & 0.006 & 0.328 & 0.006 & 0.0816 & & + 89b & 13:18:15.82 & 00:19:35.94 & 17.103 & 0.013 & 0.082 & 0.958 & 0.049 & 0.521 & 0.016 & 0.260 & 0.016 & 0.110 & 0.045 & & & + 89c & 13:18:14.65 & 00:19:35.99 & 17.664 & 0.009 & 0.083 & 1.169 & 0.035 & 0.501 & 0.011 & 0.284 & 0.011 & 0.206 & 0.024 & & & + 89d & 13:18:15.15 & 00:19:28.17 & 18.610 & 0.014 & 0.083 & 1.621 & 0.151 & 0.824 & 0.020 & 0.317 & 0.018 & 0.345 & 0.042 & & & + & & & & & & & & & & & & & & & & + 90a & 13:18:44.74 & -00:44:01.60 & 16.029 & 0.008 & 0.076 & 2.051 & 0.045 & 0.929 & 0.008 & 0.446 & 0.008 & 0.421 & 0.005 & 0.0859 & & + 90b & 13:18:44.07 & -00:43:55.98 & 18.335 & 0.011 & 0.076 & 1.743 & 0.101 & 0.904 & 0.013 & 0.404 & 0.012 & 0.545 & 0.013 & & & + 90c & 13:18:45.06 & -00:44:07.36 & 18.386 & 0.021 & 0.077 & 0.836 & 0.144 & 1.017 & 0.034 & 0.403 & 0.027 & 0.301 & 0.052 & & & + 90d & 13:18:45.55 & -00:43:25.95 & 18.548 & 0.013 & 0.075 & 1.410 & 0.122 & 0.832 & 0.017 & 0.401 & 0.015 & 0.386 & 0.027 & & & + & & & & & & & & & & & & & & & & + 91a & 13:18:46.54 & -00:45:27.13 & 17.233 & 0.008 & 0.079 & 2.093 & 0.073 & 0.949 & 0.009 & 0.448 & 0.009 & 0.441 & 0.007 & 0.0870 & & + 91b & 13:18:46.47 & -00:45:35.43 & 17.854 & 0.011 & 0.079 & 1.666 & 0.102 & 0.906 & 0.013 & 0.393 & 0.012 & 0.414 & 0.016 & & & + 91c & 13:18:47.73 & -00:45:27.31 & 19.566 & 0.023 & 0.079 & 1.696 & 0.337 & 0.877 & 0.034 & 0.406 & 0.029 & 0.423 & 0.053 & & & + 91d & 13:18:47.29 & -00:45:24.93 & 20.020 & 0.048 & 0.079 & 0.991 & 0.977 & 1.826 & 0.151 & 0.587 & 0.061 & 0.523 & 0.100 & & & + & & & & & & & & & & & & & & & & + 92a & 13:19:34.20 & -01:05:20.06 & 15.692 & 0.006 & 0.072 & 1.931 & 0.024 & 0.921 & 0.007 & 0.472 & 0.007 & 0.324 & 0.003 & & & + 92b & 13:19:34.04 & -01:05:51.25 & 16.400 & 0.009 & 0.073 & 1.768 & 0.072 & 0.887 & 0.012 & 0.383 & 0.011 & 0.273 & 0.014 & 0.0802 & & + 92c & 13:19:31.48 & -01:06:11.74 & 17.797 & 0.010 & 0.073 & 1.516 & 0.069 & 0.843 & 0.013 & 0.442 & 0.012 & 0.279 & 0.016 & & & + 92d & 13:19:31.98 & -01:06:05.29 & 18.459 & 0.012 & 0.073 & 0.980 & 0.056 & 0.752 & 0.016 & 0.425 & 0.014 & 0.216 & 0.023 & & & + & & & & & & & & & & & & & & & & + 93a & 13:24:39.79 & 00:22:15.96 & 16.152 & 0.006 & 0.066 & 2.104 & 0.047 & 1.019 & 0.007 & 0.447 & 0.007 & 0.380 & 0.006 & 0.1080 & & + 93b & 13:24:38.83 & 00:22:12.68 & 18.266 & 0.015 & 0.066 & 1.120 & 0.078 & 0.634 & 0.018 & 0.486 & 0.018 & 0.379 & 0.036 & & & + 93c & 13:24:38.32 & 00:21:51.72 & 18.794 & 0.012 & 0.065 & 2.036 & 0.193 & 1.029 & 0.016 & 0.463 & 0.014 & 0.407 & 0.025 & & & + 93d & 13:24:38.72 & 00:22:03.13 & 19.142 & 0.017 & 0.066 & 1.574 & 0.200 & 0.946 & 0.024 & 0.354 & 0.021 & 0.452 & 0.047 & & & + & & & & & & & & & & & & & & & & + 94a & 13:24:49.00 & -00:00:35.68 & 16.556 & 0.006 & 0.065 & 2.033 & 0.045 & 1.011 & 0.007 & 0.495 & 0.007 & 0.411 & 0.006 & 0.0816 & & + 94b & 13:24:48.74 & -00:00:54.77 & 17.450 & 0.009 & 0.065 & 1.880 & 0.084 & 0.908 & 0.011 & 0.381 & 0.010 & 0.326 & 0.016 & & & + 94c & 13:24:50.01 & -00:00:53.71 & 19.153 & 0.016 & 0.065 & 2.062 & 0.263 & 0.875 & 0.022 & 0.370 & 0.020 & 0.297 & 0.043 & & & + 94d & 13:24:50.17 & -00:00:31.12 & 19.404 & 0.020 & 0.065 & 1.783 & 0.259 & 0.866 & 0.029 & 0.360 & 0.026 & 0.317 & 0.057 & & & + & & & & & & & & & & & & & & & & + 95a & 13:25:07.78 & -00:38:23.69 & 16.967 & 0.009 & 0.084 & 1.898 & 0.080 & 1.081 & 0.010 & 0.536 & 0.009 & 0.433 & 0.008 & 0.0849 & & + 95b & 13:25:09.33 & -00:38:26.99 & 18.346 & 0.011 & 0.084 & 1.740 & 0.101 & 0.870 & 0.014 & 0.427 & 0.013 & 0.338 & 0.016 & & & + 95c & 13:25:08.60 & -00:38:19.90 & 19.010 & 0.144 & 0.084 & -0.618 & 0.339 & 1.210 & 0.244 & 0.297 & 0.189 & -0.045 & 0.492 & & & + 95d & 13:25:08.89 & -00:38:42.30 & 19.647 & 0.027 & 0.084 & 1.096 & 0.138 & 0.433 & 0.033 & 0.314 & 0.036 & 0.088 & 0.085 & & & + & & & & & & & & & & & & & & & & + 96a & 13:26:34.98 & 00:26:44.27 & 17.856 & 0.011 & 0.071 & 1.852 & 0.169 & 1.400 & 0.017 & 0.522 & 0.012 & 0.331 & 0.016 & & & + 96b & 13:26:34.55 & 00:26:37.96 & 20.203 & 0.036 & 0.071 & 1.631 & 0.635 & 1.303 & 0.075 & 0.536 & 0.047 & 0.416 & 0.076 & & & + 96c & 13:26:34.02 & 00:26:50.54 & 20.485 & 0.056 & 0.071 & 1.030 & 0.355 & 0.737 & 0.083 & 0.253 & 0.084 & 0.322 & 0.170 & & & + 96d & 13:26:34.17 & 00:26:46.83 & 20.704 & 0.081 & 0.071 & 4.305 & 0.959 & 1.150 & 0.162 & 0.633 & 0.106 & 0.244 & 0.199 & & & + & & & & & & & & & & & & & & & & + 97a & 13:28:20.16 & 00:26:18.66 & 17.285 & 0.011 & 0.077 & 2.531 & 0.280 & 1.462 & 0.016 & 0.592 & 0.012 & 0.407 & 0.012 & & & + 97b & 13:28:19.00 & 00:26:11.87 & 19.328 & 0.017 & 0.077 & 2.615 & 0.627 & 1.410 & 0.030 & 0.666 & 0.020 & 0.343 & 0.025 & & & + 97c & 13:28:19.69 & 00:26:17.54 & 19.578 & 0.027 & 0.077 & 2.234 & 0.794 & 1.297 & 0.052 & 0.508 & 0.036 & 0.496 & 0.053 & & & + 97d & 13:28:19.28 & 00:26:09.73 & 20.147 & 0.031 & 0.077 & 2.038 & 0.776 & 1.322 & 0.062 & 0.628 & 0.040 & 0.267 & 0.066 & & & + & & & & & & & & & & & & & & & & + 98a & 13:29:06.66 & 00:25:36.30 & 17.860 & 0.010 & 0.077 & 0.998 & 0.028 & 0.424 & 0.011 & 0.288 & 0.012 & 0.128 & 0.020 & & & + 98b & 13:29:06.38 & 00:25:29.16 & 17.985 & 0.011 & 0.077 & 2.083 & 0.199 & 1.374 & 0.016 & 0.477 & 0.012 & 0.366 & 0.015 & & & + 98c & 13:29:06.30 & 00:25:40.53 & 19.672 & 0.033 & 0.077 & 0.856 & 0.321 & 1.361 & 0.069 & 0.651 & 0.042 & 0.449 & 0.059 & & & + 98d & 13:29:05.40 & 00:25:37.74 & 20.324 & 0.039 & 0.077 & 1.802 & 0.502 & 0.757 & 0.057 & 0.428 & 0.054 & 0.193 & 0.107 & & & + & & & & & & & & & & & & & & & & + 99a & 13:34:36.92 & -01:03:41.08 & 16.928 & 0.007 & 0.087 & 1.374 & 0.022 & 0.692 & 0.008 & 0.406 & 0.008 & 0.285 & 0.006 & & 0.0768 & 2dfgrsn267z146 + 99b & 13:34:35.84 & -01:03:10.61 & 17.680 & 0.012 & 0.087 & 1.058 & 0.049 & 0.688 & 0.015 & 0.346 & 0.014 & 0.218 & 0.023 & & & + 99c & 13:34:37.23 & -01:03:27.12 & 19.791 & 0.068 & 0.087 & 3.261 & 3.127 & 2.003 & 0.256 & 0.466 & 0.085 & 0.263 & 0.158 & & & + 99d & 13:34:36.61 & -01:03:23.53 & 19.867 & 0.031 & 0.087 & 1.108 & 0.163 & 0.351 & 0.041 & 0.265 & 0.044 & 0.042 & 0.115 & & & + & & & & & & & & & & & & & & & & + 100a & 13:35:07.92 & 00:28:18.30 & 15.533 & 0.010 & 0.068 & 1.253 & 0.038 & 0.673 & 0.011 & 0.319 & 0.011 & 0.136 & 0.017 & & 0.0869 & 2dfgrsn334z162 + 100b & 13:35:08.02 & 00:28:38.40 & 17.015 & 0.008 & 0.068 & 1.145 & 0.022 & 0.571 & 0.009 & 0.489 & 0.009 & 0.200 & 0.009 & & & + 100c & 13:35:05.78 & 00:28:25.48 & 17.927 & 0.010 & 0.069 & 1.831 & 0.097 & 0.807 & 0.012 & 0.453 & 0.012 & 0.264 & 0.016 & & & + 100d & 13:35:04.60 & 00:28:04.00 & 18.074 & 0.012 & 0.069 & 1.579 & 0.125 & 0.791 & 0.017 & 0.472 & 0.016 & 0.189 & 0.028 & & & + & & & & & & & & & & & & & & & & + 101a & 13:38:05.29 & 00:57:47.09 & 17.406 & 0.010 & 0.067 & 1.620 & 0.092 & 1.077 & 0.013 & 0.450 & 0.011 & 0.327 & 0.015 & & 0.1430 & 2dfgrs n402z076 + 101b & 13:38:04.47 & 00:58:01.07 & 18.310 & 0.033 & 0.067 & 0.952 & 0.214 & 1.008 & 0.048 & 0.194 & 0.045 & 0.389 & 0.082 & & & + 101c & 13:38:03.81 & 00:57:59.40 & 19.594 & 0.024 & 0.067 & 2.268 & 1.063 & 1.826 & 0.066 & 0.565 & 0.030 & 0.398 & 0.050 & & & + 101d & 13:38:04.19 & 00:58:07.01 & 20.038 & 0.040 & 0.067 & 1.244 & 0.340 & 0.799 & 0.059 & 0.366 & 0.056 & 0.422 & 0.103 & & & + & & & & & & & & & & & & & & & & + 102a & 13:38:42.68 & -00:44:22.02 & 17.055 & 0.011 & 0.089 & 2.165 & 0.333 & 1.803 & 0.022 & 0.570 & 0.012 & 0.409 & 0.013 & 0.3462 & & + 102b & 13:38:41.85 & -00:44:35.82 & 18.537 & 0.018 & 0.089 & 1.925 & 0.798 & 2.142 & 0.060 & 0.670 & 0.021 & 0.050 & 0.036 & & & + 102c & 13:38:41.55 & -00:44:36.52 & 18.776 & 0.018 & 0.089 & 1.680 & 0.765 & 2.364 & 0.072 & 0.562 & 0.021 & 0.007 & 0.041 & & & + 102d & 13:38:42.12 & -00:44:45.93 & 19.752 & 0.028 & 0.088 & 1.445 & 0.674 & 1.807 & 0.080 & 0.618 & 0.034 & 0.421 & 0.049 & & & + & & & & & & & & & & & & & & & & + 103a & 13:38:44.08 & 00:16:33.82 & 17.712 & 0.009 & 0.080 & 1.280 & 0.046 & 0.767 & 0.011 & 0.372 & 0.010 & 0.332 & 0.019 & & 0.1290 & 2dfgrsn335z192 + 103b & 13:38:44.81 & 00:16:21.54 & 17.796 & 0.025 & 0.080 & 1.688 & 0.216 & 0.632 & 0.031 & 0.123 & 0.034 & 0.190 & 0.096 & & & + 103c & 13:38:44.56 & 00:16:48.09 & 20.109 & 0.256 & 0.080 & 0.188 & 0.877 & 1.200 & 0.379 & 0.435 & 0.301 & 0.867 & 0.383 & & & + 103d & 13:38:45.09 & 00:16:19.58 & 20.416 & 0.041 & 0.080 & 1.154 & 0.225 & 0.479 & 0.052 & 0.400 & 0.054 & 0.420 & 0.115 & & & + & & & & & & & & & & & & & & & & + 104a & 13:39:39.20 & -00:01:02.01 & 17.989 & 0.010 & 0.073 & 2.095 & 0.201 & 1.362 & 0.016 & 0.480 & 0.012 & 0.304 & 0.018 & & & + 104b & 13:39:38.81 & -00:01:13.28 & 18.961 & 0.031 & 0.073 & 2.888 & 1.502 & 1.231 & 0.056 & 0.621 & 0.038 & 0.335 & 0.067 & & & + 104c & 13:39:38.69 & -00:01:03.44 & 20.183 & 0.037 & 0.073 & 2.397 & 1.103 & 1.288 & 0.072 & 0.435 & 0.048 & 0.138 & 0.116 & & & + 104d & 13:39:38.88 & -00:00:51.07 & 20.691 & 0.059 & 0.073 & 0.811 & 0.376 & 0.921 & 0.094 & 0.341 & 0.082 & 0.324 & 0.183 & & & + & & & & & & & & & & & & & & & & + 105a & 13:40:29.51 & -01:12:13.56 & 17.098 & 0.009 & 0.073 & 1.802 & 0.042 & 1.026 & 0.010 & 0.437 & 0.009 & 0.339 & 0.006 & & & + 105b & 13:40:30.07 & -01:12:17.48 & 17.273 & 0.010 & 0.073 & 0.908 & 0.017 & 0.507 & 0.010 & 0.402 & 0.010 & 0.162 & 0.011 & & & + 105c & 13:40:28.27 & -01:11:54.68 & 17.370 & 0.010 & 0.073 & 1.961 & 0.073 & 1.017 & 0.011 & 0.430 & 0.010 & 0.345 & 0.009 & & & + 105d & 13:40:27.75 & -01:11:56.59 & 19.667 & 0.040 & 0.072 & 0.591 & 0.316 & 1.345 & 0.100 & 0.637 & 0.050 & 0.451 & 0.076 & & & + & & & & & & & & & & & & & & & & + 106a & 13:40:58.27 & 00:45:28.10 & 17.454 & 0.009 & 0.079 & 1.673 & 0.121 & 1.407 & 0.014 & 0.532 & 0.011 & 0.348 & 0.013 & 0.2417 & & + 106b & 13:40:58.13 & 00:45:31.34 & 18.020 & 0.014 & 0.079 & 1.864 & 0.266 & 1.341 & 0.025 & 0.448 & 0.018 & 0.237 & 0.031 & & 0.0902 & 2dfgrsn336z062 + 106c & 13:40:59.05 & 00:45:44.80 & 18.990 & 0.016 & 0.079 & 0.900 & 0.064 & 0.523 & 0.019 & 0.289 & 0.021 & 0.224 & 0.042 & & & + 106d & 13:40:58.39 & 00:45:18.45 & 20.113 & 0.028 & 0.079 & 1.244 & 0.368 & 1.373 & 0.055 & 0.666 & 0.034 & 0.332 & 0.054 & & & + & & & & & & & & & & & & & & & & + 107a & 13:41:06.68 & -00:01:39.22 & 17.635 & 0.008 & 0.080 & 1.955 & 0.066 & 0.915 & 0.009 & 0.405 & 0.009 & 0.341 & 0.010 & 0.1003 & & + 107b & 13:41:06.15 & -00:01:47.93 & 18.459 & 0.013 & 0.080 & 1.759 & 0.269 & 1.488 & 0.025 & 0.580 & 0.015 & 0.405 & 0.023 & & & + 107c & 13:41:06.29 & -00:01:31.52 & 19.071 & 0.023 & 0.080 & 1.458 & 0.200 & 0.713 & 0.031 & 0.386 & 0.030 & 0.252 & 0.063 & & & + 107d & 13:41:06.64 & -00:01:22.35 & 19.835 & 0.040 & 0.080 & 0.710 & 0.245 & 1.017 & 0.066 & 0.390 & 0.054 & 0.271 & 0.118 & & & + & & & & & & & & & & & & & & & & + 108a & 13:41:25.48 & -01:10:29.66 & 15.334 & 0.008 & 0.075 & 1.720 & 0.028 & 0.859 & 0.009 & 0.442 & 0.009 & 0.289 & 0.005 & & & + 108b & 13:41:27.13 & -01:09:56.12 & 15.930 & 0.008 & 0.075 & 1.975 & 0.024 & 0.908 & 0.008 & 0.456 & 0.008 & 0.332 & 0.003 & 0.0888 & & + 108c & 13:41:28.09 & -01:10:59.94 & 17.428 & 0.011 & 0.075 & 1.594 & 0.079 & 1.036 & 0.014 & 0.531 & 0.011 & 0.323 & 0.013 & & & + 108d & 13:41:26.65 & -01:09:42.75 & 17.934 & 0.012 & 0.075 & 1.910 & 0.154 & 1.107 & 0.018 & 0.562 & 0.013 & 0.370 & 0.016 & & & + & & & & & & & & & & & & & & & & + 109a & 13:42:29.65 & -00:51:29.30 & 17.724 & 0.010 & 0.078 & 2.236 & 0.193 & 1.212 & 0.018 & 0.540 & 0.012 & 0.312 & 0.016 & & & + 109b & 13:42:30.71 & -00:51:25.50 & 19.267 & 0.020 & 0.078 & 1.127 & 0.151 & 1.003 & 0.034 & 0.288 & 0.027 & 0.403 & 0.049 & & & + 109c & 13:42:30.81 & -00:51:35.18 & 19.578 & 0.023 & 0.079 & 3.268 & 1.043 & 1.020 & 0.041 & 0.523 & 0.029 & 0.279 & 0.052 & & & + 109d & 13:42:30.05 & -00:51:40.51 & 19.809 & 0.026 & 0.078 & 3.909 & 1.024 & 1.591 & 0.071 & 0.581 & 0.032 & 0.338 & 0.054 & & & + 109e & 13:42:30.16 & -00:51:18.40 & 20.509 & 0.046 & 0.078 & 0.937 & 0.341 & 0.986 & 0.089 & 0.435 & 0.064 & 0.100 & 0.149 & & & + & & & & & & & & & & & & & & & & + 110a & 13:42:29.88 & 00:21:38.19 & 16.200 & 0.012 & 0.082 & 1.352 & 0.072 & 0.856 & 0.015 & 0.331 & 0.014 & 0.413 & 0.025 & & & + 110b & 13:42:31.11 & 00:21:43.91 & 16.956 & 0.009 & 0.082 & 2.155 & 0.182 & 1.484 & 0.013 & 0.521 & 0.010 & 0.354 & 0.015 & 0.2434 & 0.2434 & 2dfgrsn336z227 + 110c & 13:42:29.92 & 00:20:54.72 & 18.599 & 0.017 & 0.081 & 1.635 & 0.281 & 1.308 & 0.030 & 0.513 & 0.021 & 0.373 & 0.042 & & & + 110d & 13:42:30.95 & 00:21:24.65 & 18.988 & 0.016 & 0.082 & 1.667 & 0.265 & 1.318 & 0.027 & 0.465 & 0.019 & 0.341 & 0.041 & & & + & & & & & & & & & & & & & & & & + 111a & 13:42:43.36 & 00:45:56.41 & 15.528 & 0.012 & 0.082 & 1.410 & 0.065 & 0.829 & 0.015 & 0.154 & 0.015 & 0.125 & 0.027 & 0.0732 & 0.0733 & 2dfgrsn335z003 + 111b & 13:42:45.27 & 00:45:21.27 & 18.187 & 0.012 & 0.083 & 1.809 & 0.144 & 1.061 & 0.017 & 0.448 & 0.014 & 0.337 & 0.021 & & & + 111c & 13:42:42.47 & 00:46:02.24 & 18.304 & 0.014 & 0.082 & 1.389 & 0.094 & 0.720 & 0.018 & 0.393 & 0.018 & 0.259 & 0.032 & & & + 111d & 13:42:43.25 & 00:45:52.36 & 18.526 & 0.021 & 0.082 & 2.298 & 0.619 & 1.343 & 0.041 & 0.741 & 0.025 & 0.192 & 0.040 & & & + & & & & & & & & & & & & & & & & + 112a & 13:45:06.50 & -00:10:14.63 & 17.956 & 0.011 & 0.069 & 1.641 & 0.086 & 0.877 & 0.014 & 0.395 & 0.013 & 0.341 & 0.019 & & & + 112b & 13:45:06.57 & -00:10:25.39 & 18.845 & 0.019 & 0.069 & 5.013 & 1.575 & 1.694 & 0.047 & 0.600 & 0.023 & 0.437 & 0.036 & & & + 112c & 13:45:07.10 & -00:10:11.62 & 20.193 & 0.035 & 0.069 & 0.831 & 0.510 & 1.814 & 0.109 & 0.571 & 0.045 & 0.513 & 0.075 & & & + 112d & 13:45:07.38 & -00:10:09.38 & 20.726 & 0.063 & 0.069 & 0.692 & 0.443 & 1.110 & 0.118 & 0.359 & 0.089 & 0.521 & 0.168 & & & + & & & & & & & & & & & & & & & & + 113a & 13:45:52.34 & -00:01:16.42 & 17.597 & 0.009 & 0.074 & 0.992 & 0.019 & 0.424 & 0.009 & 0.339 & 0.010 & 0.111 & 0.015 & 0.0908 & & + 113b & 13:45:51.36 & -00:01:22.94 & 19.615 & 0.024 & 0.074 & 2.225 & 0.705 & 1.319 & 0.045 & 0.467 & 0.030 & 0.270 & 0.063 & & & + 113c & 13:45:52.47 & -00:01:08.65 & 20.072 & 0.031 & 0.074 & 1.071 & 0.280 & 1.042 & 0.052 & 0.404 & 0.041 & 0.497 & 0.076 & & & + 113d & 13:45:51.89 & -00:01:21.60 & 20.143 & 0.042 & 0.074 & 0.570 & 0.190 & 0.790 & 0.061 & 0.168 & 0.062 & 0.631 & 0.118 & & & + & & & & & & & & & & & & & & & & + 114a & 13:49:56.20 & -00:27:56.57 & 17.642 & 0.010 & 0.082 & 2.089 & 0.269 & 1.582 & 0.018 & 0.559 & 0.011 & 0.403 & 0.014 & 0.2602 & & + 114b & 13:49:56.14 & -00:28:19.89 & 18.748 & 0.019 & 0.082 & 2.342 & 0.696 & 1.419 & 0.036 & 0.516 & 0.023 & 0.126 & 0.046 & & & + 114c & 13:49:55.77 & -00:28:20.12 & 19.789 & 0.030 & 0.082 & 1.089 & 0.193 & 0.628 & 0.040 & 0.257 & 0.042 & 0.004 & 0.114 & & & + 114d & 13:49:56.29 & -00:28:19.16 & 20.379 & 0.050 & 0.082 & 0.600 & 0.541 & 1.635 & 0.128 & 0.767 & 0.060 & 0.410 & 0.088 & & & + & & & & & & & & & & & & & & & & + 115a & 13:50:51.64 & -00:31:05.89 & 16.787 & 0.007 & 0.097 & 1.651 & 0.040 & 0.876 & 0.008 & 0.450 & 0.008 & 0.371 & 0.007 & & & + 115b & 13:50:49.89 & -00:30:58.02 & 17.467 & 0.011 & 0.097 & 1.458 & 0.079 & 0.983 & 0.013 & 0.468 & 0.013 & 0.375 & 0.015 & 0.1504 & & + 115c & 13:50:50.26 & -00:31:02.61 & 18.264 & 0.045 & 0.097 & 1.218 & 0.204 & 0.331 & 0.052 & -0.230 & 0.076 & 1.396 & 0.089 & & & + 115d & 13:50:50.46 & -00:31:01.68 & 19.323 & 0.024 & 0.097 & 2.303 & 0.900 & 1.374 & 0.048 & 0.488 & 0.031 & -0.183 & 0.086 & & & + & & & & & & & & & & & & & & & & + 116a & 13:51:33.13 & 00:44:18.26 & 17.795 & 0.013 & 0.088 & 1.335 & 0.068 & 0.652 & 0.015 & 0.405 & 0.016 & 0.252 & 0.025 & & & + 116b & 13:51:31.55 & 00:44:11.80 & 17.921 & 0.008 & 0.088 & 2.463 & 0.201 & 1.588 & 0.012 & 0.514 & 0.009 & 0.395 & 0.009 & & 0.0881 & 2dfgrsn405z061 + 116c & 13:51:32.46 & 00:44:11.82 & 18.466 & 0.014 & 0.088 & 0.938 & 0.043 & 0.257 & 0.016 & 0.178 & 0.020 & 0.133 & 0.044 & & & + 116d & 13:51:31.87 & 00:44:19.28 & 20.045 & 0.025 & 0.088 & 1.016 & 0.101 & 0.258 & 0.029 & 0.247 & 0.038 & 0.046 & 0.094 & & & + & & & & & & & & & & & & & & & & + 117a & 13:52:35.53 & -01:01:42.59 & 17.356 & 0.008 & 0.127 & 1.462 & 0.058 & 1.014 & 0.011 & 0.424 & 0.010 & 0.309 & 0.013 & 0.1412 & & + 117b & 13:52:35.42 & -01:01:34.71 & 17.947 & 0.010 & 0.126 & 1.740 & 0.101 & 1.125 & 0.014 & 0.416 & 0.011 & 0.137 & 0.018 & & & + 117c & 13:52:35.74 & -01:01:47.00 & 18.850 & 0.014 & 0.127 & 1.513 & 0.121 & 0.900 & 0.020 & 0.356 & 0.018 & 0.240 & 0.032 & & & + 117d & 13:52:34.65 & -01:01:52.95 & 19.659 & 0.029 & 0.127 & 1.740 & 0.705 & 1.591 & 0.076 & 0.471 & 0.038 & 0.307 & 0.069 & & & + & & & & & & & & & & & & & & & & + 118a & 13:53:50.83 & 00:11:01.85 & 17.708 & 0.012 & 0.111 & 2.043 & 0.220 & 1.436 & 0.018 & 0.554 & 0.013 & 0.370 & 0.019 & & & + 118b & 13:53:50.25 & 00:10:41.44 & 18.799 & 0.019 & 0.111 & 0.992 & 0.086 & 0.653 & 0.024 & 0.448 & 0.023 & 0.222 & 0.052 & & & + 118c & 13:53:50.11 & 00:10:46.44 & 19.326 & 0.023 & 0.111 & 1.724 & 0.408 & 1.268 & 0.041 & 0.493 & 0.028 & 0.206 & 0.067 & & & + 118d & 13:53:51.21 & 00:10:54.16 & 20.665 & 0.052 & 0.111 & 0.376 & 0.194 & 0.645 & 0.074 & 0.017 & 0.085 & 0.445 & 0.227 & & & + & & & & & & & & & & & & & & & & + 119a & 13:55:57.84 & 00:15:26.30 & 17.359 & 0.015 & 0.108 & 1.388 & 0.075 & 0.645 & 0.018 & 0.388 & 0.018 & 0.138 & 0.036 & & & + 119b & 13:55:58.09 & 00:15:30.64 & 17.858 & 0.012 & 0.108 & 1.451 & 0.094 & 1.083 & 0.015 & 0.459 & 0.013 & 0.337 & 0.023 & 0.1333 & & + 119c & 13:55:57.36 & 00:15:10.46 & 19.398 & 0.019 & 0.108 & 1.186 & 0.128 & 0.771 & 0.025 & 0.389 & 0.024 & 0.223 & 0.061 & & & + 119d & 13:55:58.88 & 00:15:25.41 & 19.739 & 0.030 & 0.108 & 1.013 & 0.157 & 0.543 & 0.039 & 0.372 & 0.041 & 0.105 & 0.123 & & & + 119e & 13:55:57.26 & 00:15:34.54 & 20.037 & 0.027 & 0.108 & 0.944 & 0.150 & 0.695 & 0.037 & 0.385 & 0.036 & 0.301 & 0.087 & & & + & & & & & & & & & & & & & & & & + 120a & 14:05:08.08 & -00:09:02.15 & 17.176 & 0.011 & 0.119 & 1.943 & 0.232 & 1.610 & 0.020 & 0.510 & 0.013 & 0.396 & 0.018 & 0.2427 & 0.2427 & 2masx j14050810 - 0009015 + 120b & 14:05:08.59 & -00:08:56.83 & 19.784 & 0.026 & 0.119 & 1.336 & 0.671 & 2.139 & 0.095 & 0.482 & 0.032 & 0.212 & 0.069 & & & + 120c & 14:05:08.38 & -00:08:58.91 & 20.025 & 0.039 & 0.119 & 1.715 & 0.667 & 1.212 & 0.073 & 0.823 & 0.045 & -0.113 & 0.109 & & & + 120d & 14:05:08.20 & -00:08:54.21 & 20.117 & 0.041 & 0.118 & 1.887 & 0.635 & 0.878 & 0.064 & 0.672 & 0.051 & 0.250 & 0.101 & & & + & & & & & & & & & & & & & & & & + 121a & 14:07:52.93 & -00:29:19.07 & 17.610 & 0.011 & 0.107 & 2.033 & 0.247 & 1.504 & 0.018 & 0.474 & 0.012 & 0.404 & 0.016 & & & + 121b & 14:07:52.47 & -00:29:25.84 & 17.989 & 0.015 & 0.108 & 2.497 & 0.587 & 1.434 & 0.027 & 0.533 & 0.017 & 0.422 & 0.026 & & & + 121c & 14:07:51.80 & -00:29:24.08 & 18.889 & 0.016 & 0.108 & 2.358 & 0.632 & 1.483 & 0.032 & 0.502 & 0.020 & 0.390 & 0.032 & & & + 121d & 14:07:51.51 & -00:29:16.15 & 20.018 & 0.042 & 0.108 & 0.398 & 0.328 & 1.408 & 0.093 & 0.353 & 0.060 & 0.374 & 0.118 & & & + & & & & & & & & & & & & & & & & + 122a & 14:10:54.95 & 00:59:08.92 & 17.536 & 0.009 & 0.091 & 2.138 & 0.182 & 1.182 & 0.013 & 0.492 & 0.010 & 0.368 & 0.015 & 0.1803 & & + 122b & 14:10:54.77 & 00:58:57.22 & 18.476 & 0.017 & 0.091 & 1.707 & 0.355 & 1.351 & 0.033 & 0.601 & 0.020 & 0.548 & 0.028 & & & + 122c & 14:10:54.41 & 00:59:06.29 & 20.240 & 0.041 & 0.091 & 1.005 & 0.321 & 0.768 & 0.063 & 0.371 & 0.058 & 0.213 & 0.135 & & & + 122d & 14:10:54.96 & 00:59:15.08 & 20.277 & 0.036 & 0.091 & 0.424 & 0.133 & 0.554 & 0.050 & 0.336 & 0.051 & -0.138 & 0.159 & & & + & & & & & & & & & & & & & & & & + 123a & 14:12:05.86 & -01:03:21.74 & 17.138 & 0.011 & 0.158 & 2.453 & 0.307 & 1.228 & 0.021 & 0.447 & 0.013 & 0.381 & 0.019 & 0.1822 & & + 123b & 14:12:06.90 & -01:03:26.11 & 18.884 & 0.019 & 0.159 & 1.060 & 0.129 & 0.831 & 0.033 & 0.440 & 0.024 & 0.108 & 0.052 & & & + 123c & 14:12:05.50 & -01:03:23.89 & 19.026 & 0.049 & 0.158 & 0.808 & 0.300 & 0.857 & 0.093 & 0.349 & 0.068 & -0.422 & 0.252 & & & + 123d & 14:12:05.78 & -01:03:36.80 & 19.506 & 0.019 & 0.159 & 2.292 & 0.581 & 1.285 & 0.046 & 0.439 & 0.023 & 0.466 & 0.036 & & & + & & & & & & & & & & & & & & & & + 124a & 14:14:00.24 & 00:23:35.34 & 17.167 & 0.009 & 0.112 & 2.245 & 0.169 & 1.285 & 0.012 & 0.573 & 0.010 & 0.362 & 0.014 & 0.1892 & & + 124b & 14:13:58.89 & 00:23:39.08 & 18.956 & 0.031 & 0.112 & 1.226 & 0.221 & 0.702 & 0.041 & 0.396 & 0.042 & -0.141 & 0.148 & & & + 124c & 14:13:59.49 & 00:23:53.53 & 19.030 & 0.017 & 0.111 & 2.197 & 0.396 & 1.212 & 0.026 & 0.426 & 0.021 & 0.317 & 0.044 & & & + 124d & 14:14:00.17 & 00:23:52.94 & 19.552 & 0.020 & 0.111 & 1.067 & 0.110 & 0.694 & 0.025 & 0.401 & 0.026 & 0.043 & 0.070 & & & + & & & & & & & & & & & & & & & & + 125a & 14:14:06.22 & 00:21:36.96 & 17.929 & 0.010 & 0.112 & 1.021 & 0.037 & 0.541 & 0.012 & 0.354 & 0.012 & 0.165 & 0.027 & & & + 125b & 14:14:07.18 & 00:21:22.29 & 19.316 & 0.017 & 0.112 & 1.584 & 0.301 & 1.454 & 0.032 & 0.545 & 0.021 & 0.317 & 0.041 & & & + 125c & 14:14:06.32 & 00:21:24.36 & 19.647 & 0.023 & 0.112 & 2.660 & 1.079 & 1.603 & 0.051 & 0.665 & 0.028 & 0.373 & 0.050 & & & + 125d & 14:14:06.52 & 00:21:36.05 & 19.806 & 0.027 & 0.112 & 3.238 & 1.371 & 1.690 & 0.069 & 0.674 & 0.034 & 0.399 & 0.059 & & & + & & & & & & & & & & & & & & & & + 126a & 14:15:03.15 & -00:22:42.89 & 17.233 & 0.008 & 0.117 & 2.120 & 0.115 & 1.044 & 0.010 & 0.432 & 0.009 & 0.401 & 0.011 & 0.1428 & & + 126b & 14:15:03.62 & -00:22:31.57 & 18.080 & 0.011 & 0.117 & 2.048 & 0.195 & 1.068 & 0.016 & 0.424 & 0.012 & 0.389 & 0.020 & & & + 126c & 14:15:04.44 & -00:22:20.89 & 18.522 & 0.018 & 0.117 & 1.806 & 0.287 & 0.871 & 0.029 & 0.400 & 0.024 & 0.358 & 0.047 & & & + 126d & 14:15:04.90 & -00:22:40.98 & 19.564 & 0.027 & 0.116 & 1.998 & 0.520 & 0.862 & 0.044 & 0.289 & 0.038 & 0.442 & 0.076 & & & + & & & & & & & & & & & & & & & & + 127a & 14:15:55.07 & 00:16:06.24 & 17.399 & 0.008 & 0.109 & 1.796 & 0.074 & 1.108 & 0.010 & 0.406 & 0.009 & 0.378 & 0.013 & 0.1260 & & + 127b & 14:15:54.09 & 00:15:50.18 & 18.374 & 0.010 & 0.109 & 2.134 & 0.162 & 1.088 & 0.014 & 0.444 & 0.012 & 0.421 & 0.019 & & & + 127c & 14:15:53.32 & 00:16:04.37 & 20.228 & 0.042 & 0.110 & 1.763 & 0.640 & 0.944 & 0.067 & 0.367 & 0.058 & 0.262 & 0.155 & & & + 127d & 14:15:55.04 & 00:16:13.06 & 20.341 & 0.042 & 0.109 & 1.741 & 0.544 & 0.797 & 0.061 & 0.549 & 0.053 & 0.439 & 0.111 & & & + & & & & & & & & & & & & & & & & + 128a & 14:22:03.76 & -00:57:25.65 & 15.924 & 0.007 & 0.129 & 1.874 & 0.041 & 0.996 & 0.008 & 0.448 & 0.007 & 0.342 & 0.006 & 0.1030 & & + 128b & 14:22:02.50 & -00:57:17.01 & 16.378 & 0.008 & 0.129 & 1.727 & 0.055 & 1.010 & 0.010 & 0.438 & 0.008 & 0.358 & 0.009 & & & + 128c & 14:22:02.64 & -00:57:57.00 & 16.655 & 0.008 & 0.128 & 1.581 & 0.037 & 0.840 & 0.009 & 0.498 & 0.009 & 0.319 & 0.008 & & & + 128d & 14:22:01.97 & -00:58:16.51 & 18.799 & 0.017 & 0.128 & 1.774 & 0.227 & 0.981 & 0.028 & 0.275 & 0.023 & 0.244 & 0.051 & & & + & & & & & & & & & & & & & & & & + 129a & 14:24:16.53 & -00:08:57.80 & 15.825 & 0.009 & 0.102 & 1.030 & 0.037 & 0.833 & 0.011 & 0.174 & 0.011 & 0.197 & 0.023 & 0.1736 & 0.1732 & 2dfgrsn344z15 + 129b & 14:24:18.75 & -00:08:58.27 & 17.324 & 0.008 & 0.101 & 2.000 & 0.081 & 1.093 & 0.010 & 0.445 & 0.009 & 0.375 & 0.010 & & & + 129c & 14:24:18.18 & -00:08:45.06 & 18.584 & 0.021 & 0.101 & 2.108 & 0.728 & 1.658 & 0.051 & 0.669 & 0.025 & 0.382 & 0.041 & & & + 129d & 14:24:14.98 & -00:09:03.08 & 18.701 & 0.019 & 0.102 & 1.056 & 0.094 & 0.618 & 0.025 & 0.287 & 0.026 & 0.051 & 0.068 & & & + & & & & & & & & & & & & & & & & + 130a & 14:25:15.67 & -00:29:58.32 & 17.292 & 0.008 & 0.133 & 1.665 & 0.051 & 0.919 & 0.009 & 0.452 & 0.008 & 0.379 & 0.009 & 0.1080 & & + 130b & 14:25:16.18 & -00:29:34.67 & 19.674 & 0.023 & 0.132 & 2.040 & 0.484 & 1.056 & 0.037 & 0.478 & 0.030 & 0.325 & 0.055 & & & + 130c & 14:25:15.53 & -00:29:49.67 & 19.946 & 0.029 & 0.132 & 0.558 & 0.139 & 0.887 & 0.043 & 0.352 & 0.039 & 0.338 & 0.076 & & & + 130d & 14:25:15.41 & -00:29:45.93 & 20.059 & 0.026 & 0.132 & 0.521 & 0.113 & 0.784 & 0.037 & 0.238 & 0.038 & 0.623 & 0.062 & & & + & & & & & & & & & & & & & & & & + 131a & 14:25:27.65 & 00:20:16.10 & 16.267 & 0.006 & 0.088 & 2.022 & 0.051 & 1.127 & 0.007 & 0.437 & 0.007 & 0.397 & 0.007 & 0.1343 & & + 131b & 14:25:27.45 & 00:20:19.89 & 17.479 & 0.008 & 0.088 & 1.972 & 0.088 & 1.041 & 0.010 & 0.355 & 0.009 & -0.018 & 0.022 & & & + 131c & 14:25:29.62 & 00:20:47.59 & 18.488 & 0.011 & 0.088 & 2.077 & 0.162 & 1.043 & 0.015 & 0.455 & 0.013 & 0.348 & 0.024 & & & + 131d & 14:25:28.16 & 00:20:34.37 & 19.143 & 0.452 & 0.088 & 1.735 & 2.239 & 1.051 & 0.490 & -0.140 & 0.499 & 0.437 & 0.747 & & & + & & & & & & & & & & & & & & & & + 132a & 14:25:34.36 & 00:19:56.93 & 15.817 & 0.006 & 0.089 & 1.968 & 0.034 & 0.943 & 0.007 & 0.395 & 0.006 & 0.365 & 0.006 & 0.0845 & 0.0843 & 2dfgrsn344z098 + 132b & 14:25:33.09 & 00:19:58.98 & 18.000 & 0.009 & 0.089 & 1.883 & 0.096 & 0.930 & 0.012 & 0.402 & 0.011 & 0.309 & 0.020 & & & + 132c & 14:25:33.14 & 00:19:17.20 & 18.116 & 0.010 & 0.089 & 2.106 & 0.175 & 1.065 & 0.014 & 0.404 & 0.012 & 0.387 & 0.025 & & & + 132d & 14:25:33.48 & 00:19:51.44 & 18.600 & 0.013 & 0.089 & 1.715 & 0.135 & 0.828 & 0.017 & 0.439 & 0.016 & 0.340 & 0.034 & & & + & & & & & & & & & & & & & & & & + 133a & 14:29:32.63 & 00:29:37.72 & 16.660 & 0.009 & 0.102 & 1.722 & 0.067 & 0.932 & 0.010 & 0.561 & 0.009 & 0.451 & 0.008 & 0.0554 & & + 133b & 14:29:33.95 & 00:29:33.88 & 17.559 & 0.019 & 0.102 & 0.801 & 0.072 & 0.449 & 0.023 & 0.204 & 0.027 & 0.185 & 0.058 & & & + 133c & 14:29:34.06 & 00:29:44.19 & 18.785 & 0.014 & 0.102 & 1.854 & 0.219 & 0.922 & 0.020 & 0.441 & 0.018 & 0.394 & 0.027 & & & + 133d & 14:29:32.40 & 00:30:07.83 & 19.085 & 0.038 & 0.102 & 0.747 & 0.151 & 0.416 & 0.047 & 0.230 & 0.056 & 0.182 & 0.122 & & & + & & & & & & & & & & & & & & & & + 134a & 14:29:40.63 & 00:21:59.04 & 14.360 & 0.005 & 0.105 & 2.131 & 0.010 & 0.899 & 0.005 & 0.440 & 0.005 & 0.376 & 0.002 & 0.0556 & & + 134b & 14:29:39.14 & 00:22:39.90 & 16.378 & 0.006 & 0.105 & 1.947 & 0.027 & 0.868 & 0.006 & 0.440 & 0.006 & 0.340 & 0.005 & & & + 134c & 14:29:39.54 & 00:22:03.56 & 17.020 & 0.010 & 0.105 & 1.677 & 0.080 & 0.875 & 0.013 & 0.349 & 0.012 & 0.196 & 0.025 & & & + 134d & 14:29:41.83 & 00:22:08.51 & 17.253 & 0.008 & 0.105 & 1.830 & 0.060 & 0.827 & 0.009 & 0.432 & 0.009 & 0.324 & 0.014 & 0.0537 & & + 134e & 14:29:36.96 & 00:22:07.17 & 17.253 & 0.009 & 0.105 & 2.031 & 0.120 & 0.849 & 0.012 & 0.366 & 0.011 & 0.363 & 0.023 & & & + & & & & & & & & & & & & & & & & + 135a & 14:30:57.38 & 00:47:17.77 & 17.623 & 0.009 & 0.104 & 1.487 & 0.063 & 0.968 & 0.011 & 0.496 & 0.010 & 0.314 & 0.014 & & & + 135b & 14:30:58.09 & 00:47:30.52 & 18.561 & 0.012 & 0.104 & 1.845 & 0.181 & 1.202 & 0.018 & 0.445 & 0.015 & 0.331 & 0.024 & & & + 135c & 14:30:57.84 & 00:47:31.23 & 19.497 & 0.032 & 0.104 & 1.086 & 0.502 & 1.696 & 0.085 & 0.744 & 0.039 & 0.194 & 0.073 & & & + 135d & 14:30:58.18 & 00:47:26.13 & 19.558 & 0.025 & 0.104 & 0.893 & 0.115 & 0.636 & 0.032 & 0.573 & 0.032 & 0.278 & 0.056 & & & + & & & & & & & & & & & & & & & & + 136a & 14:32:25.98 & 00:03:27.21 & 16.323 & 0.006 & 0.113 & 2.108 & 0.042 & 1.023 & 0.007 & 0.444 & 0.006 & 0.358 & 0.005 & 0.0929 & 0.0930 & 2dfgrsn346z007 + 136b & 14:32:26.40 & 00:02:51.23 & 17.037 & 0.007 & 0.114 & 1.671 & 0.068 & 1.267 & 0.009 & 0.436 & 0.008 & 0.346 & 0.011 & & & + 136c & 14:32:27.68 & 00:03:11.16 & 17.767 & 0.009 & 0.114 & 2.298 & 0.108 & 1.067 & 0.010 & 0.388 & 0.009 & 0.300 & 0.012 & & & + 136d & 14:32:26.86 & 00:02:52.00 & 17.963 & 0.011 & 0.114 & 1.775 & 0.130 & 1.032 & 0.015 & 0.475 & 0.013 & 0.022 & 0.029 & & & + 136e & 14:32:25.81 & 00:03:14.77 & 18.588 & 0.015 & 0.113 & 1.210 & 0.077 & 0.497 & 0.018 & 0.284 & 0.019 & 0.111 & 0.054 & & & + & & & & & & & & & & & & & & & & + 137a & 14:33:27.16 & -00:07:08.17 & 16.696 & 0.014 & 0.108 & 1.132 & 0.053 & 1.195 & 0.016 & 0.122 & 0.015 & -0.273 & 0.035 & & 0.0344&apmuks(bj ) b143053.50 +
000603.0 + 137b & 14:33:27.19 & -00:07:09.26 & 17.952 & 0.046 & 0.108 & 1.270 & 0.044 & -1.243 & 0.047 & 0.799 & 0.052 & 0.851 & 0.049 & & & + 137c & 14:33:27.91 & -00:06:53.03 & 19.234 & 0.063 & 0.107 & 0.669 & 0.120 & -0.166 & 0.069 & 0.202 & 0.096 & 0.072 & 0.275 & & & + 137d & 14:33:27.87 & -00:07:14.01 & 19.260 & 0.045 & 0.108 & -0.086 & 0.643 & 2.666 & 0.286 & 0.098 & 0.075 & 1.367 & 0.088 & & & + & & & & & & & & & & & & & & & & + 138a & 14:33:57.20 & 00:35:10.74 & 17.008 & 0.009 & 0.103 & 2.116 & 0.136 & 1.427 & 0.012 & 0.555 & 0.010 & 0.410 & 0.008 & 0.2221 & & + 138b & 14:33:56.34 & 00:34:49.44 & 19.642 & 0.022 & 0.102 & 1.890 & 0.537 & 1.376 & 0.043 & 0.489 & 0.029 & 0.286 & 0.049 & & & + 138c & 14:33:57.25 & 00:34:59.08 & 19.700 & 0.024 & 0.102 & 1.547 & 0.401 & 1.332 & 0.043 & 0.531 & 0.030 & 0.425 & 0.044 & & & + 138d & 14:33:56.87 & 00:35:09.48 & 19.786 & 0.025 & 0.103 & 1.279 & 0.315 & 1.312 & 0.045 & 0.643 & 0.030 & -1.514 & 0.226 & & & + & & & & & & & & & & & & & & & & + 139a & 14:35:47.68 & -00:10:06.25 & 17.356 & 0.014 & 0.114 & 1.290 & 0.059 & 0.676 & 0.016 & 0.274 & 0.016 & 0.288 & 0.030 & & 0.1066 & 2dfgrsn347z210 + 139b & 14:35:48.03 & -00:09:55.91 & 17.442 & 0.008 & 0.114 & 1.965 & 0.062 & 0.983 & 0.009 & 0.462 & 0.008 & 0.421 & 0.009 & 0.0925 & 0.0930 & 2dfgrsn346z169 + 139c & 14:35:47.25 & -00:09:45.56 & 18.989 & 0.016 & 0.114 & 2.157 & 0.397 & 1.373 & 0.031 & 0.537 & 0.019 & 0.399 & 0.034 & & & + 139d & 14:35:47.79 & -00:10:18.11 & 20.248 & 0.036 & 0.114 & 1.575 & 0.485 & 1.122 & 0.066 & 0.691 & 0.043 & 0.385 & 0.080 & & & + & & & & & & & & & & & & & & & & + 140a & 14:37:03.25 & -00:36:02.64 & 17.930 & 0.011 & 0.103 & 2.025 & 0.212 & 1.319 & 0.017 & 0.572 & 0.012 & 0.337 & 0.017 & & 0.2137 & 2dfgrsn280z122 + 140b & 14:37:03.03 & -00:35:59.83 & 19.003 & 0.018 & 0.103 & 1.271 & 0.174 & 0.967 & 0.028 & 0.402 & 0.024 & 0.282 & 0.047 & & & + 140c & 14:37:03.67 & -00:35:51.02 & 19.232 & 0.019 & 0.103 & 1.630 & 0.275 & 1.042 & 0.029 & 0.581 & 0.022 & 0.303 & 0.040 & & & + 140d & 14:37:03.49 & -00:35:42.99 & 19.340 & 0.020 & 0.103 & 2.071 & 0.554 & 1.280 & 0.037 & 0.562 & 0.025 & 0.347 & 0.044 & & & + & & & & & & & & & & & & & & & & + 141a & 14:37:35.62 & -00:33:24.20 & 16.099 & 0.008 & 0.098 & 2.180 & 0.094 & 1.231 & 0.009 & 0.556 & 0.008 & 0.405 & 0.006 & 0.1799 & 0.1797 & 2masx j14373563 - 0033239 + 141b & 14:37:34.42 & -00:33:02.09 & 17.467 & 0.012 & 0.098 & 2.808 & 0.551 & 1.137 & 0.020 & 0.511 & 0.015 & 0.327 & 0.026 & & & + 141c & 14:37:34.68 & -00:33:40.80 & 18.637 & 0.013 & 0.098 & 1.958 & 0.259 & 1.197 & 0.019 & 0.525 & 0.015 & 0.375 & 0.022 & & & + 141d & 14:37:33.91 & -00:33:24.66 & 19.056 & 0.015 & 0.098 & 2.010 & 0.348 & 1.179 & 0.024 & 0.531 & 0.018 & 0.393 & 0.028 & & & + & & & & & & & & & & & & & & & & + 142a & 14:38:03.69 & -00:19:18.65 & 17.505 & 0.011 & 0.107 & 2.167 & 0.197 & 1.439 & 0.016 & 0.556 & 0.012 & 0.395 & 0.013 & 0.2470 & & + 142b & 14:38:03.32 & -00:19:30.73 & 19.612 & 0.020 & 0.106 & 2.560 & 0.711 & 1.409 & 0.037 & 0.520 & 0.023 & 0.411 & 0.036 & & & + 142c & 14:38:04.29 & -00:19:29.88 & 19.883 & 0.029 & 0.107 & 1.014 & 0.168 & 0.724 & 0.040 & 0.207 & 0.040 & 0.232 & 0.097 & & & + 142d & 14:38:04.26 & -00:19:35.75 & 20.413 & 0.041 & 0.107 & 1.119 & 0.430 & 1.191 & 0.077 & 0.365 & 0.055 & 0.225 & 0.129 & & & + & & & & & & & & & & & & & & & & + 143a & 14:41:19.04 & 00:58:12.77 & 17.866 & 0.010 & 0.122 & 2.035 & 0.236 & 1.465 & 0.017 & 0.515 & 0.012 & 0.424 & 0.015 & & & + 143b & 14:41:18.45 & 00:58:08.59 & 18.123 & 0.011 & 0.123 & 2.378 & 0.308 & 1.398 & 0.017 & 0.489 & 0.012 & 0.381 & 0.017 & & & + 143c & 14:41:19.81 & 00:58:13.46 & 18.918 & 0.020 & 0.122 & 1.745 & 0.395 & 1.312 & 0.036 & 0.496 & 0.024 & 0.309 & 0.043 & & & + 143d & 14:41:18.82 & 00:58:26.75 & 20.396 & 0.066 & 0.122 & 0.714 & 0.498 & 1.106 & 0.118 & 0.417 & 0.090 & 0.412 & 0.166 & & & + & & & & & & & & & & & & & & & & + 144a & 14:43:18.35 & -00:48:03.12 & 17.470 & 0.011 & 0.110 & 1.594 & 0.083 & 0.990 & 0.013 & 0.478 & 0.012 & 0.357 & 0.012 & 0.1466 & 0.1464 & 2dfgrsn282z141 + 144b & 14:43:17.37 & -00:48:12.92 & 18.125 & 0.013 & 0.110 & 2.245 & 0.223 & 1.018 & 0.017 & 0.419 & 0.014 & 0.323 & 0.019 & & & + 144c & 14:43:17.18 & -00:48:06.11 & 20.237 & 0.039 & 0.110 & 2.400 & 1.221 & 1.256 & 0.073 & 0.702 & 0.046 & 0.503 & 0.062 & & & + 144d & 14:43:17.23 & -00:48:10.98 & 20.379 & 0.038 & 0.110 & 0.837 & 0.199 & 0.593 & 0.053 & 0.431 & 0.050 & -0.208 & 0.140 & & & + & & & & & & & & & & & & & & & & + 145a & 14:44:35.86 & -01:08:18.14 & 16.948 & 0.009 & 0.123 & 2.108 & 0.089 & 1.119 & 0.012 & 0.515 & 0.010 & 0.380 & 0.008 & & & + 145b & 14:44:34.96 & -01:07:51.32 & 17.537 & 0.019 & 0.123 & 1.395 & 0.155 & 1.078 & 0.032 & 0.499 & 0.023 & 0.443 & 0.033 & & & + 145c & 14:44:36.55 & -01:07:50.70 & 18.107 & 0.011 & 0.123 & 2.026 & 0.134 & 1.086 & 0.016 & 0.498 & 0.012 & 0.356 & 0.014 & & & + 145d & 14:44:36.25 & -01:08:25.98 & 18.834 & 0.064 & 0.123 & 0.645 & 0.357 & 0.997 & 0.123 & 1.040 & 0.072 & 0.404 & 0.084 & & & + 145e & 14:44:35.89 & -01:07:58.77 & 19.208 & 0.019 & 0.123 & 1.660 & 0.247 & 1.043 & 0.035 & 0.489 & 0.024 & 0.383 & 0.037 & & & + 145f & 14:44:36.38 & -01:08:27.98 & 19.898 & 0.026 & 0.123 & 3.486 & 1.160 & 0.969 & 0.048 & 0.518 & 0.033 & 0.097 & 0.068 & & & + & & & & & & & & & & & & & & & & + 146a & 14:45:18.27 & 00:33:59.15 & 17.722 & 0.012 & 0.143 & 1.464 & 0.093 & 0.997 & 0.016 & 0.460 & 0.014 & 0.353 & 0.018 & & & + 146b & 14:45:19.49 & 00:34:16.20 & 19.157 & 0.013 & 0.143 & 1.747 & 0.132 & 0.845 & 0.017 & 0.239 & 0.017 & 0.209 & 0.032 & & & + 146c & 14:45:18.83 & 00:34:08.04 & 19.178 & 0.021 & 0.143 & 1.344 & 0.172 & 0.856 & 0.030 & 0.434 & 0.027 & 0.379 & 0.043 & & & + 146d & 14:45:19.22 & 00:33:51.34 & 20.298 & 0.045 & 0.142 & 2.361 & 0.949 & 0.863 & 0.070 & 0.442 & 0.060 & 0.437 & 0.099 & & & + 146e & 14:45:19.32 & 00:34:03.99 & 20.313 & 0.036 & 0.143 & 2.741 & 1.179 & 1.521 & 0.086 & 0.548 & 0.046 & 0.376 & 0.075 & & & + & & & & & & & & & & & & & & & & + 147a & 14:45:35.36 & -01:02:01.79 & 17.417 & 0.009 & 0.121 & 2.254 & 0.136 & 1.065 & 0.013 & 0.462 & 0.010 & 0.300 & 0.014 & & & + 147b & 14:45:36.76 & -01:02:16.34 & 18.820 & 0.023 & 0.121 & 1.613 & 0.269 & 0.977 & 0.039 & 0.478 & 0.030 & 0.405 & 0.050 & & & + 147c & 14:45:35.74 & -01:01:59.68 & 19.788 & 0.031 & 0.121 & 1.550 & 0.435 & 1.201 & 0.063 & 0.256 & 0.045 & 0.275 & 0.100 & & & + 147d & 14:45:36.61 & -01:02:05.37 & 20.192 & 0.152 & 0.121 & 0.387 & 0.656 & 0.952 & 0.272 & 0.432 & 0.206 & -0.014 & 0.529 & & & + & & & & & & & & & & & & & & & & + 148a & 14:47:16.50 & 00:55:24.15 & 16.743 & 0.008 & 0.142 & 2.127 & 0.102 & 1.072 & 0.010 & 0.486 & 0.009 & 0.425 & 0.009 & & 0.1373 & 2dfgrsn349z047 + 148b & 14:47:15.70 & 00:55:34.97 & 17.448 & 0.009 & 0.142 & 1.729 & 0.081 & 1.061 & 0.011 & 0.526 & 0.009 & 0.430 & 0.010 & 0.1373 & & + 148c & 14:47:13.93 & 00:55:23.10 & 18.213 & 0.012 & 0.142 & 2.264 & 0.241 & 1.070 & 0.016 & 0.476 & 0.013 & 0.412 & 0.019 & & & + 148d & 14:47:15.20 & 00:55:48.11 & 19.535 & 0.200 & 0.142 & 1.994 & 2.325 & 0.450 & 0.247 & 0.236 & 0.278 & -2.512 & 6.258 & & & + & & & & & & & & & & & & & & & & + 149a & 14:48:38.55 & 00:38:45.38 & 16.262 & 0.009 & 0.112 & 1.491 & 0.060 & 1.010 & 0.011 & 0.427 & 0.010 & 0.294 & 0.013 & 0.1405 & & + 149b & 14:48:35.87 & 00:38:54.66 & 18.846 & 0.014 & 0.112 & 1.205 & 0.187 & 1.589 & 0.030 & 0.589 & 0.017 & 0.347 & 0.026 & & & + 149c & 14:48:37.24 & 00:38:50.56 & 18.875 & 0.013 & 0.112 & 1.806 & 0.185 & 1.169 & 0.020 & 0.451 & 0.016 & 0.290 & 0.026 & & & + 149d & 14:48:38.78 & 00:38:53.71 & 18.920 & 0.015 & 0.112 & 1.873 & 0.200 & 1.034 & 0.021 & 0.399 & 0.018 & 0.346 & 0.030 & & & + & & & & & & & & & & & & & & & & + 150a & 14:53:11.13 & 00:58:23.39 & 16.498 & 0.012 & 0.145 & 1.377 & 0.075 & 0.847 & 0.015 & 0.327 & 0.014 & 0.304 & 0.021 & & & + 150b & 14:53:09.04 & 00:58:29.63 & 16.625 & 0.011 & 0.144 & 1.074 & 0.044 & 0.435 & 0.013 & 0.170 & 0.014 & 0.175 & 0.031 & 0.1193 & & + 150c & 14:53:09.85 & 00:58:38.57 & 16.878 & 0.010 & 0.144 & 1.423 & 0.056 & 0.920 & 0.012 & 0.509 & 0.011 & 0.357 & 0.012 & & & + 150d & 14:53:09.29 & 00:58:05.77 & 19.098 & 0.023 & 0.144 & 1.106 & 0.125 & 0.538 & 0.030 & -0.012 & 0.037 & 0.409 & 0.080 & & & + & & & & & & & & & & & & & & & & + 151a & 14:53:19.49 & -00:37:42.91 & 17.763 & 0.009 & 0.149 & 0.540 & 0.016 & 0.379 & 0.010 & 0.308 & 0.010 & 0.010 & 0.019 & 0.1857 & & + 151b & 14:53:19.00 & -00:37:29.74 & 19.348 & 0.022 & 0.149 & 1.249 & 0.250 & 1.210 & 0.039 & 0.446 & 0.028 & 0.434 & 0.047 & & & + 151c & 14:53:19.89 & -00:37:25.92 & 19.855 & 0.026 & 0.150 & 2.064 & 0.595 & 1.159 & 0.047 & 0.557 & 0.033 & 0.269 & 0.060 & & & + 151d & 14:53:20.20 & -00:37:22.50 & 20.681 & 0.052 & 0.150 & 2.067 & 1.108 & 1.160 & 0.099 & 0.354 & 0.074 & 0.575 & 0.125 & & & + & & & & & & & & & & & & & & & & + 152a & 14:54:08.05 & -00:10:57.09 & 17.333 & 0.016 & 0.143 & 1.352 & 0.074 & 0.454 & 0.019 & 0.269 & 0.020 & 0.078 & 0.052 & & & + 152b & 14:54:08.87 & -00:11:02.58 & 19.059 & 0.015 & 0.143 & 2.594 & 0.542 & 1.405 & 0.028 & 0.556 & 0.018 & 0.466 & 0.028 & & & + 152c & 14:54:08.38 & -00:11:03.70 & 19.693 & 0.027 & 0.143 & 1.481 & 0.228 & 0.717 & 0.038 & 0.456 & 0.035 & 0.199 & 0.083 & & & + 152d & 14:54:07.41 & -00:11:12.25 & 19.753 & 0.025 & 0.144 & 1.172 & 0.173 & 0.798 & 0.036 & 0.457 & 0.032 & 0.093 & 0.084 & & & + & & & & & & & & & & & & & & & & + 153a & 14:56:16.53 & 00:19:55.11 & 17.280 & 0.008 & 0.121 & 2.042 & 0.083 & 1.111 & 0.010 & 0.465 & 0.009 & 0.362 & 0.010 & 0.1393 & & + 153b & 14:56:16.36 & 00:19:55.86 & 18.047 & 0.012 & 0.121 & 2.426 & 0.308 & 1.265 & 0.017 & 0.466 & 0.014 & 0.325 & 0.025 & & & + 153c & 14:56:15.97 & 00:19:46.72 & 19.472 & 0.017 & 0.121 & 2.263 & 0.399 & 1.159 & 0.025 & 0.383 & 0.021 & 0.346 & 0.040 & & & + 153d & 14:56:15.61 & 00:19:50.36 & 19.752 & 0.027 & 0.121 & 1.155 & 0.139 & 0.365 & 0.032 & 0.243 & 0.038 & 0.021 & 0.125 & & & + 153e & 14:56:16.06 & 00:20:02.90 & 19.906 & 0.025 & 0.121 & 1.956 & 0.482 & 1.019 & 0.038 & 0.323 & 0.035 & 0.596 & 0.065 & & & + & & & & & & & & & & & & & & & & + 154a & 14:56:30.81 & 00:13:34.93 & 17.621 & 0.010 & 0.128 & 1.952 & 0.146 & 1.215 & 0.014 & 0.503 & 0.012 & 0.406 & 0.017 & 0.1849 & & + 154b & 14:56:30.99 & 00:13:42.36 & 19.217 & 0.016 & 0.129 & 1.521 & 0.196 & 1.143 & 0.024 & 0.571 & 0.019 & 0.510 & 0.029 & & & + 154c & 14:56:32.16 & 00:13:29.37 & 20.250 & 0.038 & 0.129 & 1.212 & 0.284 & 0.652 & 0.051 & 0.344 & 0.053 & 0.035 & 0.173 & & & + 154d & 14:56:32.28 & 00:13:35.04 & 20.534 & 0.069 & 0.129 & 0.882 & 0.410 & 0.706 & 0.097 & 0.129 & 0.110 & -0.300 & 0.538 & & & + & & & & & & & & & & & & & & & & + 155a & 14:57:38.10 & -00:04:37.89 & 17.684 & 0.017 & 0.154 & 1.162 & 0.043 & 0.266 & 0.018 & 0.172 & 0.021 & 0.001 & 0.050 & 0.0420 & & + 155b & 14:57:37.84 & -00:04:48.69 & 18.482 & 0.014 & 0.154 & 1.653 & 0.132 & 1.012 & 0.019 & 0.512 & 0.016 & 0.389 & 0.025 & & & + 155c & 14:57:37.35 & -00:04:38.80 & 18.791 & 0.014 & 0.154 & 2.562 & 0.443 & 1.413 & 0.024 & 1.028 & 0.015 & 0.573 & 0.014 & & & + 155d & 14:57:37.70 & -00:04:26.60 & 19.958 & 0.042 & 0.153 & 1.032 & 0.237 & 0.769 & 0.059 & 0.398 & 0.057 & 0.201 & 0.130 & & & + & & & & & & & & & & & & & & & & + 156a & 15:03:42.69 & -00:46:47.01 & 17.439 & 0.010 & 0.154 & 1.222 & 0.093 & 1.107 & 0.015 & 0.505 & 0.012 & 0.415 & 0.017 & & & + 156b & 15:03:43.35 & -00:46:36.62 & 18.555 & 0.015 & 0.154 & 1.706 & 0.245 & 1.403 & 0.024 & 0.514 & 0.017 & 0.381 & 0.021 & & & + 156c & 15:03:43.45 & -00:46:50.61 & 19.216 & 0.016 & 0.154 & 3.517 & 1.287 & 1.494 & 0.030 & 0.526 & 0.018 & 0.423 & 0.026 & & & + 156d & 15:03:43.15 & -00:46:29.49 & 19.902 & 0.032 & 0.154 & 1.968 & 0.783 & 1.208 & 0.058 & 0.466 & 0.040 & 0.405 & 0.068 & & & + & & & & & & & & & & & & & & & & + 157a & 15:07:37.54 & 00:02:24.29 & 17.441 & 0.012 & 0.147 & 2.019 & 0.193 & 1.543 & 0.017 & 0.542 & 0.013 & 0.356 & 0.017 & 0.2314 & & + 157b & 15:07:37.59 & 00:02:27.94 & 18.727 & 0.020 & 0.147 & 1.896 & 0.398 & 1.387 & 0.035 & 0.656 & 0.023 & 0.033 & 0.053 & & & + 157c & 15:07:37.85 & 00:02:22.00 & 19.530 & 0.022 & 0.147 & 1.872 & 0.516 & 1.617 & 0.045 & 0.469 & 0.026 & 0.315 & 0.053 & & & + 157d & 15:07:37.71 & 00:02:33.69 & 19.824 & 0.025 & 0.148 & 2.198 & 0.670 & 1.435 & 0.048 & 0.567 & 0.030 & 0.276 & 0.061 & & & + & & & & & & & & & & & & & & & & + 158a & 15:07:46.98 & -00:30:04.30 & 17.675 & 0.010 & 0.167 & 2.252 & 0.158 & 1.127 & 0.012 & 0.509 & 0.011 & 0.330 & 0.013 & 0.1525 & & + 158b & 15:07:47.00 & -00:30:08.36 & 18.821 & 0.015 & 0.167 & 1.831 & 0.221 & 1.114 & 0.022 & 0.455 & 0.018 & 0.330 & 0.028 & & & + 158c & 15:07:46.73 & -00:30:07.68 & 19.367 & 0.023 & 0.167 & 1.814 & 0.581 & 1.475 & 0.050 & 0.511 & 0.030 & 0.147 & 0.061 & & & + 158d & 15:07:47.87 & -00:30:01.37 & 19.997 & 0.030 & 0.168 & 2.072 & 0.622 & 0.926 & 0.046 & 0.423 & 0.040 & 0.256 & 0.083 & & & + & & & & & & & & & & & & & & & & + 159a & 15:08:30.67 & -00:14:49.33 & 16.724 & 0.009 & 0.156 & 1.921 & 0.046 & 1.005 & 0.010 & 0.497 & 0.010 & 0.410 & 0.006 & 0.0948 & & + 159b & 15:08:29.86 & -00:14:48.37 & 18.565 & 0.022 & 0.157 & 1.683 & 0.217 & 0.825 & 0.029 & 0.521 & 0.026 & 0.363 & 0.043 & & & + 159c & 15:08:30.90 & -00:14:29.61 & 18.887 & 0.019 & 0.156 & 1.300 & 0.231 & 1.368 & 0.035 & 0.651 & 0.022 & 0.496 & 0.031 & & & + 159d & 15:08:30.51 & -00:14:43.65 & 19.187 & 0.036 & 0.156 & 0.701 & 0.098 & 0.291 & 0.043 & 0.418 & 0.046 & -0.073 & 0.124 & & & + & & & & & & & & & & & & & & & & + 160a & 15:09:28.03 & 00:53:54.90 & 17.972 & 0.011 & 0.149 & 2.518 & 0.358 & 1.544 & 0.017 & 0.553 & 0.012 & 0.308 & 0.016 & & & + 160b & 15:09:27.89 & 00:53:49.54 & 18.953 & 0.016 & 0.149 & 2.656 & 0.677 & 1.483 & 0.028 & 0.532 & 0.019 & 0.334 & 0.030 & & & + 160c & 15:09:26.81 & 00:53:49.34 & 20.209 & 0.044 & 0.149 & 0.729 & 0.461 & 1.526 & 0.098 & 0.586 & 0.056 & 0.347 & 0.101 & & & + 160d & 15:09:27.11 & 00:53:57.44 & 20.805 & 0.045 & 0.149 & 1.023 & 0.541 & 1.485 & 0.095 & 0.528 & 0.058 & 0.475 & 0.098 & & & + & & & & & & & & & & & & & & & & + 161a & 15:09:37.81 & 00:48:25.34 & 15.373 & 0.006 & 0.149 & 1.983 & 0.016 & 0.905 & 0.006 & 0.478 & 0.006 & 0.331 & 0.002 & & & + 161b & 15:09:39.71 & 00:47:39.45 & 16.889 & 0.007 & 0.149 & 2.010 & 0.036 & 0.947 & 0.008 & 0.475 & 0.007 & 0.338 & 0.005 & 0.0725 & & + 161c & 15:09:38.35 & 00:48:24.19 & 16.935 & 0.007 & 0.149 & 1.980 & 0.039 & 0.907 & 0.008 & 0.470 & 0.008 & 0.291 & 0.007 & & & + 161d & 15:09:35.98 & 00:48:29.79 & 17.282 & 0.011 & 0.150 & 1.101 & 0.028 & 0.420 & 0.011 & 0.330 & 0.012 & 0.122 & 0.021 & 0.0943 & & + & & & & & & & & & & & & & & & & + 162a & 15:12:34.36 & 00:19:36.51 & 17.240 & 0.009 & 0.133 & 1.945 & 0.086 & 0.998 & 0.010 & 0.453 & 0.010 & 0.389 & 0.014 & & & + 162b & 15:12:35.03 & 00:19:41.76 & 17.461 & 0.009 & 0.133 & 2.012 & 0.085 & 0.963 & 0.010 & 0.484 & 0.010 & 0.312 & 0.014 & 0.1174 & & + 162c & 15:12:35.22 & 00:19:39.82 & 18.931 & 0.017 & 0.133 & 1.978 & 0.237 & 0.807 & 0.023 & 0.444 & 0.022 & 0.134 & 0.056 & & & + 162d & 15:12:33.01 & 00:19:41.60 & 20.104 & 0.048 & 0.133 & 0.692 & 0.285 & 0.951 & 0.075 & 0.407 & 0.066 & 0.087 & 0.208 & & & + 162e & 15:12:33.24 & 00:19:21.65 & 20.145 & 0.034 & 0.133 & 1.048 & 0.289 & 1.010 & 0.054 & 0.552 & 0.043 & 0.404 & 0.093 & & & + & & & & & & & & & & & & & & & & + 163a & 15:14:31.91 & -00:13:34.46 & 17.831 & 0.021 & 0.142 & 1.143 & 0.067 & 0.381 & 0.024 & 0.353 & 0.025 & 0.187 & 0.046 & & & + 163b & 15:14:31.80 & -00:13:37.43 & 19.304 & 0.051 & 0.142 & 2.602 & 3.261 & 2.023 & 0.192 & 0.596 & 0.063 & 0.422 & 0.110 & & & + 163c & 15:14:32.32 & -00:13:33.79 & 19.565 & 0.028 & 0.142 & 0.853 & 0.196 & 1.188 & 0.050 & 0.467 & 0.035 & 0.463 & 0.058 & & & + 163d & 15:14:31.77 & -00:13:43.52 & 19.790 & 0.023 & 0.142 & 1.921 & 0.442 & 1.362 & 0.043 & 0.395 & 0.028 & 0.191 & 0.057 & & & + 163e & 15:14:32.42 & -00:13:52.70 & 20.405 & 0.036 & 0.142 & 0.828 & 0.182 & 0.730 & 0.053 & 0.153 & 0.052 & 0.326 & 0.121 & & & + & & & & & & & & & & & & & & & & + 164a & 15:14:46.28 & 00:03:22.34 & 17.158 & 0.013 & 0.142 & 1.608 & 0.066 & 0.879 & 0.015 & 0.577 & 0.014 & 0.364 & 0.015 & 0.0716 & & + 164b & 15:14:45.20 & 00:03:04.64 & 19.438 & 0.028 & 0.142 & 1.566 & 0.479 & 1.344 & 0.056 & 0.723 & 0.033 & 0.434 & 0.057 & & & + 164c & 15:14:44.91 & 00:03:22.01 & 19.443 & 0.028 & 0.142 & 0.465 & 0.109 & 0.819 & 0.040 & 0.390 & 0.036 & 0.162 & 0.098 & & & + 164d & 15:14:45.37 & 00:03:01.08 & 19.895 & 0.037 & 0.142 & 1.260 & 0.338 & 0.950 & 0.057 & 0.495 & 0.046 & 0.458 & 0.094 & & & + & & & & & & & & & & & & & & & & + 165a & 15:14:47.32 & -00:09:01.89 & 17.138 & 0.010 & 0.146 & 1.494 & 0.040 & 0.589 & 0.011 & 0.403 & 0.011 & 0.227 & 0.016 & 0.1001 & & + 165b & 15:14:48.43 & -00:09:11.78 & 19.545 & 0.021 & 0.146 & 2.248 & 0.638 & 1.451 & 0.043 & 0.645 & 0.025 & 0.399 & 0.042 & & & + 165c & 15:14:47.21 & -00:08:58.10 & 19.561 & 0.039 & 0.146 & 0.058 & 0.101 & 0.653 & 0.058 & 0.413 & 0.054 & -0.321 & 0.199 & & & + 165d & 15:14:46.50 & -00:09:05.02 & 19.689 & 0.027 & 0.145 & 1.061 & 0.219 & 1.036 & 0.044 & 0.420 & 0.036 & 0.381 & 0.073 & & & + & & & & & & & & & & & & & & & & + 166a & 15:16:10.16 & 00:27:41.02 & 15.854 & 0.008 & 0.141 & 2.161 & 0.032 & 1.019 & 0.008 & 0.450 & 0.008 & 0.380 & 0.003 & & & + 166b & 15:16:09.06 & 00:28:15.05 & 16.727 & 0.009 & 0.141 & 2.340 & 0.084 & 1.127 & 0.010 & 0.519 & 0.009 & 0.433 & 0.006 & 0.1091 & & + 166c & 15:16:11.89 & 00:27:28.79 & 17.123 & 0.011 & 0.141 & 2.042 & 0.094 & 1.016 & 0.013 & 0.464 & 0.012 & 0.372 & 0.011 & 0.1061 & & + 166d & 15:16:09.28 & 00:27:46.66 & 18.202 & 0.013 & 0.141 & 2.058 & 0.157 & 0.942 & 0.017 & 0.444 & 0.015 & 0.234 & 0.023 & & & + 166e & 15:16:12.10 & 00:27:43.58 & 18.311 & 0.012 & 0.140 & 2.110 & 0.130 & 0.972 & 0.014 & 0.451 & 0.013 & 0.332 & 0.016 & & & + & & & & & & & & & & & & & & & & + 167a & 15:16:23.10 & -00:18:05.10 & 15.196 & 0.009 & 0.158 & 2.006 & 0.021 & 0.926 & 0.009 & 0.486 & 0.009 & 0.385 & 0.003 & 0.0509 & & + 167b & 15:16:23.33 & -00:16:51.78 & 16.124 & 0.011 & 0.158 & 1.666 & 0.055 & 0.991 & 0.012 & 0.452 & 0.012 & 0.302 & 0.011 & 0.1217 & & + 167c & 15:16:22.92 & -00:17:27.76 & 16.531 & 0.018 & 0.158 & 1.384 & 0.117 & 1.055 & 0.023 & 0.265 & 0.021 & 0.355 & 0.033 & & & + 167d & 15:16:21.67 & -00:18:14.00 & 17.981 & 0.016 & 0.158 & 1.830 & 0.148 & 0.754 & 0.020 & 0.360 & 0.019 & 0.297 & 0.034 & & & + & & & & & & & & & & & & & & & & + 168a & 15:16:23.48 & 00:06:09.12 & 16.782 & 0.010 & 0.153 & 1.942 & 0.067 & 1.046 & 0.011 & 0.485 & 0.011 & 0.421 & 0.009 & 0.1152 & & + 168b & 15:16:23.60 & 00:05:58.29 & 17.773 & 0.013 & 0.153 & 2.174 & 0.160 & 0.982 & 0.015 & 0.487 & 0.014 & 0.401 & 0.020 & & & + 168c & 15:16:24.15 & 00:06:12.39 & 19.543 & 0.025 & 0.153 & 1.175 & 0.213 & 0.972 & 0.038 & 0.432 & 0.032 & 0.334 & 0.073 & & & + 168d & 15:16:24.54 & 00:05:57.13 & 19.634 & 0.029 & 0.154 & 1.381 & 0.418 & 1.334 & 0.055 & 0.601 & 0.036 & 0.344 & 0.074 & & & + & & & & & & & & & & & & & & & & + 169a & 15:16:48.17 & -00:48:30.42 & 17.391 & 0.011 & 0.186 & 1.933 & 0.113 & 1.078 & 0.013 & 0.508 & 0.012 & 0.468 & 0.011 & 0.1191 & & + 169b & 15:16:49.37 & -00:48:37.04 & 18.042 & 0.015 & 0.186 & 1.341 & 0.082 & 0.760 & 0.017 & 0.434 & 0.016 & 0.310 & 0.023 & & & + 169c & 15:16:47.95 & -00:48:40.80 & 19.069 & 0.017 & 0.186 & 1.664 & 0.216 & 1.072 & 0.024 & 0.455 & 0.020 & 0.408 & 0.029 & & & + 169d & 15:16:48.03 & -00:48:22.76 & 19.324 & 0.023 & 0.186 & 1.341 & 0.191 & 0.760 & 0.030 & 0.417 & 0.028 & 0.487 & 0.044 & & & + & & & & & & & & & & & & & & & & + 170a & 15:17:22.36 & 00:00:00.52 & 17.476 & 0.011 & 0.166 & 1.832 & 0.077 & 0.998 & 0.012 & 0.440 & 0.011 & 0.364 & 0.013 & 0.1380 & & + 170b & 15:17:23.34 & 00:00:05.45 & 18.159 & 0.016 & 0.166 & 1.313 & 0.100 & 0.770 & 0.019 & 0.508 & 0.018 & 0.254 & 0.035 & & & + 170c & 15:17:23.49 & 00:00:12.10 & 18.571 & 0.021 & 0.166 & 0.898 & 0.113 & 0.809 & 0.028 & 0.323 & 0.027 & 0.264 & 0.067 & & & + 170d & 15:17:21.88 & 00:00:24.10 & 19.147 & 0.023 & 0.165 & 0.735 & 0.086 & 0.556 & 0.028 & 0.193 & 0.031 & -0.040 & 0.111 & & & + 170e & 15:17:21.61 & 00:00:13.78 & 20.125 & 0.044 & 0.165 & 1.499 & 0.491 & 0.797 & 0.064 & 0.214 & 0.063 & 0.708 & 0.128 & & & + & & & & & & & & & & & & & & & & + 171a & 15:22:42.19 & 00:41:30.15 & 17.348 & 0.009 & 0.157 & 2.165 & 0.084 & 0.766 & 0.010 & 0.359 & 0.010 & 0.235 & 0.014 & 0.0761 & & + 171b & 15:22:42.36 & 00:41:33.22 & 17.697 & 0.009 & 0.157 & 1.062 & 0.023 & 0.394 & 0.010 & 0.420 & 0.010 & 0.110 & 0.016 & & & + 171c & 15:22:42.70 & 00:41:06.04 & 18.776 & 0.018 & 0.157 & 0.447 & 0.071 & 0.872 & 0.026 & 0.486 & 0.023 & 0.416 & 0.040 & & & + 171d & 15:22:43.60 & 00:41:06.74 & 19.507 & 0.023 & 0.157 & 4.299 & 1.228 & 1.527 & 0.051 & 0.559 & 0.029 & 0.364 & 0.050 & & & + & & & & & & & & & & & & & & & & + 172a & 15:31:28.88 & -01:07:09.93 & 16.368 & 0.009 & 0.440 & 2.623 & 0.167 & 1.309 & 0.012 & 0.511 & 0.010 & 0.382 & 0.008 & & & + 172b & 15:31:28.43 & -01:07:10.82 & 18.093 & 0.011 & 0.440 & 2.070 & 0.186 & 1.329 & 0.019 & 0.526 & 0.013 & 0.313 & 0.016 & & & + 172c & 15:31:30.46 & -01:07:26.30 & 18.186 & 0.010 & 0.437 & 1.870 & 0.165 & 1.237 & 0.019 & 0.441 & 0.012 & 0.398 & 0.018 & & & + 172d & 15:31:27.63 & -01:07:40.34 & 18.890 & 0.016 & 0.439 & 2.059 & 0.259 & 1.136 & 0.026 & 0.506 & 0.018 & 0.384 & 0.026 & & & + 172e & 15:31:28.76 & -01:07:04.94 & 18.991 & 0.023 & 0.440 & 0.558 & 0.291 & 2.109 & 0.096 & 0.525 & 0.029 & 0.399 & 0.045 & & & + & & & & & & & & & & & & & & & & + 173a & 15:31:39.62 & 00:31:20.14 & 17.075 & 0.009 & 0.176 & 1.813 & 0.057 & 0.892 & 0.010 & 0.439 & 0.010 & 0.332 & 0.009 & 0.0800 & & + 173b & 15:31:38.72 & 00:31:15.55 & 17.183 & 0.009 & 0.177 & 2.073 & 0.076 & 0.957 & 0.010 & 0.462 & 0.010 & 0.345 & 0.009 & & & + 173c & 15:31:40.22 & 00:31:01.10 & 17.855 & 0.013 & 0.176 & 1.414 & 0.069 & 0.770 & 0.015 & 0.493 & 0.015 & 0.345 & 0.017 & & & + 173d & 15:31:40.19 & 00:31:24.43 & 18.894 & 0.044 & 0.175 & 0.691 & 0.434 & 1.520 & 0.106 & 0.718 & 0.054 & 0.412 & 0.081 & & & + & & & & & & & & & & & & & & & & + 174a & 15:33:03.01 & 00:17:01.31 & 15.109 & 0.006 & 0.189 & 1.988 & 0.020 & 0.962 & 0.006 & 0.497 & 0.006 & 0.377 & 0.003 & & & + 174b & 15:33:01.59 & 00:17:01.77 & 16.262 & 0.007 & 0.188 & 2.044 & 0.032 & 0.969 & 0.007 & 0.462 & 0.007 & 0.379 & 0.005 & & & + 174c & 15:33:00.92 & 00:17:02.81 & 17.521 & 0.008 & 0.188 & 1.665 & 0.044 & 0.831 & 0.009 & 0.380 & 0.009 & 0.363 & 0.011 & & & + 174d & 15:32:58.17 & 00:16:56.81 & 18.087 & 0.010 & 0.186 & 1.274 & 0.053 & 0.697 & 0.012 & 0.477 & 0.012 & 0.227 & 0.023 & & & + & & & & & & & & & & & & & & & & + 175a & 15:36:50.89 & 00:34:04.09 & 17.808 & 0.011 & 0.220 & 1.862 & 0.152 & 1.313 & 0.016 & 0.567 & 0.012 & 0.372 & 0.014 & & & + 175b & 15:36:50.12 & 00:34:20.30 & 18.558 & 0.013 & 0.220 & 1.946 & 0.222 & 1.332 & 0.021 & 0.492 & 0.015 & 0.440 & 0.020 & & & + 175c & 15:36:49.77 & 00:34:12.57 & 18.682 & 0.015 & 0.219 & 2.102 & 0.304 & 1.234 & 0.024 & 0.469 & 0.018 & 0.335 & 0.029 & & & + 175d & 15:36:50.70 & 00:34:05.24 & 19.856 & 0.036 & 0.220 & 1.527 & 0.506 & 1.190 & 0.067 & 0.612 & 0.045 & 0.083 & 0.094 & & & + 175e & 15:36:50.25 & 00:34:16.37 & 20.782 & 0.051 & 0.220 & 1.033 & 0.333 & 0.759 & 0.076 & 0.353 & 0.072 & -0.650 & 0.329 & & & + & & & & & & & & & & & & & & & & + & & & & & & & & & & & & & & & & + tri1a & 11:01:14.24 & + 00:22:48.88 & 17.638 & 0.013 & 0.079 & 0.542 & 0.024 & 0.441 & 0.014 & 0.356 & 0.015 & 0.136 & 0.033 & 0.214802 & 0.2148 & sdssqsoj110114.24 + 002248.9 + tri1b & 11:01:14.18 & + 00:22:38.65 & 19.063 & 0.017 & 0.079 & 1.049 & 0.074 & 0.580 & 0.019 & 0.357 & 0.021 & 0.189 & 0.051 & & & + tri1c & 11:01:13.59 & + 00:22:42.15 & 19.585 & 0.021 & 0.079 & 1.406 & 0.202 & 0.951 & 0.029 & 0.621 & 0.025 & 0.344 & 0.049 & & & + tri1d & 11:01:14.47 & + 00:22:43.86 & 20.634 & 0.052 & 0.079 & 0.562 & 0.170 & 0.390 & 0.064 & 0.354 & 0.074&-0.257 & 0.322 & & & + & & & & & & & & & & & & & & & & + tri2a & 13:10:35.46 & -00:14:03.19 & 17.118 & 0.010 & 0.069 & 1.987 & 0.077 & 0.982 & 0.012 & 0.515 & 0.010 & 0.429 & 0.009 & & & + tri2b & 13:10:36.45 & -00:14:25.87 & 19.245 & 0.015 & 0.069 & 0.205 & 0.025 & 0.320 & 0.017 & 0.082 & 0.019 & 0.454 & 0.034 & & 0.29 & 2qzj131036.4 - 001426 + tri2c & 13:10:36.23 & -00:14:05.83 & 19.873 & 0.053 & 0.070 & 1.461 & 0.283 & 0.458 & 0.063 & 0.257 & 0.067 & 0.302 & 0.134 & & & + tri2d & 13:10:35.90 & -00:14:30.94 & 19.901 & 0.028 & 0.069 & 3.180 & 1.326 & 1.349 & 0.060 & 0.567 & 0.034 & 0.437 & 0.057 & & & + & & & & & & & & & & & & & & & & + l r ll ll ll ll ll sdsscg & 175 & [email protected] & 4.00 & [email protected] & 23.51 & [email protected] & 0.126 & [email protected] & 0.0073 & 58@xmath632 & 55 + pcg & 84 & [email protected] & 4.00 & [email protected] & 23.32 & & @xmath164 & [email protected] & 0.0113 & & + scg & 121 & [email protected] & 4.00 & [email protected] & 25.68 & & & [email protected] & 0.125 & & + uzc cg & 291 & [email protected] & 3.00 & & & [email protected] & 0.0171 & & & 147@xmath633 & 152 + lccg & 76 & [email protected] & 3.00 & & & [email protected] & 0.075 & [email protected] & 0.013 & 53@xmath632 & 52 + rscg & 89 & [email protected] & 3.00 & & & [email protected] & 0.014 & [email protected] & 0.08 & 60@xmath633 & 55 + hcg & 92 & [email protected] & 4.00 & [email protected] & 23.4 & [email protected] & 0.030 & [email protected] & 0.048 & 84@xmath635 & 70 + lrrrr @xmath83 & [email protected] & 0.120 & [email protected] & 0.120 + @xmath165 & @[email protected] & @xmath16621.264 & @[email protected] & @xmath16621.293 + @xmath157 & [email protected] & 1.659 & [email protected] & 1.575 + @xmath154 & [email protected] & 0.756 & [email protected] & 0.730 + @xmath160 & [email protected] & 0.393 & [email protected] & 0.378 + @xmath161 & [email protected] & 0.272 & [email protected] & 0.268 + lcc @xmath83 vs. @xmath168 & 0.05316 & 0.90385 + & & + @xmath169 vs. @xmath170 & 0.19937 & 0.00054 + @xmath168 vs. @xmath170 & 0.17627 & 0.00296 + @xmath171 vs. @xmath170 & 0.14778 & 0.02209 + @xmath172 vs. @xmath170 & 0.07911 & 0.54591 + & & + @xmath168 vs. @xmath169 & 0.23006 & 0.00002 + @xmath171 vs. @xmath168 & 0.18354 & 0.00109 + @xmath172 vs. @xmath171 & 0.15791 & 0.00971 + | compact groups ( cgs ) of galaxies relatively poor groups of galaxies in which the typical separations between members is of the order of a galaxy diameter offer an exceptional laboratory for the study of dense galaxian environments with short ( @xmath0 1 gyr ) dynamical time - scales . in this paper , we present an objectively defined catalog of cgs in 153 sq deg of the sloan digital sky survey early data release ( sdss edr ) . to identify cgs , we applied a modified version of hickson s @xcite criteria aimed at finding the highest density cgs and thus reducing the number of chance alignments .
our catalog contains 175 cgs down to a limiting galaxy magnitude of @xmath1 .
the resulting catalog has a median depth of @xmath2 , substantially deeper than previous cg catalogs .
since the sdss will eventually image up to one quarter of the celestial sphere , we expect our final catalog , based upon the completed sdss , will contain on the order of 5,000 10,000 cgs
. this catalog will be useful for conducting studies of the general characteristics of cgs , their environments , and their component galaxies . |
the determination of the time - history of accretion is crucial to our understanding of how supermassive black holes form and evolve .
however , much of the accretion power in the universe is absorbed ( e.g. , almaini , lawrence , & boyle 1999 ) , making it difficult to measure at optical wavelengths .
the _ chandra _
( weisskopf et al . 2002 ) and _ xmm - newton _ ( jansen et al . 2001 ) _ x - ray observatories _ have revolutionized distant active galactic nucleus ( agn ) studies by making it possible to map the history of the agn population using hard ( @xmath10 kev ) x - ray surveys .
hard x - rays can directly probe agn activity , are uncontaminated by star formation processes at the x - ray luminosities of interest , and detect all but the most absorbed sources . thus , hard x - ray surveys provide as complete and unbiased a sample of agns as is presently possible ( e.g. , mushotzky 2004 ) , though they will still miss compton - thick sources . not surprisingly , hard x - ray luminosity functions constructed from _
chandra _ and _ xmm - newton _
samples have revealed that previous optically - selected and soft x - ray selected agn samples substantially undercount the agn population , at least at low and intermediate x - ray luminosities ( @xmath11 ergs s@xmath2 ; cowie et al .
2003 ; hasinger 2003 ; steffen et al . 2003
; fiore et al . 2003 ;
ueda et al . 2003 ) .
in fact , optically - selected broad - line agns comprise only about a third ( e.g. , barger et al .
2003b ) of the x - ray background ( giacconi et al .
1962 ) , and many x - ray sources show no signs of agn activity at all in their optical spectra ( e.g. , barger et al . 2001b ; tozzi et al . 2001 ; hornschemeier et al . 2001 ) .
theoretical models of supermassive black hole formation ( e.g. , haehnelt , natarajan , & rees 1998 ; kauffmann & haehnelt 2000 , 2002 ) have historically relied on comparisons with the optical quasar luminosity function , so these models now need to be reworked .
in addition , the hard x - ray luminosity functions have revealed that broad - line agns dominate the number densities at the higher x - ray luminosities , while non broad - line agns dominate at the lower x - ray luminosities ( steffen et al .
2003 ; we hereafter refer to this as the `` steffen effect '' ) . although we do not yet have a physical explanation for the steffen effect , there are two simple possibilities to consider .
one possibility is that the simple unified model for agns , where the differences between broad - line agns and non broad - line agns are only due to orientation effects , needs to be modified to include an x - ray luminosity dependent covering factor .
the second , more speculative possibility is that the steffen effect is a consequence of the broad - line agns being intrinsically different than the non broad - line agns .
we are now in a position to do a very thorough study of the nature and evolution of hard x - ray selected agns .
the primary aims of this paper are to explore the origin of the steffen effect and to determine the bolometric luminosities of all agns in order to understand the energy release history of supermassive black hole accretion .
the recent advances that make this comprehensive study possible are the high - resolution , multicolor observations that have been obtained with the acs camera on the _ hubble space telescope ( hst ) _ of fields with deep _
chandra _ x - ray data , and the extensive , high - quality optical spectroscopic follow - up observations that have been made of x - ray sources detected in both deep and wide - area _
chandra _ surveys .
the structure of the paper is as follows . in
[ secsamp ] , we describe the x - ray samples used in our analysis and the completeness of the spectroscopic and photometric redshift identifications .
we then determine the x - ray luminosities and median redshifts versus x - ray flux . in [ secclass ] , we spectrally classify the optical counterparts to the x - ray sources and investigate the dependence of the optical spectral types on x - ray obscuration . in
[ secnuc ] , we use multicolor _ hst _ acs great observatories origins deep survey ( goods - north ; giavalisco et al . 2004 )
observations of the cdf - n to separate the nuclear component of each source from the host galaxy light , and we then compare the ( nuclear galaxy ) colors with the optical spectral types . in [ sechxlf][sechizhxlf ] , we construct up - to - date low and high - redshift hard x - ray luminosity functions ( all spectral types and broad - line agns alone ) for our highly spectroscopically complete samples , do maximum likelihood fits over the redshift range @xmath12 , and examine the evolution of the hard x - ray luminosity functions . in [ seclf ] , we directly compare the broad - line agn hard x - ray luminosity functions with the optical qso luminosity functions . in
[ secenergy ] , we determine the evolution of the rest - frame hard x - ray comoving energy density production rate for all spectral types together as well as separated by spectral type . in [ secsmbh ] , we estimate black hole masses for a small sample of broad - line agns using the measured mgii 2800 line widths and nuclear optical luminosities .
we then explore `` mass starvation '' versus `` agn downsizing '' as explanations for the observed rapid decline in the energy density production rates between @xmath13 and @xmath14 .
we show in [ secfailure ] that the simple unified model can not explain the steffen effect . in [ seclum ]
, we postulate that the simplest interpretation of the steffen effect is a luminosity dependent unified model , although another interpretation might be intrinsic differences in the source populations . in
[ sechow ] , we discuss how the bolometric corrections to go from x - ray luminosities to bolometric luminosities can be determined . to infer what these corrections are ,
we need long - wavelength data . in [ seclongwave ] , we use mid - infrared ( mir ) and far - infrared ( fir)/submillimeter data obtained with the isophot and isocam instruments on the _ infrared space observatory ( iso ) _ and the scuba bolometer array on the james clerk maxwell telescope ( jcmt ) to observe directly any enhancements at these wavelengths due to absorption and reradiation by gas and dust .
we note that the observational situation in the mir / fir may be expected to improve with the _ spitzer space telescope_. since we do not know for sure what the origin of the steffen effect is , in [ secspectype ] and [ sechxlum ] , we cover both possibilities ( intrinsic differences in the source populations or a luminosity dependent unified model ) by constructing composite spectral energy distributions of the sources first by optical spectral type and then by x - ray luminosity . in [ secbolcorrtype ] and [ secbolcorrlum ] , we infer the bolometric corrections by spectral type and by x - ray luminosity . in [ secacc ] , we use our bolometric corrections and the rest - frame hard x - ray comoving energy density production rate to determine the accretion history of the universe .
we summarize our results in [ secsummary ] .
we assume @xmath15 , @xmath16 , and @xmath17 km s@xmath2 mpc@xmath2 .
all magnitudes are in the ab magnitude system .
a detailed study of the cosmic evolution of the hard x - ray source population requires the complementarity of deep and wide - field x - ray surveys .
fortunately , both types of x - ray surveys are now becoming available . since the _ chandra _ angular resolution is critical to the identification of the optical counterparts to the x - ray sources , in this paper we only consider _ chandra _ surveys ( except at the very brightest x - ray fluxes , where we supplement our data with an _ asca _ sample ) . to make sure our analysis is as robust as possible , we include only the three most spectroscopically complete _ chandra _ surveys available .
the two deep x - ray surveys that we use are the _ chandra _ deep field - north ( cdf - n ) and the _ chandra _ deep field - south ( cdf - s ) , the deepest x - ray images ever taken .
the @xmath18 ms cdf - n survey samples a large , distant cosmological volume down to very faint x - ray flux limits of @xmath19 ergs @xmath20 s@xmath2 and @xmath21 ergs @xmath20 s@xmath2 ( alexander et al .
2003b ) , while the @xmath22 ms cdf - s survey samples only a factor of two shallower ( giacconi et al .
2002 ) .
we supplement these deep surveys with a wide - area , intermediate depth survey in order to sample a large , low - redshift cosmological volume and to detect the rare , high - luminosity population .
the low - redshift volume enables us to probe robustly the evolution of agns between @xmath23 and @xmath0 . the only wide - area , intermediate depth survey published to date that has a high level of spectroscopic completeness is the _ chandra _ large - area synoptic x - ray survey , or clasxs .
this survey covers an @xmath24 deg@xmath25 region in the lockman hole - northwest , imaged to x - ray flux limits of @xmath26 ergs @xmath20 s@xmath2 and @xmath27 ergs @xmath20 s@xmath2 ( yang et al .
2004 ) .
the high spatial resolution of the _ chandra _ x - ray images allows a generally unambiguous optical counterpart to be found for the great majority of x - ray sources having counterparts @xmath28 .
( note that about 20% of the optical counterpart identifications in the @xmath29 range will be spurious ; see [ secphotz ] . ) only a few have multiple counterparts .
most of the @xmath30 sources , and many of the @xmath31 sources , can be spectroscopically identified ( barger et al .
2003b ; szokoly et al .
2004 ; steffen et al .
2004 ) . in table
[ tab1 ] , we show for the three _ chandra _ fields in our sample ( the most intensively spectroscopically observed _ chandra _ fields to date ) the total number of sources in the x - ray catalogs , the number of such sources observed spectroscopically , and the number of such sources identified spectroscopically .
if we want to explore the highest x - ray luminosities , we also need much wider - field , higher - flux samples than these _ chandra _ data can provide .
_ asca _ , _ bepposax _ , and _ rxte _ provide such samples , but with their low spatial resolutions , cross - identifications to the optical counterparts are more difficult and ambiguous .
ultimately , many of the sources in these surveys will be pinned down with higher resolution _
chandra _ or _ xmm - newton _ observations and their counterparts robustly identified . fortunately , however , since these high x - ray flux sources generally do have brighter optical counterparts than the sources in lower x - ray flux surveys , in many cases it is already possible to identify the counterparts , even with the positional uncertainties . in particular , the _ asca _ sample of akiyama et al .
( 2003 ) have nearly complete identifications ( see table [ tab1 ] ) , albeit with some ambiguous cases , while roughly 70% of the bright _ rxte _ sample have identifications ( sazonov & revnivtsev 2004 ) .
hereafter , the cdf - n , cdf - s , clasxs , and akiyama et al .
( 2003 ) _ asca _ hard x - ray
selected samples will constitute this paper s `` total hard x - ray sample '' , and the cdf - n , cdf - s , and clasxs soft x - ray
selected samples will constitute this paper s `` total soft x - ray sample '' .
lrrrr total & 503 & 346 & 525 & 32 observed & 451 & 247 & 467 & 32 identified & 306 & 137 & 272 & 31 broad - line & 43 & 32 & 106 & 30 high - excitation & 39 & 23 & 45 & 0 star formers & 148 & 55 & 73 & 0 absorbers & 58 & 20 & 28 & 0 stars & 14 & 7 & 20 & 1 [ tab1 ] figure [ figsamp ] shows the useful flux ranges of the three _ chandra _ surveys and the _ asca _ survey ( which is only in the hard x - ray sample ) that make up this paper s total hard and soft x - ray samples .
the surveys provide good coverage over the flux ranges @xmath32 ergs @xmath20 s@xmath2 and @xmath33 ergs @xmath20 s@xmath2 .
note that for all of the _ chandra _ surveys , the observed x - ray counts were converted to @xmath10 kev fluxes in the original papers using individual power - law indices determined from the ratios of the hard - to - soft x - ray counts ( i.e. , the hardness ratios ) .
for the _ asca _ survey , akiyama et al .
( 2003 ) assumed an intrinsic @xmath34 photon index to compute @xmath35 kev fluxes , so we have converted their @xmath35 kev fluxes to @xmath10 kev assuming their @xmath34 .
figure [ figfract ] shows how the contributions to the @xmath10 kev total resolved x - ray background , obtained by extrapolating the number counts to faint and bright x - ray fluxes , are strongly peaked around @xmath36 ergs @xmath20 s@xmath2 .
in fact , nearly 75% arises in the @xmath32 ergs @xmath20 s@xmath2 range ( e.g. , campana et al . 2001 ; cowie et al . 2002 ; rosati et al .
2002 ; alexander et al . 2003b ) .
approximately 70% of the @xmath10 kev light in the flux range @xmath32 ergs @xmath20 s@xmath2 is spectroscopically identified , while about 80% of the @xmath37 kev light in the flux range @xmath33 ergs @xmath20 s@xmath2 is spectroscopically identified .
figure [ figsamp]a also shows that the fraction of observed sources that can be spectroscopically identified is a strong function of the hard x - ray flux . above @xmath36 ergs @xmath20 s@xmath2 ,
@xmath38% of the sources have spectroscopic redshifts , while below this , the fraction drops to about 60% . while this is partly a consequence of the fainter x - ray sources being optically fainter , it is also a consequence of the fraction of broad - line agns being much higher at the brighter x - ray luminosities .
the hard x - ray flux limit of @xmath36 ergs @xmath20 s@xmath2 , above which the redshift identifications are very complete ( @xmath38% ) , corresponds to a rest - frame @xmath10 kev luminosity of @xmath39 ergs s@xmath2 at @xmath13 , while the soft x - ray flux limit of @xmath40 ergs @xmath20 s@xmath2 , above which the surveys are about 80% spectroscopically complete , corresponds to a rest - frame @xmath10 kev luminosity of @xmath41 ergs s@xmath2 at @xmath42 .
thus , nearly all of the high x - ray luminosity sources have been identified , and incompleteness has very little effect on the bright end of the hard x - ray luminosity function determinations ( see [ sechxlf ] ) . in figure [ figbroad]a , we show @xmath10 kev flux versus redshift for the spectroscopically observed sources in the total hard x - ray sample .
the clasxs sources nicely fill in the almost an order of magnitude flux gap between the cdf - n / cdf - s and _ asca _ samples .
the two solid curves correspond to loci of constant rest - frame @xmath10 kev luminosity , @xmath43 .
any source more luminous than @xmath44 ergs s@xmath2 _ ( lower curve ) _ is very likely to be an agn on energetic grounds ( zezas , georgantopoulos , & ward 1998 ; moran , lehnert , & helfand 1999 ) , though many of the intermediate luminosity sources do not show obvious agn signatures in their optical spectra .
sources with @xmath45 ergs s@xmath2 _ ( upper curve ) _ are often called quasars , since this x - ray luminosity roughly corresponds to the absolute optical magnitude of @xmath46 that is the traditional dividing line between quasars and seyfert 1 galaxies .
however , the distinction between seyfert 1 galaxies and quasars seems to have little physical significance . in this paper
, we simply refer to all objects which have lines with full width half maximum ( fwhm ) greater than 2000 km s@xmath2 in their optical spectra as broad - line agns ( see [ secclass ] ) .
the broad - line agns are denoted by large symbols in figure [ figbroad]a . from the figure
, we can see that the great majority of the most luminous sources , where we are very spectroscopically complete , are broad - line agns . in figure [ figbroad]b , we show rest - frame @xmath10 kev luminosity versus redshift for the broad - line agns alone . at @xmath47 ,
the luminosities were calculated from the observed - frame @xmath10 kev fluxes _ ( squares ) _ , and at @xmath48 , the luminosities were calculated from the observed - frame @xmath37 kev fluxes _ ( diamonds)_. one advantage of using the observed - frame soft x - ray fluxes at high redshifts is the increased sensitivity , since the @xmath37 kev _ chandra _ images are deeper than the @xmath10 kev images . in addition , at @xmath42 , observed - frame @xmath37 kev corresponds to rest - frame @xmath10 kev , providing the best possible match to the lower redshift data . in calculating the rest - frame luminosities ,
we have assumed an intrinsic @xmath49 , for which there is only a small differential @xmath50-correction to rest - frame @xmath10 kev .
note that using the individual photon indices ( rather than the universal power - law index of @xmath49 adopted here ) to calculate the @xmath50-corrections would result in only a small difference in the rest - frame luminosities ( barger et al .
the curves in figure [ figbroad]b show the luminosity limits ( using the on - axis flux limits and the @xmath50-corrections ) of the various x - ray surveys used in our analysis .
the break in the curves at @xmath42 results from switching from observed - frame @xmath10 kev flux to observed - frame @xmath37 kev flux in calculating the rest - frame luminosities ( due to the increased sensitivity at @xmath37 kev ) .
we can see the steffen effect from figure [ figbroad]b the population of broad - line agns drops off dramatically at the lower x - ray luminosities .
since broad - line agns are straightforward to identify spectroscopically , even in the so - called redshift `` desert '' at @xmath51 , and since the bulk of the sources in the total x - ray samples have now been spectroscopically observed ( see table [ tab1 ] ) , we do not need to worry that broad - line agns are making up a substantial fraction of the unidentified population .
in fact , at @xmath52 , deep _
observations have picked up all of the color - selected quasars identified by the combo-17 survey in their fields of view ( wolf et al .
2004 ) . moreover , deep _
observations have picked up all of the spectroscopically identified broad - line agns in the highly spectroscopically complete ( to @xmath53 ; cowie et al .
2004b ; wirth et al .
2004 ) acs goods - north region of the cdf - n ( see [ secclass ] ) . at @xmath54 ,
steidel et al . (
2002 ) proposed that there might be a subsample of agns in their lyman break galaxy survey that are relatively x - ray faint and hence would not be detected in even the deepest x - ray pointings .
they based this on a comparison of the agns detected in their lyman break galaxy survey with the x - ray sources detected in a restricted region of the 1 ms exposure of the cdf - n .
four of their 148 lyman break galaxy agn candidates were detected in x - rays , and two of these were spectroscopically identified .
one was found to be an optically - faint broad - line agn at @xmath55 . using their estimated spectroscopic completeness for agns and their detection of this one source
, they concluded that there could be about 11 such optically - faint broad - line agns in a redshift interval of @xmath56 near @xmath42 in a full _ chandra _ acis - i field .
as can be seen from figure [ figbroad]b , we also easily spectroscopically identified the @xmath55 broad - line agn in the cdf - n , and it does indeed have a low x - ray luminosity . however ,
other than this one source , the low x - ray luminosity regime at these high redshifts is devoid of broad - line agns , despite the x - ray sensitivity to such sources and the ease of spectroscopic identifications of broad - line agns . we therefore conclude that there are not very many low x - ray luminosity broad - line agns at @xmath57 .
we will come back to the steffen effect again when we construct the hard x - ray luminosity functions in [ sechxlf ] . to illustrate the range of luminosities covered by the hard and soft x - ray samples , in figure [ figlum ] we show redshift versus rest - frame ( a ) @xmath10 kev and ( b ) @xmath37 kev luminosity for the total hard and soft x - ray samples , respectively . here
the rest - frame luminosities of the x - ray sources _ ( squares ) _ were determined from the observed - frame ( a ) hard and ( b ) soft x - ray fluxes and the @xmath50-corrections calculated using a @xmath49 spectrum .
the rest - frame luminosity limits of the x - ray surveys _ ( curves ) _ were determined from the on - axis ( a ) @xmath10 kev and ( b ) @xmath37 kev flux limits of the surveys and the @xmath50-corrections .
it is possible to extend the redshift information to fainter magnitudes using photometric redshifts rather than spectroscopic redshifts , and this has been done for both the cdf - n ( barger et al .
2002 , 2003b ; p. capak et al .
, in preparation ) and the cdf - s ( wolf et al . 2004 ; zheng et al .
these redshifts are robust and suprisingly accurate ( often to better than 8% ) for non broad - line agns .
the cdf - n photometric redshifts were computed using broadband galaxy colors and the bayesian code of bentez ( 2000 ) , and only sources with probabilities for the photometric redshift of greater than 90% were included .
zheng et al .
( 2004 ) computed their photometric redshifts based on a variety of codes and data sets and included sources with very large offsets from the x - ray positions .
however , the positional offsets should not be large , given the accuracy of the _ chandra _ x - ray positions . moreover , as we move to optically fainter sources , the cross - identifications to the x - ray sources become progressively more insecure .
even with a @xmath58 match radius , about 20% of optical identifications in the @xmath29 range will be spurious , and with larger match radii , many of the identifications will be incorrect ( barger et al .
therefore , in our analysis , we only include a cdf - s photometric redshift from zheng et al .
( 2004 ) where the optical counterpart to the x - ray source lies within @xmath58 of the x - ray position . despite the different methodologies and the slightly fainter flux limits of the cdf - n , both samples have very similar photometric redshift success rates , with just under 85% of the entire x - ray sample being identified in each of the fields .
treister et al .
( 2004 ) claimed that by using the photometric redshifts of mobasher et al .
( 2004 ) for the acs goods - south region of the cdf - s , they could achieve 100% spectroscopic plus photometric redshift completeness in that field . however , mobasher et al .
do not include any photometric redshift reliability measures ( like the @xmath59% probabilities used in the cdf - n ) , and hence treister et al .
are likely including photometric redshifts for sources that are really too optically faint for reliable determinations . in figure [ photz_rmags ]
, we show the spectroscopic ( barger et al . 2003b ; szokoly et al .
2004 ) and photometric ( barger et al . 2003b ; zheng et al .
2004 ) redshift identifications versus @xmath60 magnitude for the cdf - n and cdf - s .
as can be seen from the figure , most sources brighter than @xmath61 can be spectroscopically identified , and photometric redshifts can extend this by almost 2 magnitudes . given the absence , as yet , of a similar set of photometric redshifts for the clasxs region , we have chosen to analyze our samples primarily by assuming that the redshifts of the optically fainter , spectroscopically unidentified sources are essentially unknown .
nevertheless , we may use the photometric redshifts to assess where the unidentified sources are likely to lie , and hence where incompleteness is likely to be a serious concern in computing the luminosity functions ( see [ sechxlf ] ) .
we show a comparison of the photometrically identified sources with the spectroscopically identified sources in figure [ photz_hist ] . as might be expected ,
nearly all of the @xmath4 sources with photometric redshifts are also spectroscopically identified , and incompleteness only becomes a serious concern at @xmath62 , where the sources become optically faint and much harder to identify spectroscopically .
figure [ figlumz ] shows the luminosity versus redshift plot of figure [ figlum]a for just the cdf - n , with the spectroscopic and photometric redshifts denoted by solid and open squares , respectively .
many of the photometrically identified ( but spectroscopically unidentified ) sources correspond to lower x - ray luminosity sources in the @xmath63 range .
we shall assume hereafter that the spectroscopically unidentified sources lie primarily at @xmath62 .
fiore et al .
( 2003 ) have pointed out a correlation between the x - ray to optical flux ratios and the hard x - ray luminosities of non broad - line agns , such that higher x - ray luminosity sources tend to have higher x - ray to optical flux ratios .
this correlation arises due to the obscuration of the nuclear uv / optical light
but not the hard x - ray light in these systems , such that the host galaxy light dominates .
the redshift determination then relies on the fact that the host galaxies have a narrow range in their rest - frame optical magnitudes .
we can use the spectroscopic plus photometric redshift samples in the cdf - n and cdf - s to test the fiore et al .
( 2003 ) relation . in figure
[ fiore_test ] , we show the x - ray to optical flux ratios versus @xmath43 for the non broad - line agns , separated into three @xmath60 magnitude intervals .
unfortunately , we find that the relationship depends on both variables , with the fainter @xmath60 magnitude sources having a higher normalization of the x - ray to optical flux ratio versus @xmath43 .
this means that applying the fiore et al .
( 2003 ) relation to the sources that do not have redshift determinations ( generally because they are faint ) using a calibration that is based only on sources with redshifts ( generally because they are bright ) will result in an overestimate of the redshifts of these sources .
thus , estimates of the number of obscured sources with quasar x - ray luminosities based on this type of analysis will be too high ( see , e.g. , padovani et al .
2004 ) .
it might be possible to make a better estimate of the redshifts using the dependence of the rest - frame absolute magnitude on @xmath43 , together with typical @xmath50-corrections .
however , given the small fraction of unidentified sources in our sample , and the relatively large uncertainties in this type of determination , we do not pursue this in the present paper .
we are now in a position to be able to compare the median redshifts of the total hard and soft x - ray samples at a range of fluxes . in figure
[ figmedz ] , we show spectroscopic redshift distributions for three flux intervals in the total hard x - ray sample .
we also show the median redshifts _
( solid squares ) _ and @xmath64 median redshift ranges _ ( solid bars)_. the @xmath64 upper and lower limits on the median redshifts were determined by placing all of the spectroscopically unidentified sources at arbitrarily high and low redshifts , respectively . at the brighter x - ray fluxes , where most of the sources have been spectroscopically identified , the median redshifts are fairly well determined , while at the fainter x - ray fluxes , where the spectroscopic incompleteness is more substantial , the median redshifts are less well determined . in figure
[ figmedz]c , we also show the median redshift _
( open square ) _ and @xmath64 median redshift range _
( dotted bars ) _ when we use only the cdf - n and cdf - s data and include the photometric redshifts in the computation , allowing for sources with neither spectroscopic nor photometric redshifts in the same way as above .
this substantially improves the accuracy of the median redshift determination in the lowest x - ray flux bin .
steffen et al . ( 2004 ) determined median redshifts for the surveys used in the present paper combined with other _
( lehmann et al . 2001 ) and _ xmm - newton _ (
mainieri et al .
2002 ; fiore et al .
2003 ) surveys .
their redshift versus @xmath37 kev and @xmath10 kev flux diagrams ( their figures 10a and 10b ) showed that the median redshifts of the surveys remained about constant at @xmath0 and were not strongly correlated with either the soft or hard x - ray fluxes .
they cautioned , however , that since the magnitudes of high - redshift sources may be faint , the spectroscopic surveys may be biased against finding high - redshift sources , particularly at the fainter x - ray fluxes where the surveys are more incomplete .
we examine this issue in more detail in figures [ figfxmedz]a and [ figfxmedz]b , where we show the spectroscopic median redshifts _
( solid squares ) _ and @xmath64 median redshift ranges _ ( solid bars ) _ , calculated as in figure [ figmedz ] , versus @xmath10 kev and @xmath37 x - ray flux , respectively , for the total hard and soft x - ray samples . at the lower x - ray fluxes , we also show the median redshifts _ ( open squares ) _ and @xmath64 ranges _ ( dotted bars ) _ for the spectroscopic plus photometric cdf - n and cdf - s data .
again we see that the median redshifts of the hard and soft x - ray samples are very similar .
moreover , with the tighter spectroscopic plus photometric median redshift ranges , we see that the median redshifts are indeed fairly constant with flux .
the dotted curves show the standard redshift - luminosity relation for a source with rest - frame @xmath10 kev luminosity @xmath65 ergs s@xmath2 .
the lower x - ray flux sources are clearly dominated by sources with luminosities less than this value and redshifts near one .
we shall return to this point when we compute the hard x - ray luminosity functions in [ sechxlf ] .
the optical spectra of the x - ray sources in our sample , while generally of high quality , span different rest - frame wavelengths and suffer varying degrees of agn mixing with the host galaxy spectrum at different redshifts
. it would therefore be quite hard to classify the x - ray sources in any uniform way using a conventional agn classification scheme . recognizing this , we have instead only roughly classified the spectroscopically identified _
sources into four optical spectral classes .
we call sources without any strong emission lines ( ew([oii])@xmath66 or ew(h@xmath67nii)@xmath68 ) _ absorbers _ ; sources with strong balmer lines and no broad or high - ionization lines _ star formers _ ; sources with [ nev ] or civ lines or strong [ oiii ] ( ew([oiii ] 5007 @xmath69 ew(h@xmath70 ) _ high - excitation ( hex ) sources _ ; and , finally , sources with optical lines having fwhm line widths @xmath71 km s@xmath2 _ broad - line agns_. we have chosen these four classes to roughly match those used by szokoly et al .
( 2004 ) for the cdf - s ( they call the second category _
low - excitation or lex _ sources ) in order to combine our own classifications of the cdf - n and clasxs sources with their classifications of the cdf - s sources . in this paper
, we will sometimes combine the absorber and the star former classes into a _ normal galaxy _ class .
table [ tab1 ] gives the breakdown of the cdf - n , cdf - s , and clasxs samples by spectral type .
note that four of the cdf - n redshifts are from the literature ( see barger et al .
2003b for references ) and hence do not have spectral typings .
hereafter , we call all of the sources that do not show broad - line ( fwhm@xmath71 km s@xmath2 ) signatures `` optically - narrow '' agns
. however , we note that there may be a few sources where our wavelength coverage is such that we are missing lines which would result in us defining the spectrum as broad - line . for the cdf - n and clasxs fields
, we made a more quantitative analysis by measuring the fwhm line widths for each of the civ 1550 , [ ciii ] 1909 , mgii 2800 , h@xmath72 , and h@xmath73 lines that were in the observed spectra by fitting gaussian profiles to the lines .
where more than one of these lines was within the spectrum , we took the maximum of the measured fwhms to be the fwhm .
there were only significant differences in the measured widths of the various lines for a small number of cases where h@xmath72 was narrow and mgii 2800 was wide . for a small number of other spectra ,
none of the lines were in the observed range , and hence we were not able to classify them .
for the cases where the wavelengths of the lines were within our spectra , but no emission lines were visible , we set the widths to zero .
figure [ figclwid ] shows line width versus optical spectral class for the cdf - n and clasxs sources .
the choice of 2000 km s@xmath2 as the dividing line between the high - excitation sources and the broad - line agns is in itself rather arbitrary , and many of the sources lying in the high - excitation category would rather naturally fall into the narrow - line seyfert 1 galaxy definition ( osterbrock & pogge 1985 ; goodrich 1989 ) . to investigate this further , we ran the same classification scheme on the large optical spectroscopic sample given in cowie et al .
( 2004b ) and wirth et al .
( 2004 ) for the acs goods - north region of the cdf - n field .
figure [ field_widths ] shows soft x - ray flux versus fwhm line width for these sources . of the 1718 sources in the goods region where we could measure line widths
, 20 had fwhm@xmath74 km s@xmath2 , and 13 had fwhm@xmath71 km s@xmath2 .
all of these sources are x - ray sources .
we may make two points from this .
first , x - ray selected samples at the flux limit of the cdf - n essentially find all of the broad - line agns .
however , some of the intermediate width ( @xmath75 km s@xmath2 ) sources would not be found in a field with the clasxs depth , though they are well above the on - axis cdf - n flux limit .
second , a split at 800 km s@xmath2 might seem , in some ways , to be more objective than one at 2000 km s@xmath2 .
however , when we turn to the x - rays , it is clear that the primary distinction is between the broad - line agns and the optically - narrow agns .
indeed , this is the only distinction that can easily be made from the x - ray colors or x - ray spectra ( szokoly et al .
2004 ) . in figure
[ softwid ] , we show the @xmath37 kev to @xmath10 kev flux ratio for the sample of sources with strong @xmath10 kev fluxes in the cdf - n and clasxs fields versus fwhm line width . above 2000
km s@xmath2 , almost all of the sources are soft ( @xmath49 ) , while below this line width , there is a wide span of x - ray colors . the intermediate width sources show a very similar spread in their x - ray colors relative to all of the remaining optically - narrow sources .
this result is consistent with recent work by williams , mathur , & pogge ( 2004 ) , who used _
_ observations of a sample of optically - selected , x - ray
weak , narrow - line seyfert 1 galaxies to show that strong , ultrasoft x - ray emission is not a universal characteristic of narrow - line seyfert 1 galaxies , and , indeed , that many narrow - line seyfert 1 galaxies have weak or hard x - ray emission . for the cdf - n sources , which have the deepest exposures , we can perform a more sophisticated color analysis . in figure [ figcol ]
, we show the color - color plot of the @xmath76 kev to @xmath77 kev counts ratio versus the @xmath37 kev to @xmath10 kev counts ratio for the @xmath77 kev _ chandra _ sources that have more than 100 counts in that band , separated by optical spectral type ( broad - line agns in figure [ figcol]a ; optically - narrow agns in figure [ figcol]b ) .
the broad - line agns are nearly all soft with mean photon indices @xmath49 and essentially no visible absorption in x - rays .
by contrast , the optically - narrow agns are well described by a simple model in which a power - law spectrum with @xmath49 is suppressed at low energies by photoelectric absorption spread over a very wide range of absorbing column densities ( barger et al .
2002 ; alexander et al .
there is little dependence of the absorbing column densities on optical spectral type ( shown by the different symbols in figure [ figcol]b ) or on the line widths of the optically - narrow agns .
thus , the presence of optical seyfert 2 or narrow - line seyfert 1 galaxy characteristics does not seem to be dependent on the absorbing column density to the x - ray source . from figures [ softwid ] and [ figcol ] , we see that it is possible to separate roughly the broad - line agns and the optically - narrow agns on the basis of the x - ray colors alone ( e.g. , szokoly et al .
2004 ) , without knowing the optical spectra .
however , there will be a small amount of contamination from stars , from the small number of optically - narrow agns that have soft x - ray colors , and from the small number of broad - line agns that have hard x - ray colors .
( note that perola et al .
2004 found that about 10% of their broad - line agns showed some x - ray obscuration , with estimated @xmath78 @xmath20 . )
to study the nuclear uv / optical properties of the @xmath10 kev selected sources , we use multicolor observations of the cdf - n taken with the acs camera on _ hst _ in four bands ( acs f435w , f606w , f814w , and f850lp ) as part of goods ( giavalisco et al .
( grogin et al .
2003 used earlier _ hst _ observations of the cdf - s for this type of analysis . ) because of the high spatial resolution of these data , we can separate the nuclear component of each source from the host galaxy light , even at the higher redshifts , in order to analyze the nuclear colors . thus , _ hst
_ makes it possible for us to reproduce the types of analyses that have been done for decades on low - redshift agns and optically - bright agns . since the galaxies often have complex morphologies ( see figure [ figimages ] for examples ) , we decided to use a very simple prescription to separate the nuclear magnitudes from the galaxy light .
for each source , we first located the peak of the optical light within a @xmath58 radius around the x - ray position .
we then measured the magnitude in a @xmath79 aperture radius centered on this optical position .
based on measurements of 196 spectroscopically confirmed stars in the field , we found that this radius corresponded to the 80 percent enclosed energy fraction .
( this is also consistent with the enclosed energy curves in the acs handbook . )
thus , the @xmath79 aperture magnitude corrected for this enclosed energy fraction provides a first estimate of the nuclear magnitude . in order to provide an improved estimate of the nuclear magnitude with the galaxy contribution removed , we followed an iterative procedure .
we first measured the light within a @xmath80 annulus around the peak position of the optical light .
we then subtracted from this measurement the expected contribution from the nucleus ( 10% of the @xmath79 radius light , as determined from the psf ) in order to leave only the galaxy contribution .
based on measurements of objects without strong nuclei that otherwise appear to have similar spectral characteristics , we found that the value of the galaxy light in the central @xmath79 radius relative to that in the @xmath80 annulus is @xmath81 .
( we note that this number is rather rough and does not take into account redshift effects . )
thus , we subtracted this estimate of the galaxy light from the @xmath79 aperture magnitude to obtain a revised estimate of the nuclear magnitude .
the whole process was then repeated until the estimates converged .
the division between the nucleated sources and those without strong nuclei is generally quite clear .
we have illustrated this for three redshifts in figure [ figimages ] .
some objects ( such as those with double nuclei ) can be problematic , but rather than correct these by hand , we have chosen to stay with the objectively measured values .
we ran the above procedures on all four color bands in the acs sample , determining nuclear magnitudes for all 286 objects in the cdf - n 2 ms list that lie within the goods - north region .
the galaxy magnitudes are then just the isophotal acs goods - north magnitudes minus the nuclear magnitudes .
there is a good correlation between the fraction of nuclear uv light in the galaxy and the optical spectral type , which gives us confidence in the procedure , since these are completely independently measured properties . in figure
[ nuc_gal_type ] , we show nuclear uv magnitude minus galaxy magnitude for all of the x - ray sources versus optical spectral type . the broad - line agns and the stars are all strongly nucleated , while the absorbers and star formers have much weaker nuclei , and the high - excitation sources lie at intermediate values .
of the 286 sources measured , 130 have no detectable nucleus in the @xmath82-band , and 126 have no detectable nucleus in the @xmath83-band .
only 4 of the sources without nuclei are spectrally classified as high - excitation sources , and none are classified as broad - line agns .
with the advent of the _ chandra _ and _ xmm - newton _ data , there have been several computations of the evolution of the hard x - ray luminosity functions with redshift ( e.g. , cowie et al .
2003 ; hasinger 2003 ; steffen et al . 2003
; ueda et al . 2003 ; fiore et al . 2003 ) .
it is a measure of how rapidly the science is evolving that the _ chandra _ spectroscopic samples are now much improved over those used in even these recent calculations .
the _ chandra _ samples of table [ tab1 ] now contain 1165 spectroscopically observed sources and 715 spectroscopically identified sources . in calculating the hard x - ray luminosity functions ,
we use only the spectroscopically observed sources .
this approach is strictly valid for clasxs and the cdf - s , where the x - ray sources were randomly observed . for the cdf - n
, there may be a small bias since the spectroscopy does not come purely from targeted observations of the x - ray sample . however , as there are now only 52 x - ray sources in the cdf - n that have not been observed , it would make very little difference to the results if we were instead to use all of the sources .
there are a total of 698 spectroscopically observed sources in the
cdf - n and cdf - s samples . of these
, 601 have either spectroscopic or photometric redshifts .
we can test for the effects of incompleteness in our analysis by using the spectroscopic plus photometric redshifts in these two fields and the spectroscopic redshifts for the _ asca _ data , where only one source is unidentified . here
we recompute the hard x - ray luminosity functions following cowie et al .
( 2003 ) , who used the traditional @xmath84 method of felten ( 1977 ) .
we use the spectroscopically observed samples of table [ tab1 ] .
we define the hard x - ray luminosity function versus rest - frame @xmath10 kev x - ray luminosity and redshift , @xmath85 , as the number of x - ray sources per unit comoving volume per unit base 10 logarithmic luminosity that lie in the redshift interval .
we determine the solid angle covered by the observed sources at a given flux by comparing the observed number of sources versus flux with the averaged number counts in the appropriate energy band from cowie et al .
( 2002 ) and yang et al .
this method allows a simple treatment of the incompleteness that was modeled in computing the counts .
however , the counts in cowie et al
. also include the low cdf - s counts , which may affect the normalization at the 10% level relative to the clasxs average ( yang et al .
we consider this to be a reasonable estimate of the systematic errors . at @xmath86 ergs @xmath20 s@xmath2 ( @xmath10 kev ) , where the contribution to the @xmath10 kev x - ray background peaks ( see [ seccomplete ] ) , the solid angle is dominantly from clasxs . in figure [ figarea ] ,
we show solid angle versus ( a ) @xmath10 kev and ( b ) @xmath37 kev flux for the observed hard and soft x - ray samples , respectively . in figures
[ figlowhxlf]a , [ figlowhxlf]b , and [ figlowhxlf]c , we show the hard x - ray luminosity functions for all spectral types ( _ squares _ ; this includes all of the spectroscopically identified x - ray sources with @xmath3 ergs s@xmath2 , without regard to the optical spectroscopic classifications ) and for broad - line agns ( _ diamonds _ ; this includes all of the spectroscopically identified broad - line agns ) for three low - redshift intervals .
these hard x - ray luminosity functions were computed from the observed - frame @xmath10 kev measurements , assuming an intrinsic @xmath49 .
the poissonian @xmath64 uncertainties are based on the number of galaxies in each luminosity bin . at these low redshifts
, there is relatively little uncertainty from incompleteness , since the sources have x - ray fluxes lying in the range where most sources are spectroscopically identified .
the triangles show the values computed for what we hereafter refer to as the `` spectroscopic plus photometric '' sample . for this sample ,
we consider only the cdf - n , cdf - s , and _ asca _ spectroscopically observed sources and assign cdf - n and cdf - s photometric redshifts , where available , to those sources that are spectroscopically unidentified .
the resulting identified sample is therefore a combination of spectroscopic and photometric redshifts .
note that because this sample does not contain the clasxs data , in some cases the values _ ( triangles ) _ lie below the spectroscopic values _ ( squares ) _ due to poor sampling by the deep fields of the low - redshift volume and of the high - luminosity sources . in figure [ figlowhxlf]d
, we show the local @xmath87 kev luminosity function from the _ rxte _ analysis of sazonov & revnivtsev ( 2004 ) .
the forms of the hard x - ray luminosity functions for all spectral types and for broad - line agns are clearly very different .
the hard x - ray luminosity functions for broad - line agns peak at a characteristic luminosity , such that the dominant population at the higher x - ray luminosities is broad - line agns , while the dominant population at the lower x - ray luminosities is optically - narrow agns .
this reproduces the steffen effect discussed in [ secxlum ] .
we show this more clearly in figure [ figtype ] , where we plot the ratio _ ( squares ) _ of the hard x - ray luminosity functions for all spectral types to the hard x - ray luminosity functions for broad - line agns . above @xmath65
ergs s@xmath2 , the fraction of sources that are broad - line agns is very high , while below @xmath88 ergs s@xmath2 , the fraction of sources that are broad - line agns is very low .
this result still holds when we include in the hard x - ray luminosity functions for all spectral types all of the spectroscopically unidentified sources .
this is done by placing the unidentified sources at the center of each redshift interval and then including them in the luminosity bin where their luminosities at that redshift would put them _ ( dashed curves)_. thus , we have clearly detected a luminosity dependence in optical spectral type . in figures
[ figlowhxlf]e g , we show the division between the optically normal galaxies ( our absorber and star former classes ) and the sources which show clear agn signatures in their optical spectra ( our high - excitation and broad - line agn classes ) . including the high - excitation sources slightly raises the normalization of the broad - line agn luminosity function , but it does not change its shape much . at low x - ray luminosities ,
the optically normal galaxies comprise nearly the entire population .
the cross - over point lies around @xmath89 ergs s@xmath2 at @xmath90 and rises to @xmath91 ergs s@xmath2 near @xmath13 .
all of the curves in figure [ figlowhxlf ] show pure luminosity evolution in both the luminosity functions and the spectral type mix .
this must be understood in any model that seeks to explain the difference between the optically normal galaxies and those with agn characteristics in their optical spectra .
this type of parallel evolution is probably most easily understood for a luminosity dependent unified model ( see [ seclum ] ) .
the low - redshift hard x - ray luminosity functions are well represented ( _ solid and dashed curves _ ; figure [ figlowhxlf ] ) by a conventional double power - law fit ( piccinotti et al .
1982 ) of the form @xmath92 parameterizing the redshift evolution as @xmath93 and @xmath94 we determined the six parameters and their @xmath64 uncertainties for both the hard x - ray luminosity function for all spectral types and the hard x - ray luminosity function for broad - line agns using maximum likelihood fits over the redshift range @xmath12 ( cash 1979 ; marshall et al . 1983 ) .
the parameters are summarized in table [ tab2 ] .
the evolution of both hard x - ray luminosity functions is consistent with pure luminosity evolution as is the evolution of the hard x - ray luminosity function for optically - selected broad - line agns ( boyle et al .
2000)where the values of @xmath95 , @xmath96 , and @xmath97 remain constant and only @xmath5 evolves .
the fits are acceptable throughout the redshift range , as shown in figures [ figlowhxlf]a d .
lcc @xmath98 & @xmath99 & @xmath100 @xmath101 & @xmath102 & @xmath103 @xmath104 & @xmath105 & @xmath106 @xmath107 & @xmath108 & @xmath109 @xmath110 & @xmath111 & @xmath112 @xmath82 & @xmath113 & @xmath114 [ tab2 ] the above fit to the hard x - ray luminosity function for all spectral types agrees well with that of ueda et al .
( 2003 ) at higher luminosities but is shallower below the break , where ueda et al .
find @xmath115 for their pure luminosity evolution model .
in fact , the ueda et al .
slope overpredicts our low - luminosity counts throughout the @xmath12 redshift range .
our data sample is much larger than the sample used in ueda et al .
, particularly in this redshift range , because of our inclusion of the wide - field clasxs data ; correspondingly , our fit should be more robust . our @xmath116 evolution at @xmath4 is steeper than the @xmath117 evolution found by ueda et al .
( 2003 ) , though consistent within the uncertainties .
sazonov & revnivtsev ( 2004 ) found that the ueda et al .
( 2003 ) model lay significantly above their local determination .
in contrast , our steeper luminosity evolution matches the sazonov & revnivtsev ( 2004 ) local determination within the uncertainties in the determination of the @xmath110 parameter ( see figure [ figlowhxlf]d ) .
thus , agns drop in luminosity by almost an order of magnitude over the @xmath118 redshift range .
this drop applies equally to all agns , regardless of optical spectral type .
we shall return to the question of whether this is an evolution of the accretion rate or of the active population of supermassive black holes in [ secsmbh ] .
we also determined the hard x - ray luminosity functions for two high - redshift intervals , @xmath119 and @xmath120 , using a @xmath37 kev observed - frame sample to provide the best possible match to the lower redshift data .
figures [ fighihxlf]a and [ fighihxlf]b show these measured hard x - ray luminosity functions ( _ squares_all spectral types ; _ diamonds_broad - line agns ) .
it is very difficult to measure host galaxy redshifts for sources that lie in the high - redshift intervals , so incompleteness is potentially a large source of error .
we therefore computed maximal hard x - ray luminosity functions for all spectral types by assigning redshifts at the center of each redshift interval to all of the spectroscopically unidentified sources ( i.e. , we included all of the unidentified sources in both redshift intervals ) .
we show these as dot - dashed curves in the figures .
it is important to keep in mind that because all of the unidentified sources have been included in both redshift intervals ( provided that the sources , when assigned those redshifts , are at @xmath3 ergs s@xmath2 ) , the curves are not consistent with one another . because the spectroscopic identifications are much more complete at higher x - ray fluxes , the associated systematic uncertainties are larger at lower @xmath43 .
( note that any low x - ray flux source assigned to a given redshift interval will have a low @xmath43 in that redshift interval . )
we also show as triangles the spectroscopic plus photometric hard x - ray luminosity functions determined from the cdf - n , cdf - s , and _ asca _ data only . in the @xmath119 redshift interval ,
the spectroscopic plus photometric hard x - ray luminosity function is much closer to the maximal hard x - ray luminosity function than is the spectroscopic hard x - ray luminosity function , whereas in the @xmath121 redshift interval , the spectroscopic plus photometric hard x - ray luminosity function is very similar to the spectroscopic hard x - ray luminosity function .
this suggests that many of the spectroscopically unidentified sources lie in the @xmath119 interval rather than at higher redshifts .
even with our maximal incompleteness corrections , at @xmath91 ergs s@xmath2 , the hard x - ray luminosity functions for all spectral types in the two high - redshift intervals lie below the maximum likelihood fits to the @xmath12 hard x - ray luminosity function computed at @xmath13 ( _ solid curve _ ) , suggesting a peak in the universal agn energy density production rate near @xmath13 .
however , at @xmath45 ergs s@xmath2 , the maximal hard x - ray luminosity functions in both high - redshift intervals appear to continue the increasing trend with redshift . before proceeding ,
we want to make sure that we would not have gotten different results if we had used the @xmath10 kev observed - frame sample instead of the @xmath37 kev observed - frame sample in determining the high - redshift hard x - ray luminosity functions .
we performed this check by computing the @xmath119 hard x - ray luminosity function for broad - line agns using the @xmath37 kev observed - frame sample and also using the @xmath10 kev observed - frame sample . in figure
[ figshcomp ] , we show that the agreement between these two computations is very good .
since the x - ray selected broad - line agns recover essentially all of the optically - selected type 1 agns ( see [ secclass ] ) , we should be able to compare directly the broad - line agn hard x - ray luminosity functions computed above with the optical qso luminosity functions . in order to compare the two
, we calculate the bolometric luminosities using the bolometric corrections determined by elvis et al .
( 1994 ) ( i.e. , @xmath122 for the broad - line agn hard x - ray luminosity functions , and @xmath123 for the qso luminosity functions ) .
in figure [ blagn_opt ] , we show for six redshift intervals the broad - line agn hard x - ray luminosity functions _ ( open diamonds ) _ and the rest - frame @xmath82-band qso luminosity functions ( croom et al . 2004 ; _ solid squares _ ) versus the calculated bolometric luminosities .
the croom et al .
luminosity functions have been renormalized to our assumed geometry .
the bright end luminosity functions agree extremely well at all redshifts , confirming that the two methodologies are measuring the same sample and that the bolometric corrections used are appropriate .
the optical qso luminosity functions do not probe faint enough to see the downturn in the broad - line agn hard x - ray luminosity functions .
moreover , they may be missing some sources at the very lowest luminosities to which they probe , as can be seen from the lowest redshift panels .
the optical qso luminosity function in the @xmath124 to @xmath125 range is well described by pure luminosity evolution over the redshift range @xmath126 ( boyle et al . 2000 ) .
croom et al . ( 2004 ) parameterize this as @xmath127 evolving as @xmath128 , where @xmath129 and @xmath130 . in figure
[ lumevol_comp ] , we compare the croom et al . evolution with the @xmath131 law we determined in [ secevol ] for the broad - line agns ( see table [ tab2 ] ) .
the croom et al .
evolution is slightly steeper , though part of the difference may lie in the adopted functional forms and in the different redshift ranges over which the laws have been fitted .
however , within the uncertainties , the two determinations are consistent over the @xmath12 redshift interval .
we further investigate the pure luminosity evolution model for broad - line agns by using the croom et al .
( 2004 ) evolution law ( which was fit over a wider redshift range , @xmath126 , than our maximum likelihood fit was , @xmath12 ) to correct all of the x - ray luminosities to their values at @xmath13 .
we then compute the hard x - ray luminosity functions for broad - line agns over the wide redshift ranges @xmath132 , @xmath133 , and @xmath134 .
the resulting hard x - ray luminosity functions are compared in figure [ delum_comp ] . the lower redshift functions _ ( open triangles and open diamonds )
_ match each other and our maximum likelihood fit computed at @xmath13 _
( dashed curve ) _ throughout the luminosity range , while the highest redshift function _ ( solid squares ) _ matches the lower redshift functions and our maximum likelihood fit only at the bright end , where the optical qso determinations are made .
it is clear that there are fewer intermediate luminosity sources in the highest redshift interval , and hence that the pure luminosity evolution model can not in fact be carried reliably to the higher redshifts .
this is another statement that the lower luminosity sources peak in number at lower redshifts than the high - luminosity sources ( e.g. , cowie et al .
2003 ; hasinger 2003 ; barger et al . 2003a
; steffen et al . 2003 ; fiore et al .
2003 ; ueda et al .
in figure [ figevol ] , we show the evolution with redshift of the @xmath10 kev comoving energy density production rate , @xmath135 , of all spectroscopically identified @xmath3 ergs s@xmath2 agns _ ( solid squares ) _ and broad - line agns alone _ ( open diamonds)_. the production rate rises rapidly from @xmath14 to @xmath13 .
the solid and dashed curves show the @xmath136 and @xmath131 laws found for all spectral types and for broad - line agns alone , respectively , in [ secevol ] .
there is no reason why the evolution should be well fitted by a power of @xmath137 , and , indeed , at @xmath138 , the evolution is turning over in redshift , at least in part due to incompleteness .
if we had instead fitted over a lower redshift range ( say @xmath14 to @xmath13 ) , then we would have obtained a slightly steeper relation .
such an extremely rapid evolution with redshift bears a striking resemblance to the overall redshift evolution of the star formation rate density in @xmath139 galaxies .
the uv luminosity density in small star - forming galaxies is falling as @xmath140 ( wilson et al .
2002 ; note that there have been a whole slew of relations determined , including some that are steeper ) .
however , agns are more likely to be related to the bulge luminosity of galaxies , and hence to the submillimeter and ultraluminous infrared galaxy @xmath141 populations .
the evolution of the star - forming luminosity function at @xmath142 , as determined from radio sources with the equivalent radio power of at least @xmath143 , is steeper and reasonably well described by a @xmath144 evolution ( cowie et al .
2004a ) .
the peak of the hard x - ray energy density production rate of all spectroscopically identified @xmath3 ergs s@xmath2 agns is in the interval @xmath145 and is about @xmath146 ergs s@xmath2 mpc@xmath147 .
about one - third of this production arises in broad - line agns and about two - thirds in optically - narrow agns .
half of this production arises in sources more luminous than @xmath148 ergs s@xmath2 , where the majority of the sources are broad - line agns .
the above results are based on direct summation over the individual sources , but integration of the power - law fits to the hard x - ray luminosity functions give similar answers .
the largest uncertainty is the redshift distribution of the spectroscopically unidentified sources , not the small effects of extrapolation outside of the observed luminosity range . in figure
[ figevol ] , we assume that most of the spectroscopically unidentified sources in the spectroscopically observed samples lie at @xmath149 ( we examine this in more detail below ) .
we then determine the maximal incompleteness for the two high - redshift intervals _
( open squares ) _ by using the sources that have a measured spectroscopic redshift in these intervals and by assigning redshifts at the center of each interval to all of the spectroscopically unidentified sources in the spectroscopically observed samples .
note that the maximal incompleteness values are not consistent with one another , because all of the unidentified sources are included in both redshift intervals , provided that the sources , when placed at those redshifts , lie at @xmath3 ergs s@xmath2 .
even with the incompleteness uncertainty , @xmath135 is at most flat beyond @xmath13 . to explore more carefully the incompleteness at all redshifts , in figure [ figlightden ]
, we show the production rate for the spectroscopic samples in the cdf - n , cdf - s , and _ asca _ fields only _
( solid squares ) _ , and , at @xmath150 , the production rate for the spectroscopic plus photometric samples in these fields _ ( triangles)_. we see that the only substantial difference at @xmath150 is in the @xmath151 redshift interval . at @xmath13 ,
the symbols are a bit higher than the solid square was in figure [ figevol ] ( the solid curve from figure [ figevol ] has been replotted in figure [ figlightden ] to illustrate this ) because of the excess in the cdf - n at that redshift , which gets smoothed out when the clasxs data are included .
we compute the maximal incompleteness for the two high - redshift intervals _
( open squares ) _ by using the sources with either a spectroscopic or photometric redshift in these intervals and by assigning redshifts at the center of each interval to all of the remaining unidentified sources in the cdf - n , cdf - s , and _ asca _ fields .
the inclusion of the photometric data does not change the conclusion that at @xmath138 , @xmath135 is flat or , more realistically , slightly falling .
we can go one step further in studying the evolution of the energy density production rate . in figure [ figevolblagn ]
, we separate out the optically - narrow agns _
( solid circles ) _ , incompleteness correct their production rates ( see below ) , and then compare the evolution of their corrected production rates with the evolution of the production rates of the broad - line agns ( _ open diamonds _ ; values from figure [ figevol ] ) . over the redshift range @xmath152 , the incompleteness corrections were obtained by multiplying
the spectroscopically identified optically - narrow agns by the ratios of the spectroscopic plus photometric values in figure [ figlightden ] ( the triangles in that figure ) to the spectroscopic values in figure [ figlightden ] ( the solid squares in that figure ) . at @xmath150 ,
the only place where the incompleteness corrections are significant is in the redshift interval @xmath151 . for @xmath119
, we can not incompleteness correct so simply .
instead , we compute the possible range of values ( _ vertical bar _ ) by subtracting the value given in figure [ figevol ] for broad - line agns ( the diamond at @xmath119 ) from the value given in figure [ figevol ] for all spectral types ( the solid square at @xmath119 ) to get the lower value _ ( solid circle ) _ , and by subtracting the same value for broad - line agns from the value given in figure [ figevol ] for the maximal incompleteness ( the open square at @xmath119 ) to get the upper value _
( open circle)_. from figure [ figevolblagn ] , we conclude that at @xmath139 , most of the hard x - ray energy density production is coming from optically - narrow agns , while at higher redshifts , this reverses , and the broad - line agns dominate the production , even with the maximal incompleteness .
the broad - line agn production is fairly well represented by a constant at higher redshifts ( see discussion in [ secacc ] ) , as illustrated by the dashed curve at @xmath138 in the figure .
( at @xmath139 , the dashed curve shows the pure luminosity evolution maximum likelihood fit for broad - line agns over the redshift range @xmath12 from figure [ figevol ] . )
the rapid drop in the luminosities of the individual black holes ( both broad - line agns and optically - narrow agns ) with decreasing redshift between @xmath13 and @xmath14 , as well as the drop in the overall energy density production by the agns , could arise in two ways .
first , the accretion rate of all supermassive black holes , regardless of mass , could be declining .
we refer to this as `` mass starvation '' .
second , the more massive black holes could be being preferentially starved , while the less massive black holes continue to accrete .
this would parallel the behavior of star formation , which is dominated by lower mass sources at lower redshifts . following the naming introduced by cowie et al .
( 1996 ) for that process , we refer to this possibility as `` agn downsizing '' . for the broad - line agns
, we can use the line widths and nuclear optical luminosities to estimate supermassive black hole masses ( wandel , peterson , & malkan 1999 ; kaspi et al .
2000 ) . given the redshifts of the sources
, we only have a substantial sample of uniform line widths for the mgii 2800 line ( measured from sources in the acs goods - north region of the cdf - n ) .
we use the mclure & dunlop ( 2004 ) calibration based on this line and the rest - frame 3000 luminosity of the nucleus to estimate the black hole masses for those sources with mgii line widths greater than 2000 km s@xmath2 .
we note , however , that the sample is small ( 7 sources between redshifts of @xmath153 and 1.5 ) , and the simple fitting technique that we use to measure the line widths may result in some contamination by the feii line complex ( see mclure & dunlop 2004 ) .
there are also many uncertainties in the method , so the presently derived values should only be considered as a rough measure ( accurate to factors of a few ) of the black hole masses . in figure [ edd_plot ]
, we show the ratios of the bolometric luminosities ( @xmath122 ; the bolometric correction is from elvis et al .
1994 and kuraszkiewicz et al .
2003 for broad - line agns ) to the eddington luminosities that correspond to the estimated black hole masses . over nearly two orders of magnitude in @xmath43
, there is no obvious change in these eddington ratios , with individual values varying by less than a factor of two about a median value of 0.16 .
( we note , however , that the x - ray selection means that we are only seeing the high - end values of the eddington ratio ; e.g. , see alonso - herrero et al .
2002 and merloni et al . 2003 . )
this is partly a consequence of the tight relation between @xmath43 and @xmath154 ( see woo & urry 2002 for a discussion ) , as @xmath154 enters as roughly the square root in the black hole mass calculation ( e.g. , wandel et al .
1999 ; mclure & dunlop 2004 ) , but the relation is much tighter than would be expected solely from this effect .
indeed , the eddington ratio depends on the line width squared , and the line widths of the sources vary by nearly a factor of four .
this suggests that at @xmath13 , there is a fairly tight correlation between @xmath43 and black hole mass , @xmath155 for a source accreting at the median value of @xmath156 of the eddington luminosity measured for the seven sources , the mass doubling time is about @xmath157 years .
( note that the mass growth rate of a black hole accreting at the eddington rate is exponential with an e - folding timescale of @xmath158 yr , the salpeter time ; see natarajan 2004 for a summary of the basic definitions used in the accretion paradigm . ) since a few percent of luminous galaxies are active at any time ( barger et al . 2001a , 2003b ; cowie et al .
2004b ) , we would expect that the whole population of supermassive black holes would have significantly grown over timescales similar to that of the hubble time at @xmath13 .
this result is consistent with the ongoing formation of the supermassive black hole population at @xmath13 that we found in [ secenergy ] . in figure [ galevol ] ,
we show for the sources in the acs goods - north region of the cdf - n the host galaxy luminosities ( excluding the nuclear contributions ) computed at rest - frame 5000 versus redshift .
the luminosities are shown for three intervals in @xmath43 ( _ triangles_@xmath45 ergs s@xmath2 ; _
diamonds_@xmath159 ergs s@xmath2 ; _
squares_@xmath160 ergs s@xmath2 ) .
we chose rest - frame 5000 because it is the largest rest wavelength that still corresponds to an observed optical wavelength at @xmath139 .
we wanted to choose the reddest rest - frame wavelength possible to give the best representation of the mass and to minimize the susceptibility to evolution in the star formation . from the figure
, we can see that the most luminous x - ray sources lie in the most optically - luminous host galaxies .
the median absolute magnitudes in the three x - ray luminosity ranges listed above are @xmath161 , @xmath162 , and @xmath163 , respectively , while the median x - ray luminosities are @xmath164 ergs s@xmath2 , @xmath165 ergs s@xmath2 , and @xmath166 ergs s@xmath2 .
this seems to be in accord with what we would expect from the well - known bulge luminosity black hole mass relation ( kormendy & richstone 1995 ; magorrian et al .
1998 ) ; that is , the highest mass supermassive black holes ( the accretion onto which , as we found from our sample of seven , produces the highest x - ray luminosities ) should lie in the most optically - luminous galaxies . more importantly for the present discussion , the host galaxy optical luminosities for a given interval in @xmath43 do not appear to be changing much with redshift ( i.e. , the x - ray sources at low redshifts are occurring in the same host galaxies as they are at high redshifts ) , and the most optically - luminous host galaxies that existed at high redshifts have completely disappeared at low redshifts . if the accretion rates of all supermassive black holes , regardless of mass , were just declining , then we would see a different picture : the host galaxy optical luminosities for a given interval in @xmath43 would be higher at low redshifts than at high redshifts , and the most optically - luminous host galaxies would still contain the most x - ray luminous sources , even though the @xmath43 values of these sources would be lower .
the most optically - luminous host galaxies would not just cease to exist as x - ray sources .
thus , figure [ galevol ] suggests that the loss of the highest @xmath43 sources is indeed an agn downsizing effect .
the highest mass supermassive black holes in the most optically - luminous host galaxies are switching off between @xmath13 and the present time , leaving only the lower mass supermassive black holes in the less optically - luminous galaxies active .
this is consistent with the work of heckman et al .
( 2004 ) , who found from their analysis of the sloan digital sky survey data that most of the present - day accretion is occurring onto low - mass black holes ( @xmath167 ) , with massive black holes currently experiencing very little additional growth .
one possible way to account for at least part of the steffen effect , i.e. , the absence of broad - line agns at low x - ray luminosities , would be if the nuclear optical light of the optically - narrow agns were being swamped by the host galaxy light , thereby washing out the broad - line signal ( moran , filippenko , & chornock 2002 ; moran 2004 ) . in the present spectroscopic obsertvations ,
this effect would be less pronounced at lower redshifts , where the nuclear light is a larger part of the light entering the spectrograph slit than the extended galaxy . in this section ,
we explore the possibility of galaxy dilution by looking at the nuclear uv / optical properties of the _ chandra _ x - ray sources and determining how these properties relate to the spectral characteristics of the sources .
if the dilution hypothesis is correct , then we would expect the optically - narrow agns to have the same optical to x - ray ratios as the broad - line agns when we isolate their nuclear uv / optical light . in [ secnuc ]
, we found that the nuclear dominated sources are mostly observed as broad - line agns , while the galaxy dominated sources are not . at first glance , this result could be seen as consistent with the dilution hypothesis .
however , if we look more closely at the nuclear uv / optical magnitudes of the sources and compare them with the x - ray fluxes , a different picture emerges .
it is well known ( e.g. , zamorani et al . 1981 ) that there is a strong correlation between the nuclear uv magnitudes and the x - ray fluxes for broad - line agns .
we can see this correlation for the broad - line agns in our sample in figure [ figfs_nuc ] .
however , the correlation does not continue to hold for the narrow - line agns .
in fact , from figure [ figfs_nuc ] , we see that the nuclei of the narrow - line agns are much weaker relative to their x - ray light than would be expected if they were similar to the broad - line agns . in figure [ nuc_flux_z1 ]
, we plot nuclear rest - frame 3000 flux versus rest - frame @xmath10 kev flux .
we restrict to the redshift range @xmath168 , where the deconvolution procedure should be similar for all of the sources to provide a highly uniform sample , but the result holds at all redshifts . in general , the optically - narrow agns fall well below the relation for the broad - line agns , with the high - excitation sources lying closer to the relation , and the absorbers and star formers lying well below the relation , on average . from the above results ,
we are led to conclude that the absence of broad - line agns at low x - ray luminosities is not a dilution effect , but rather that the optically - narrow agns really have weaker uv / optical nuclei relative to the x - rays .
this leaves open the question of whether the lower uv / optical fluxes are a consequence of obscuration ( as would be suggested by figure [ figcol ] and by the observational evidence supporting the unified model ; e.g. , antonucci 1993 ) or a result of intrinsic differences in the source populations ( e.g. , the optically - narrow agns intrinsically have less uv emission and no broad - line region or perhaps are at a different stage in their evolutionary sequence ) . at least having ruled out dilution as a partial explanation for the steffen effect , we can state firmly that the simple unified model has failed because of the existence of the steffen effect .
probably the simplest interpretation of the steffen effect is that the covering fraction of obscuration in the agns is extremely small at high x - ray luminosities ( so all we see at these high luminosities are broad - line agns ) and extremely large ( near unity ) at low x - ray luminosities ( so we do not see broad - line agns at these low luminosities ) .
we shall hereafter refer to this as the luminosity dependent unified model .
if the luminosity depedent unified model is the correct explanation of the steffen effect , then there are important implications for the interpretation of the spectral energy distributions of the x - ray sources ( see [ sechxlum ] ) .
in any unified scheme , where the absorption is due to the orientation of the agn relative to the line of sight , any fir emission powered by the nucleus or by star formation will be emitted isotropically ; thus , the fir / submillimeter properties of the absorbed sources should be the same as those of the unabsorbed sources .
( see barvainas et al .
( 1995 ) for a nice example that illustrates this point . ) for the luminosity dependent unified model , the fir / submillimeter properties of the absorbed and unabsorbed sources should be the same _ at a given total bolometric luminosity_. we assume that the hard x - ray luminosity is an approximate tracer of the total bolometric luminosity for the present sources
. then if the luminosity dependent unified model holds , high x - ray luminosity sources , with their low covering fractions , would not be expected to radiate as much in the fir as low x - ray luminosity sources , with their high covering fractions .
this provides a good prediction that can be tested directly .
one observational study may already argue against the luminosity dependent unified model .
page et al . ( 2001 ) measured 850@xmath169 m fluxes for a sample of eight @xmath138 x - ray absorbed , radio - quiet quasars ( with column densities of @xmath170 based on the x - ray absorption ; these sources show little or no obscuration in their broad emission lines and ultraviolet continua ) discovered serendipitously in archival x - ray data , while page et al .
( 2004 ) measured 850@xmath169 m fluxes for a comparison sample of 20 unabsorbed quasars .
the latter sources were drawn from a combination of x - ray surveys and were matched in luminosity and redshift to the absorbed sample .
four of the eight absorbed sources were detected above 5 mjy at @xmath171 ( the other four were not detected at all ) , while none of the 20 unabsorbed sources were detected above 5 mjy at @xmath171 ( only one of the unabsorbed sources was significantly detected at all , i.e. , at @xmath172 ) . from the level of segregation in the submillimeter properties of
their absorbed and unabsorbed samples , page et al .
( 2004 ) ruled out orientation as the cause of the x - ray absorption .
they proposed that the two types of sources were at different stages in an evolutionary sequence , with the absorbed quasars representing an earlier phase . to put this on a more quantitative footing , we used the published submillimeter fluxes from page et al .
( 2001 , 2004 ) to calculate an error - weighted mean flux of @xmath173 mjy for their unabsorbed sample at a mean soft x - ray luminosity of @xmath174 ergs s@xmath2 , and an error - weighted mean flux of @xmath175 mjy for their absorbed sample at a mean soft x - ray luminosity ( corrected for instrinsic absorption ) of @xmath176 ergs s@xmath2 .
both samples are significantly detected at @xmath172 .
however , in agreement with the page et al .
( 2004 ) conclusions , the mean @xmath177 m flux of the absorbed agns is much larger ( about a factor of 6 ) than that of the unabsorbed agns at roughly the same mean x - ray luminosity .
this is not what would be expected in the luminosity dependent unified model .
however , given the small sample size , larger hard x - ray
selected samples are needed to make the results definitive .
other submillimeter studies have targeted large samples of optically - luminous quasars , and although it is harder to compare with these studies because of the lack of x - ray information , yuan et al .
( 1998 ) have shown that the great majority of optically - selected quasars have little or no x - ray absorption .
thus , these optically - selected samples are most likely similar to the page et al .
( 2001 ) unabsorbed sample in their x - ray properties , and we can try to make a comparison .
priddey et al . ( 2003 ) targeted a homogeneous sample of 57 radio - quiet , optically - luminous quasars in the range @xmath178 , detecting nine @xmath179 mjy sources at greater than @xmath180 significance . using their published submillimeter fluxes
, we calculate an error - weighted mean flux of @xmath181 mjy for the sample . a previous , higher redshift ( @xmath182 ) survey of 38 radio - quiet quasars by isaak et al .
( 2002 ) detected eight @xmath183 mjy sources at greater than @xmath180 sigificance . again using
their published submillimeter fluxes , we calculate an error - weighted mean flux of @xmath184 mjy for the sample . in both of these studies , the mean @xmath177 m fluxes at x - ray luminosities that are probably fairly similar to , or perhaps somewhat higher than ( based on the optical luminosities ) , those of the page et al .
( 2001 , 2004 ) studies turn out to be similar to the mean @xmath177 m flux of the page et al .
( 2004 ) absorbed sample .
this is what would be expected in the luminosity dependent unified model .
thus , it may be premature to unambiguously conclude that the absorbed and unabsorbed sources at these high luminosities are intrinsically different populations .
a major issue for mapping the accretion history of the universe with redshift is knowing what bolometric corrections to use to correct @xmath10 kev luminosities to bolometric luminosities . in the simple unified model , where both the covering fractions and the intrinsic spectral energy distributions are invariant with luminosity ( and redshift )
, it is straightforward to determine the bolometric corrections : since the fir radiation is the absorbed uv through soft x - ray light reradiated isotropically , integrating a broad - line agn spectral energy distribution , taking care to exclude the fir to avoid double counting , would give the bolometric correction .
( note that since the line of sight to the broad - line agn would be unabsorbed , the observed light would account for all of the radiation from the agn . )
an alternative approach would be to take a large , homogeneous sample that covered all possible orientations and sum the spectral energy distributions of all the sources in the sample , regardless of type , to create an average spectral energy distribution that could be integrated .
this would then give the bolometric correction of an unobscured source , which should have the same value as the bolometric correction obtained from the broad - line agn spectral energy distribution .
since we know the bolometric correction obtained by integrating a local broad - line agn spectral energy distribution is approximately 35 ( elvis et al .
1994 ; kuraszkiewicz et al .
2003 ) , if the simple unified model were right , we would already have a determination of the invariant bolometric correction .
however , we now know that the simple unified model is not right , and we need to consider the possibility that the bolometric correction is not invariant and may in fact be spectral type or x - ray luminosity dependent . in [ secenergy ] , we found that a large fraction of the low - redshift , hard x - ray energy density production is in optically - narrow agns , so we really need to determine the bolometric corrections for these sources if we are to map the accretion history with redshift .
given the fact that we do not know what the origin of the steffen effect is ( i.e. , whether the different spectral types are intrinsically different or whether the luminosity dependent unified model is correct ) , in [ secspectype ] and [ sechxlum ] , we construct the composite spectral energy distributions that are needed to calculate the bolometric corrections ( see [ secbolcorrtype ] and [ secbolcorrlum ] ) as a function of both spectral type and x - ray luminosity to cover both possibilities .
either way , we need to include long - wavelength data . in [ seclongwave ] , we describe the long - wavelength imaging data that we have available for our study .
any enhancement in the mir / fir radiation due to absorption and reradiation by gas and dust surrounding the central agn should be directly observable at mir / fir wavelengths .
the observational situation in the mir / fir may be expected to improve rapidly with the _ spitzer space telescope _
( e.g. , alonso - herrero et al .
2004 ; rigby et al . 2004 ; lonsdale et al . 2004 ; dole et al .
however , we may already use observations with the isocam and isophot instruments on the _ iso _ satellite and with the scuba bolometer array on the jcmt to determine the mir and fir / submillimeter properties versus optical spectral type .
there are deep isocam observations of the cdf - n region ( aussel et al .
1999 ) , as well as moderately deep scuba observations over much of the cdf - n region ( wang , cowie , & barger 2004 ) .
the deepest isophot observations are of two lockman hole fields ( kawara et al .
2004 ) , one of which is the clasxs field .
the submillimeter properties of the x - ray sample in the cdf - n are discussed in barger et al .
( 2001c ) and alexander et al .
( 2003a ) using earlier , more restricted scuba observations , while analyses on other fields are discussed in almaini et al .
( 2003 ) and waskett et al .
( 2003 ) . in all of these fields ,
only a small number of x - ray sources are directly detected as submillimeter sources , and , given the poor resolution of the submillimeter observations , some of these may be chance projections .
( a more detailed discussion may be found in wang et al .
2004 , who report that the fraction of @xmath185 mjy and s / n@xmath186 sources with x - ray counterparts in the cdf - n is @xmath187% .
this fraction decreases to @xmath188% once the number of random overlaps has been taken into account . ) however , the error - weighted mean submillimeter fluxes of the ensemble of hard x - ray sources may be measured . for the total ensemble of 61 @xmath189 ergs s@xmath2 sources lying in the submillimeter mapped region of the cdf - n where the 850@xmath169 m flux errors are less than 2.7 mjy
, we find an error - weighted mean 850@xmath169 m flux of @xmath190 mjy , consistent with previous measurements .
if the optically - narrow agns are reprocessing their missing uv / optical light to the mir / fir more so than the broad - line agns , possibly due to these sources being at a dustier stage of the evolutionary sequence than broad - line agns ( if optically - narrow agns are fundamentally different than broad - line agns ) , or possibly due to these , on average , lower @xmath43 sources having a higher covering factor ( if the luminosity dependent unified model holds ) , then most of the above submillimeter signal may be expected to be from the optically - narrow agns .
morever , because of the strong @xmath50-correction , which arises from the steep spectral slope at submillimeter wavelengths , the signal may be expected to be much stronger in the high - redshift sources , where the rest - frame corresponding to 850@xmath169 m lies much closer to the blackbody peak .
we find that this is marginally the case : for the 20 optically - narrow agns at @xmath138 that satisfy the conditions @xmath189 ergs s@xmath2 and 850@xmath169 m flux errors less than 2.7 mjy , we measure an error - weighted mean 850@xmath169 m signal of @xmath191 mjy .
in addition , since the optically - narrow agns are typically hard and the broad - line agns soft , most of the submillimeter signal may also be expected to arise in sources with hard x - ray spectra .
again , this effect is marginally seen .
for the 18 sources with @xmath192 that satisfy the conditions @xmath193 , @xmath189 ergs s@xmath2 , and 850@xmath169 m flux errors less than 2.7 mjy , the error - weighted mean flux is @xmath194 mjy , while for the 21 sources with @xmath195 that satisfy the same conditions , the error - weighted mean flux is @xmath196 mjy . at observed wavelengths of 90@xmath169 m and 170@xmath169 m
, we can compare the clasxs x - ray sample of yang et al .
( 2004 ) with the isophot catalog of kawara et al .
because @xmath197 is decreasing with increasing @xmath198 at these wavelengths , sources fade with increasing redshift , and it is the most nearby sources that will be detected .
we matched the x - ray sources to the @xmath199 m catalog of kawara et al .
( 2004 ) using a @xmath200 search radius . only the two brightest x - ray sources that is , the two @xmath201 ergs @xmath20 s@xmath2 , optically - narrow agns at redshifts @xmath202 and @xmath203are detected in the 90@xmath169 m catalog .
the expected number of chance projections is small : randomized samples show that there is an 8% probability of a chance projection for an individual source .
thus , it is likely that both of these sources are correctly matched , despite the poor resolution of the isophot data .
neither of these sources is detected at 170@xmath169 m .
none of the 12 broad - line agns in the field with @xmath204 ergs @xmath20 s@xmath2 is detected at either 90@xmath169 m or 170@xmath169 m , but this result is not very constraining , given the sensitivity limits at these wavelengths .
figure [ figbolo90 ] shows the @xmath205 bolometric fluxes at 90@xmath169 m for the 23 clasxs sources in the isophot field with @xmath204 ergs @xmath20 s@xmath2 and @xmath3 ergs s@xmath2 .
( we restrict our analysis to the higher x - ray flux sources , where the fir flux limits are more constraining . )
the 21 clasxs sources that do not have a match in the isophot catalog are shown at the @xmath180 upper limit of 60 mjy for the field .
if the @xmath206 to @xmath207 ratio were similar to that of the two brightest sources , then we would not expect to be able to detect most of the remaining sources , as can be seen by the solid line . at observed 15@xmath169 m wavelengths
, we can compare the cdf - n x - ray sample of alexander et al .
( 2003b ) with the isocam catalog of aussel et al .
( 1999 ; see also alexander et al .
there are six x - ray sources in the well - covered central region of the isocam image that have @xmath208 ergs @xmath20 s@xmath2 and @xmath3 ergs s@xmath2 , where again we restrict to the brightest x - ray sources , which are more likely to be detected .
all six are optically - narrow agns .
four of the six match to 15@xmath169 m sources in the aussel et al .
( 1999 ) catalog using a @xmath58 search radius .
we have generated randomized samples by placing the same number of sources at arbitrary positions within this region and running the same matching procedure . using a large number of such samples , we find that there is only a 20% probability that any identification is a chance projection .
for the broad - line agns , the strong linear correlation ( see [ secfailure ] ) of the nuclear rest - frame uv / optical fluxes with the rest - frame @xmath10 kev fluxes enables us to determine the mean uv and optical spectral energy distributions of these objects . in figure
[ nuc_broad ] , we show the correlation between the two fluxes at rest - frame wavelengths of 3000 , 2000 , and 1200 . the fits give ratios of @xmath209 , 4.7 , and 4.0 , respectively , at these wavelengths , which corresponds to @xmath197 scaling roughly as @xmath210 .
this is similar to the composite quasar spectra of vanden berk et al .
( 2001 ) and zheng et al .
( 1997 ) , which are shown in figure [ figbolo_opt ] .
individual sources scatter by approximately a multiplicative factor of 3 about these relations , as can be seen in figure [ nuc_broad ] .
we can also plot for each source the ratio of the nuclear rest - frame uv / optical flux to the rest - frame @xmath10 kev flux at the rest - frame frequencies corresponding to the four goods bands . since the sources lie at a range of redshifts , the individual measurements spread over a wide range of frequencies . as long as we assume that there is no redshift dependent evolution of the spectral energy distributions ,
we can determine for the various spectral classes the composite spectral energy distributions over the range @xmath211 normalized to the x - ray flux . in figure
[ figbolo_opt ] , we show these composite spectral energy distributions for three spectral classes ( broad - line agns , high - excitation sources , and normal galaxies ) .
these plots show the shape of the spectral energy distributions relative to the hard x - ray emission .
the sources in the broad - line agn and high - excitation classes all have uv / optical nuclei , while many of the sources in the normal galaxy class show no signs of having a uv / optical nucleus at all .
sources that have no measurable nuclei are assigned a nominal magnitude of 29.5 .
consistent with figure [ nuc_broad ] , the broad - line agns exhibit only a small amount of scatter ( roughly a factor of 3 ) about a mean spectral energy distribution that closely matches to composite quasar spectra ( zheng et al .
1997 ; vanden berk et al .
. the ratios of the rest - frame uv / optical luminosities to the rest - frame @xmath10 kev luminosities match those of the local hard x - ray broad - line agn sample analyzed by kuraszkiewicz et al .
( 2003 ) .
in contrast , the high - excitation sources scatter to much fainter uv / optical values than do the broad - line agns , and the effect is even more extreme for the normal galaxies . mean normalized fluxes versus frequency are summarized in table [ tab3 ] for each class ( broad - line agns , high - excitation sources , and normal galaxies ) and are shown in figure [ mean_bolo ] .
numbers are only given if there are more than five sources in a frequency interval .
the values for the high - excitation and normal galaxy classes drop to less than half that of the broad - line agn class at @xmath212 hz and continue to fall rapidly at higher frequencies . these results , and
the opacity ranges seen in the x - ray color - color plots of figure [ figcol ] , show that , on average , the optically - narrow agns ( high - excitation sources and normal galaxies ) are highly suppressed in the frequency range between the uv and soft x - rays ( approximately @xmath213 hz to @xmath214 hz ) relative to the broad - line agns .
ccccccc @xmath215 & & & 5.14 & & 3.62 & 5.90 @xmath216 & 5.55 & 2.13 & 2.47 & 4.02 & 1.85 & 3.28 @xmath217 & 3.48 & 1.12 & 0.664 & 2.03 & 1.64 & 0.737 @xmath218 & 4.37 & 0.588 & 0.614 & 4.03 & 1.43 & 0.511 [ tab3 ] lccr 0.4 - 1.0 & 1 & @xmath219 & @xmath220 1.0 - 2.0 & 3 & @xmath221 & @xmath222 2.0 - 4.0 & 5 & @xmath223 & @xmath224 0.4 - 1.0 & 30 & @xmath225 & @xmath226 1.0 - 2.0 & 16 & @xmath227 & @xmath228 2.0 - 4.0 & 4 & @xmath229 & @xmath230 [ tabbolosmm ] using our long - wavelength data , we may now construct composite spectral energy distributions of both the broad - line agns and the optically - narrow agns . in figure [ figbolo_all ] , we show the mean bolometric luminosities ( relative to the x - ray luminosities ) versus the rest - frame frequencies for the ( a ) broad - line agns and ( b ) optically - narrow agns .
the dashed curve that appears in both plots shows the median value for the local hard x - ray selected sample of kuraszkiewicz et al .
( 2003 ) , which is primarily broad - line agns with a small admixture of intermediate - type seyferts . from the scuba data ( see [ secsmm ] ) , we were able to compute the ratio of the bolometric luminosity , @xmath231 , to @xmath43 over most of the 100@xmath169 m to 1 mm wavelength range . to do this , we divided all of the agns by redshift bin and by spectral class ( broad - line agns and optically - narrow agns ) and then computed their mean ratios @xmath232 , weighted by the square of the submillimeter error divided by @xmath43 .
the results are given in table [ tabbolosmm ] and plotted in figure [ figbolo_all ] .
the optically - narrow agns show a strong rise with increasing frequency , which is consistent with a strong fir component in these sources , while the broad - line agns have only upper limits or weak signals , which is consistent with a flat spectral energy distribution .
the higher redshift ( frequency ) results are based on a small number of high @xmath43 sources and so may not be representative of the lower x - ray luminosity sources . in the fir ( see [ secfir ] ) , we used the two x - ray sources in the clasxs field that were detected at 90@xmath169 m to estimate a ratio of bolometric luminosity to rest - frame @xmath10 kev luminosity of @xmath233 _
( open diamond ) _ for the optically - narrow agns ( figure [ figbolo_all]b ) .
the second open diamond in the figure shows the @xmath180 upper limit at 170@xmath169 m , which is based on the non - detections of the two x - ray sources in the kawara et al .
( 2004 ) catalog . for the broad - line agns ( figure [ figbolo_all]a ) , we show the @xmath180 upper limits at both 90@xmath169 m and 170@xmath169 m , which are based on the non - detections of the x - ray bright broad - line agns in the kawara et al .
( 2004 ) catalog . in the mir
( see [ secmir ] ) , the ratio of the bolometric luminosity to the rest - frame @xmath10 kev luminosity is 6 _ ( large solid square ) _ for the six isocam detected optically - narrow agns ( figure [ figbolo_all]b ) .
the six individual points are shown as dots .
no mir point is shown in figure [ figbolo_all]a .
various bolometric corrections have been adopted for the optically - narrow agns in the literature .
barger et al . ( 2001b ) adopted a value of about 35 , based on the radio and submillimeter properties of a _
chandra _ x - ray sample , while fabian ( 2004 ) suggested a lower value of about 15 , based on the suppression of uv and optical light in the local seyfert galaxies mrk335 , ngc3783 , and ngc5548 .
however , the latter determination neglects the enhancement of the mir / fir light that results from the reradiation of the absorbed uv / optical light by gas and dust , which we can see from figure [ figbolo_all]b is important . in fact , the composite spectral energy distributions in figure [ figbolo_all ] show that the broad - line agns are consistent with the locally derived results though poorly constrained at the intermediate wavelengths while the optically - narrow agns are clearly much more strongly peaked in the fir . to illustrate this , in figure [ figbolo_all]b
, we have overplotted the highly obscured local galaxy ngc6240 ( hasinger 2000 ) , which is often used as a template for absorbed agns .
this template provides a fairly good description of the shape of the composite spectrum over the frequency range from the submillimeter to the uv . by normalizing the shape of the template to the data _ ( solid curve ) _ , we would infer a bolometric correction of about 85 for the optically - narrow sources , with most of the light emerging around 100@xmath169 m in the rest - frame .
this inferred bolometric correction is larger than that determined for the broad - line agns . as we can see from the dot - dashed curve in figure [ figbolo_all]b , a bolometric correction of 35 would be too low to fit the optically - narrow agn spectral energy distribution .
part of this effect is due to internal absorption in the @xmath10 kev band , which is higher for the optically - narrow sources and negligible for the broad - line agns . at @xmath153
, we can use the @xmath77 kev and @xmath76 kev counts to estimate a mean correction of about 1.3 for the @xmath10 kev flux in the optically - narrow sources ( see figure [ figcol ] ) . however , even with this correction , the bolometric correction inferred for the optically - narrow agns is higher than that for the broad - line agns .
one additional concern is that the fir emission from these sources may be partially due to star formation , though the correlation of @xmath43 and @xmath234 would seem to suggest that @xmath234 is related to the nuclear activity ( or that star formation is correlated with the nuclear activity ) .
given the limited mir / fir / submillimeter data sets , there are clearly considerable uncertainties in our determinations
. however , the data do suggest that it is unlikely that the bolometric corrections for the optically - narrow agns are significantly lower than the bolometric corrections for the broad - line agns , as was proposed by fabian ( 2004 ) .
if the luminosity dependent unified model holds , such that the covering fraction varies with x - ray luminosity , then the low x - ray luminosity sources with their high covering fractions should have spectral energy distributions that are , on average , suppressed in the uv to the soft x - rays with the light reradiated in the fir . in contrast , the high x - ray luminosity sources with their low covering fractions should have little reradiation . in figure [ figbolo_lx ]
, we show the uv / optical spectral energy distributions according to x - ray luminosity rather than according to spectral type .
we again summarize the mean normalized fluxes versus frequency in table [ tab3 ] , this time for each luminosity interval .
numbers are only given if there are more than five sources in a given frequency interval . at high x - ray luminosities ,
most of the sources lie close to the unobscured quasar spectral energy distribution , and the mean uv / optical to x - ray flux ratios are only slightly lower than the broad - line agn values ( table [ tab3 ] ) .
in contrast , for the low and intermediate x - ray luminosity sources , there is considerably more spread in the individual uv / optical to x - ray flux ratios , with the mean ratios lying well below the broad - line agn values at the higher frequencies .
thus , as expected , the lowest x - ray luminosity sources are the most suppressed .
this behavior is similar to that seen in samples divided by optical luminosity , where the extinctions are higher at lower luminosities ( e.g. , gaskell et al .
the same reference also shows that there may be extinction without spectral curvature , a point which emphasizes the necessity of looking at the full spectral energy distributions of the sources in deciding the bolometric corrections .
this is just another statement of the steffen effect , though one which is completely independent of the optical spectral typing : low x - ray luminosity sources have much fainter uv / optical nuclei relative to their @xmath10 kev fluxes than do high x - ray luminosity sources .
in the luminosity dependent unified model , the covering fraction of the sources in a given luminosity range is equal to the fraction of non broad - line agns in that luminosity range .
if we assume that the bolometric corrections are not dependent on x - ray luminosity , then we can predict how much luminosity should be reradiated into the fir at a given x - ray luminosity : the fir luminosity should just be the bolometric luminosity times the covering fraction .
this simple model predicts that the brightest fir sources should be the intermediate x - ray luminosity sources .
higher x - ray luminosity sources have small covering fractions and do not reradiate much light into the fir , while lower x - ray luminosity sources have low bolometric luminosities to start off with .
we illustrate this in figure [ lum_model ] , where we show the error - weighted mean submillimeter fluxes ( converted to fir luminosities , see below ) of the ensemble of x - ray sources in the redshift range @xmath235 versus their mean hard x - ray luminosities _ ( diamonds)_. we convert the submillimeter fluxes to fir luminosities assuming a fir spectrum with the shape of ngc6240 ( hasinger 2000 ) .
the solid and dashed curves show the predicted values for the luminosity dependent unified model , assuming a bolometric correction of 35 .
the solid curve was calculated using the spectroscopically identified sources only , and the dashed curve was calculated by placing all of the spectroscopically unidentified sources in the spectroscopically observed sample into the redshift interval as non broad - line agns , which gives a very extreme upper limit . the dotted curve ( shown only for the spectroscopically identified sources ) shows the same calculation for a bolometric correction of 85 ( see [ secbolcorrtype ] ) , which provides a better fit to the data at the low - luminosity end . at the high - luminosity end
, we would expect the bolometric correction of 35 to be valid , because we know that value is appropriate for broad - line agns .
thus , the data appear to be consistent with the luminosity dependent unified model , but two caveats should be kept in mind .
first , the errors on the submillimeter data are significant , and the consistency of the results with the model is not highly significant . also , bearing in mind the page et al .
( 2001 , 2004 ) results for the brighter quasars , better data are clearly required .
second , some part of the fir luminosity could be produced by star formation , which may in turn be correlated with the nuclear activity .
as was first pointed out by sotan ( 1982 ) , we may use the evolution of the parameterization of the rest - frame @xmath10 kev energy density production rate per comoving volume , @xmath135 , to infer the present - day supermassive black hole mass density accreted by agns , @xmath236 . the mass inflow rate onto a supermassive black hole , @xmath237 ,
is related to the bolometric luminosity of the agn , @xmath238 , by @xmath239 , where @xmath240 is the accretion efficiency ( typically taken to be about 0.1 ) .
assuming @xmath8 , we can use this to go from @xmath135 to the accretion rate density , @xmath236 , if we know the corrections to bolometric luminosity ( see , e.g. , barger et al .
2001a ) . in the present section
, we set @xmath241 , where @xmath242 is the properly averaged bolometric correction in the @xmath10 kev band for a given population . recognizing the uncertainties
, we use our determinations of @xmath242 .
we do not make any correction for internal absorption in the @xmath10 kev band , since this is present in both the hard x - ray luminosity functions and the bolometric corrections .
this could result in some redshift dependent correction ; in particular , we may be slightly underestimating the energy density production rate at @xmath14 relative to that at @xmath13 , since we are selecting in a higher energy band at the higher redshifts .
however , we do not expect this effect to be large , given that the overall correction for absorption is not large .
our parameterization of the production rate for all spectral types is @xmath243 where @xmath244 ergs s@xmath2 mpc@xmath147 , and @xmath245 at @xmath139 and @xmath246 at higher redshifts .
the integral of equation [ bargereq ] is only weakly sensitive to the poorly determined @xmath73 at the higher redshifts , since it is dominated by the production rate at @xmath13 ( see [ secenergy ] ) .
thus , we have chosen our value of @xmath73 at @xmath138 so that the parameterization is between the measured spectroscopic values and the incompleteness corrected values of figure [ figevol ] . at low redshifts ,
roughly 40% of the hard x - ray energy density production rate is due to broad - line agns and the remainder to optically - narrow agns . assuming this constant ratio to determine what fraction of each bolometric correction to use , and integrating through time
, we find the accreted supermassive black hole mass density for all spectral types to be @xmath247 if we use @xmath248 for the optically - narrow agns , and @xmath249 if we use @xmath250 for the optically - narrow agns .
( we always use @xmath250 for the broad - line agns . ) making the energy density production rate history flat ( @xmath251 ) at higher redshifts ( i.e. , using the incompleteness correction of figure [ figevol ] ) would increase these numbers by at most 40% .
these numbers do not depend on the hubble constant and are insensitive to the assumed cosmological geometry ( e.g. , chokshi & turner 1992 ) .
the accreted supermassive black hole mass density from the broad - line agns may be more exactly calculated , since the bolometric correction is better known , and the history of the energy production should be less susceptible to the incompleteness corrections .
the ratio of the hard x - ray energy density production rate due to broad - line agns relative to that due to optically - narrow agns is clearly higher at higher redshifts ( figure [ figevolblagn ] ) .
thus , we parameterize the rest - frame @xmath10 kev comoving energy density production rate of the broad - line agns alone as @xmath252 where @xmath253 ergs mpc@xmath147 s@xmath2 , and @xmath254 at @xmath139 and @xmath255 at higher redshifts . this fit is shown in figure [ figevolblagn ] . integrating equation [ bargereq2 ]
, we obtain @xmath256~{\rm m}_\odot~{\rm mpc}^{-3 } \
, , \label{bargereq3}\ ] ] which would say that about one - half to one - quarter of the supermassive black hole mass density was fabricated in broad - line agns and the remainder in optically - narrow agns . with the energy density production rate of equation [ bargereq ] ,
half of the accreted supermassive black hole mass density has formed by @xmath257 ( @xmath258 yr ) and the remaining amount after this redshift .
the broad - line agns ( which are the high x - ray luminosity population , and hence correspond to the more massive black holes ) form slightly earlier , with half the mass density in place by @xmath259 .
for a constant bolometric correction , regardless of type or luminosity , roughly half of the supermassive black hole mass density is formed in x - ray sources with @xmath44 ergs s@xmath2 to @xmath260 ergs s@xmath2 and half from more luminous sources . using equation [ eq4 ] , this would correspond to a black hole mass of about @xmath261 .
if the bolometeric correction is larger for the lower luminosity sources , then this would weight the result to a lower break point .
the broad - line agn supermassive black hole mass density calculation should also provide the most precise comparison to previous optical determinations , since these have been computed only for the broad - line agn population .
yu & tremaine ( 2002 ) determined the accreted supermassive black hole mass density during optically - bright quasar phases to be @xmath262~{\rm m}_\odot~{\rm mpc}^{-3 } \,,\ ] ] where @xmath263 is the bolometric correction in the @xmath82 band , defined by @xmath264 and found to be about 11.8 by elvis et al .
this value of @xmath265 is similar to that derived by chokshi & turner ( 1992 ) .
we can see that the yu & tremaine ( 2002 ) calculation is about 1.75 times higher than our value .
the reason for this difference lies in the extrapolation by yu & tremaine of the optical qso luminosity function determinations outside the measured absolute magnitude range .
this extrapolation is quite sensitive to the poorly determined shape of the faint - end optical qso luminosity function .
yu & tremaine used the optical qso luminosity function of boyle et al .
( 2000 ) , which found a faint - end slope of @xmath266 , whereas a more recent determination by croom et al .
( 2004 ) finds a shallower slope of @xmath267 .
thus , mass density estimates based on the croom et al .
paper will be lower than those obtained by yu & tremaine . however , even the croom et al .
extrapolation does not show the downturn seen in the hard x - ray luminosity functions at low luminosities . in figure [ lum_den_opt ] , we use the new croom et al .
( 2004 ) qso luminosity function determination to show the effect of the extrapolation outside the measured absolute magnitude range .
the solid curve shows the bolometric energy density production determined directly from the croom et al .
data and the dotted curve the redshift extrapolation of this determination .
the corresponding results determined from our x - ray selected population are shown by the diamonds and dashed curve .
we can see that the restriction of the measured optical qso luminosity function to absolute magnitudes @xmath46 to @xmath125 means that just using the croom et al .
fit underestimates the broad - line agn light production at @xmath139 , while above this redshift , where the light is dominated by higher luminosity sources , the two measurements agree .
extrapolating the croom et al .
data outside the fitted absolute magnitude range _ ( dot - dashed curve ) _ improves the agreement at the low - redshift end , where much of the light is in low - luminosity sources .
however , at the high - redshift end , this extrapolation results in an overestimation , because the faint - end extrapolation is flatter than the hard x - ray luminosity function determinations , which turn down at the low - luminosity end ( see figure [ blagn_opt ] ) . in terms of the effect that the extrapolation has on the accreted supermassive black hole mass density from optical qsos ,
if we only include the sources in the @xmath268 to @xmath125 range and use the croom et al .
( 2004 ) fit , we obtain @xmath269~{\rm m}_\odot~{\rm mpc}^{-3 } \,,\ ] ] while if we extrapolate to brighter and fainter luminosities using the croom et al .
fit , we instead find @xmath270~{\rm m}_\odot~{\rm mpc}^{-3 } \,.\ ] ] these values bracket our broad - line agn hard x - ray luminosity function determination . in conclusion ,
whether one uses the recent optical qso luminosity function determinations of croom et al .
( 2004 ) or our broad - line agn hard x - ray luminosity function determinations , assuming @xmath8 , both give about half the accreted supermassive black hole mass density found by yu & tremaine ( 2002 ) .
this means that there is still room for obscured accretion from the optically - narrow agns when we compare with the local supermassive black hole mass density .
in fact , for @xmath9 , we find reasonable agreement between our accreted supermassive black hole mass density for all spectral types ( equation [ eqall1 ] or [ eqall2 ] ) and the local supermassive black hole mass density determined by yu & tremaine ( 2002 ) : @xmath271 however , we note that there is very little room for further obscured sources or for any low efficiency accretion periods .
in this paper , we presented a thorough analysis of the nature and evolution of hard x - ray selected agns using both deep and wide - area _ chandra _
surveys with nearly complete optical spectroscopic follow - up observations .
our results are as follows .
@xmath272 we determined median redshifts and @xmath64 median redshift ranges with x - ray flux for the spectroscopically and spectroscopically plus photometrically identified x - ray samples .
we found that the median redshifts are fairly constant with x - ray flux at @xmath0 and that the lower x - ray flux sources , where the surveys are more spectroscopically incomplete , are dominated by sources with hard x - ray luminosities below @xmath273 ergs s@xmath2 and redshifts near one .
@xmath272 we spectrally classified the optical counterparts to the x - ray sources and measured the fwhm line widths of the spectra .
we found that the x - ray selected broad - line agns recover essentially all of the optically - selected type 1 agns and that the only distinction that can be made from the x - ray colors is between the broad - line agns and the sources that do not show broad lines ( @xmath1 km s@xmath2 ; the `` optically - narrow '' agns ) .
most of the broad - line agns show essentially no visible absorption in x - rays , while the optically - narrow agns show a wide range of absorbing column densities .
however , these absorbing column densities show little dependence on the optical spectral type or on the line widths of the optically - narrow agns .
@xmath272 we constructed up - to - date low and high - redshift hard x - ray luminosity functions for all spectral types and for broad - line agns alone using our highly spectroscopically complete observed samples .
we did maximum likelihood fits over the redshift range @xmath12 .
we found that all of the hard x - ray luminosity functions are consistent with pure luminosity evolution , with @xmath274 for all spectral types and @xmath275 for broad - line agns .
thus , all agns drop in luminosity by almost an order of magnitude over this redshift range .
@xmath272 we directly compared our broad - line agn hard x - ray luminosity functions with the optical qso luminosity functions and found that at the bright end , the luminosity functions agreed extremely well at all redshifts .
however , we found that the optical qso luminosity functions do not probe faint enough to see the downturn in the broad - line agn hard x - ray luminosity functions and that they may even be missing some sources at the very lowest luminosities to which they probe .
@xmath272 we determined the evolution with redshift of the hard x - ray energy density production rate and found that at @xmath139 , it bears a striking resemblance to the overall redshift evolution of the star formation rate density .
we also determined the evolution according to spectral type , applying incompleteness corrections to the production rates of the optically - narrow agns .
we found that at @xmath150 , most of the hard x - ray energy density production is due to optically - narrow agns , while at higher redshifts , the broad - line agns dominate the production .
@xmath272 we estimated black hole masses for the broad - line agns using our measured mgii 2800 line widths and nuclear optical luminosities of sources in the _ hst _ acs goods - north region of the cdf - n . for the resulting small sample , we found a surprisingly tight correlation between hard x - ray luminosity and black hole mass at @xmath13 , with the black holes accreting at a fairly constant rate of 16% of the eddington luminosity .
@xmath272 from the x - ray sources in the acs goods - north region of the cdf - n , we found that the most luminous x - ray sources lie in the most optically - luminous host galaxies , as would be expected from the bulge luminosity black hole mass relation .
we also found that the host galaxy optical luminosities do not change much with redshift for a given interval in hard x - ray luminosity , but the most optically - luminous host galaxies that existed at high redshifts have completely disappered at low redshifts .
we therefore concluded that the observed drop in the hard x - ray luminosities is due to agn downsizing rather than to an evolution in the accretion rates onto the supermassive black holes .
that is , the highest mass supermassive black holes in the most optically - luminous host galaxies are switching off between @xmath13 and the present time , leaving only the lower mass supermassive black holes in the less optically - luminous galaxies active .
@xmath272 a primary aim of the paper was to explore the origin of the steffen effect , which is the observation that broad - line agns dominate the number densities at the higher x - ray luminosities , while non broad - line agns dominate at the lower x - ray luminosities .
we ruled out galaxy dilution as being a partial explanation of the steffen effect by seeing how the nuclear uv / optical properties of the x - ray sources in the cdf - n ( measured from the acs goods - north data ) relate to the spectral characteristics of the sources .
we found that the uv / optical nuclei of the optically - narrow agns are much weaker than expected if the optically - narrow agns were similar to the broad - line agns .
since the simple unified model can not explain the steffen effect , we postulated that the simplest interpretation of the steffen effect is a luminosity dependent unified model .
another possible interpretation is that the broad - line agns and the optically - narrow agns are intrinsically different . regardless of which interpretation is correct , a large fraction of the low - redshift , hard x - ray energy density production is in optically - narrow agns ; thus , bolometric corrections for these sources are needed if we are to map the accretion history with redshift .
@xmath272 to infer the bolometric corrections , we constructed composite spectral energy distributions ( including long - wavelength data from the mid - infrared to the submillimeter ) as a function of both spectral type and x - ray luminosity ( to cover both interpretations of the steffen effect ) .
in contrast to the broad - line agns , we found that the optically - narrow agns are , on average , highly suppressed in the frequency range between the uv and soft x - rays and have a strong far - infrared component .
we inferred a bolometric correction of about 85 for these sources .
likewise , we found that the low x - ray luminosity sources have much fainter uv / optical nuclei relative to their @xmath10 kev fluxes than do high x - ray luminosity sources .
the far - infrared luminosities of these low x - ray luminosity sources are also consistent with a bolometric correction of 85 .
@xmath272 using our bolometric corrections and our parameterizations of the evolution of the rest - frame hard x - ray energy density production rates , we inferred the accreted supermassive black hole mass densities for all spectral types and for broad - line agns alone .
we found that only about one - half to one - quarter of the supermassive black hole mass density was fabricated in broad - line agns .
@xmath272 we compared our accreted supermassive black hole mass density for broad - line agns alone with the optical qso determination made by yu & tremaine ( 2002 ) using the boyle et al .
( 2000 ) qso luminosity function , assuming @xmath8 .
our value is 1.75 times smaller than their value , but this difference is very sensitive to their extrapolation of the optical qso luminosity function outside the optically observed luminosity range .
the new croom et al .
( 2004 ) qso luminosity function has a shallower faint - end slope than boyle et al .
we recalculated the accreted supermassive black hole mass densities for optical qsos using the croom et al .
fit , first by restricting to the measured absolute magnitude range , and second by extrapolating the fit to higher and lower luminosities , and we found that the two values bracketed our hard x - ray luminosity function determination .
thus , both the new optical qso luminosity function and the hard x - ray luminosity function determinations of the supermassive black hole mass densities are a factor of almost two lower than the yu & tremaine ( 2002 ) determination .
this leaves room for obscured accretion by the optically - narrow agns in a comparison with the local supermassive black hole mass density .
@xmath272 for @xmath9 , we found reasonable agreement between our hard x - ray luminosity function supermassive black hole mass density for all spectral types and the local supermassive black hole mass density ; however , there is little room for further obscured sources or for any low efficiency accretion periods .
we thank the referee , robert antonucci , for helpful comments that improved the manuscript .
we thank qingjuan yu and scott tremaine for helpful conversations concerning their paper .
we gratefully acknowledge support provided by nasa through grants hst - go-09425.30-a ( a. j. b. ) and hst - go-09425.03-a ( l. l. c. ) from the space telescope science institute , which is operated by the association of universities for research in astronomy , incorporated , under nasa contract nas5 - 26555 .
we also gratefully acknowledge support from nsf grants ast-0084847 and ast-0239425 ( a. j. b. ) and ast-0084816 ( l. l. c. ) , nasa cxc grants go2 - 3191a ( a. j. b. ) and df1 - 2001x and go2 - 3187b ( l. l. c. ) , the university of wisconsin research committee with funds granted by the wisconsin alumni research foundation , the alfred p. sloan foundation , and the david and lucile packard foundation ( a. j. b. ) , the ids program of r. f. m. , and nasa s national space grant college and fellowship program and the wisconsin space grant consortium ( a. t. s. ) .
alexander , d. m. , aussel , h. , bauer , f. e. , brandt , w. n. , hornschemeier , a. e. , vignali , c. , garmire , g. p. , & schneider , d. p. 2002
, , 568 , l85 alexander , d. m. , et al .
2003a , , 125 , 383 alexander , d. m. , et al .
2003b , , 126 , 539 almaini , o. , lawrence , a. , & boyle , b. j. 1999 , , 305 , l59 | we use highly spectroscopically complete deep and wide - area _
chandra _ surveys to determine the cosmic evolution of hard x - ray selected agns . for the deep fields ,
we supplement the spectroscopic redshifts with photometric redshifts to assess where the unidentified sources are likely to lie .
we find that the median redshifts are fairly constant with x - ray flux at @xmath0 .
we classify the optical spectra and measure the fwhm line widths .
most of the broad - line agns show essentially no visible absorption in x - rays , while the sources without broad lines ( @xmath1 km s@xmath2 ; `` optically - narrow '' agns ) show a wide range of absorbing column densities .
we determine hard x - ray luminosity functions for all spectral types with @xmath3 ergs s@xmath2 and for broad - line agns alone . at @xmath4 , both are well described by pure luminosity evolution , with @xmath5 evolving as @xmath6 for all spectral types and as @xmath7 for broad - line agns alone .
thus , all agns drop in luminosity by almost an order of magnitude over this redshift range .
we show that this observed drop is due to agn downsizing rather than to an evolution in the accretion rates onto the supermassive black holes .
we directly compare our broad - line agn hard x - ray luminosity functions with the optical qso luminosity functions and find that at the bright end they agree extremely well at all redshifts .
however , the optical qso luminosity functions do not probe faint enough to see the downturn in the broad - line agn hard x - ray luminosity functions and even appear to be missing some sources at the lowest luminosities they probe .
we find that broad - line agns dominate the number densities at the higher x - ray luminosities , while optically - narrow agns dominate at the lower x - ray luminosities .
we rule out galaxy dilution as a partial explanation for this effect by measuring the nuclear uv / optical properties of the _ chandra _ sources using the _ hst _ acs goods - north data .
the uv / optical nuclei of the optically - narrow agns are much weaker than expected if the optically - narrow agns were similar to the broad - line agns .
we therefore postulate the need for a luminosity dependent unified model .
an alternative possibility is that the broad - line agns and the optically - narrow agns are intrinsically different source populations .
we cover both interpretations by constructing composite spectral energy distributions including long - wavelength data from the mid - infrared to the submillimeter by spectral type and by x - ray luminosity .
we use these spectral energy distributions to infer the bolometric corrections ( from hard x - ray luminosities to bolometric luminosities ) needed to map the accretion history .
we determine the accreted supermassive black hole mass density for all spectral types and for broad - line agns alone using the observed evolution of the hard x - ray energy density production rate and our inferred bolometric corrections .
we find that only about one - half to one - quarter of the supermassive black hole mass density was fabricated in broad - line agns .
using either recent optical qso luminosity function determinations or our broad - line agn hard x - ray luminosity function determinations , we measure an accreted supermassive black hole mass density that is a factor of almost two lower than that measured by previous work , assuming @xmath8 .
this leaves room for the obscured accretion when compared with the local supermassive black hole mass density .
in fact , we find reasonable agreement between the accreted supermassive black hole mass density from all spectral types and the local supermassive black hole mass density , assuming @xmath9 .
however , there is very little room for further obscured sources or for any low efficiency accretion periods . |
a smooth two dimensional projective variety @xmath0 defined over a field @xmath1 is called a _ del pezzo _
surface if its anticanonical divisor @xmath2 is ample .
the _ degree _ @xmath3 of a del pezzo surface is the self - intersection number of its canonical class and is always between 1 and 9 . over algebraically closed fields
, del pezzo surfaces admit a description that is simple to understand and is adequate for the purposes of this paper .
* iv.24.4 ) ( * ? ? ?
* thm 1 ) .
let @xmath4 be a del pezzo surface of degree @xmath3 defined over an algebraically closed field .
* @xmath5 is isomorphic to @xmath6 .
* @xmath7 is isomorphic to @xmath8 or to the blow - up of @xmath6 at one point . * if @xmath9 , then @xmath4 is the blow - up of @xmath6 at @xmath10 points , no three of which lie on a line , no six of which lie on a conic , and no eight lie on a singular cubic having one of the points as the singularity . because any four points in @xmath6 can be moved to any four other points via an automorphism of @xmath6 , for @xmath11 ,
there is only one del pezzo surface of that degree up to isomorphism ( * ? ? ?
* iv.24.4.1 ) .
thus , from a moduli theoretic point of view , the most interesting del pezzo surfaces are those of degree less than five
. over arbitrary fields these del pezzo surfaces may not be the blow - up of points in @xmath6 , so we need a simple description that does not require the field @xmath1 to be algebraically closed .
such a description is provided by kollr ( * ? ? ? * iii.3.5 ) .
let @xmath4 be a degree @xmath3 del pezzo surface defined over an arbitrary field @xmath1 .
* @xmath12 is isomorphic to the intersection of two quadrics in @xmath13 .
* @xmath14 is isomorphic to a cubic surface in @xmath15 .
* @xmath16 is isomorphic to a hypersurface of degree four in weighted projective space @xmath17 .
* @xmath18 is isomorphic to a hypersurface of degree six in weighted projective space @xmath19 .
del pezzo surfaces contain a finite number of surprisingly rigid curves that encode much of their geometry .
they are called _ exceptional curves _ and are defined as irreducible genus zero curves with self - intersection @xmath20 , and they are the only irreducible curves on @xmath4 with negative self - intersection . a surface @xmath4 is called _ split _ over the field @xmath1 if it can be realized as the blow - up of @xmath6 at @xmath10 points .
split surfaces are of interest to us because when a surface @xmath4 is split , all of its exceptional curves are defined over @xmath1 .
the exceptional curves on @xmath4 can easily be described when @xmath21 .
let @xmath22 be the blow - up of the plane at the points @xmath23 with @xmath24 and let @xmath25 be an exceptional curve .
the image @xmath26 in @xmath6 is one of the following : * one of the points @xmath27 ; * a line passing through two of the points @xmath27 ; * a conic passing through five of the points @xmath27 ; * a cubic passing through seven of the points @xmath27 such that one of them is a double point . the number of exceptional curves on @xmath4 is given in the following table : [ cols="<,<,<,<,<,<,<",options="header " , ] using the table above and an in depth analysis of cubic surfaces , hirschfeld classified the full del pezzo surfaces of degree nine through three @xcite .
let @xmath4 be a split del pezzo surface of degree @xmath3 defined over a finite field @xmath28 .
* @xmath4 is never full when @xmath29 .
* @xmath30 is full if and only if @xmath31 or @xmath32 .
* @xmath12 is full if and only if @xmath33 . *
full cubic surfaces exist only over the fields @xmath34 and @xmath35 .
this paper gives a partial classification of full degree two del pezzo surfaces over fields of odd characteristic .
let @xmath16 be a full degree two del pezzo surface over @xmath28 where @xmath36 is odd .
then @xmath37 and up to isomorphism : * there is a unique @xmath16 defined over @xmath38 and @xmath39 ; * there are two @xmath16 s defined over @xmath40 ; * there are six @xmath16 s defined over @xmath41 ; * there are five @xmath16 s defined over @xmath42 ; * there are at least two @xmath16 s defined over @xmath43
. theoretically there can exist full degree two del pezzo surfaces over @xmath44 and @xmath45 , but we have not found them yet .
+ + * acknowledgments : * the authors are greatly indebted to the referee for writing a magma program that helped us analyze the bitangents to the curves given by kaplan .
we would also like to thank robert lazarsfeld for introducing them at the university of michigan and zachary scherr for discussing this problem with us .
from now on we will work over a finite field of odd characteristic , @xmath28 . as stated above
, a split degree two del pezzo surface @xmath16 can be thought of as the blow - up of the plane at seven points in general position or a degree four hypersurface in weighted projective space @xmath17 .
the weighted projective space we are considering has variables @xmath46 where the @xmath47 is given weight two and the other variables are weight one .
this means for our surface is that we can write @xmath16 as the zero set of an equation of the form @xmath48 where @xmath49 is a homogeneous degree four polynomial in three variables with coefficients in @xmath28 .
this representation of @xmath16 yields a third way of thinking of degree two del pezzo surfaces .
the polynomial @xmath50 defines a quartic curve @xmath51 in the plane @xmath52 and @xmath16 is a double cover of the plane branched over @xmath51 .
thus , there is a surjective map @xmath53
such that the preimage of a point in the plane is two points in @xmath16 unless the point is contained in the quartic @xmath51 , in which case the preimage is one point .
the map @xmath54 is the rational map given by the anticanonical linear system , @xmath55 , and happens to be a morphism in this case ( * ? ? ?
* iii.3.5 ) .
if @xmath56 is a line in the plane , then by bezout s theorem @xmath56 intersects @xmath51 at four points counting multiplicity and we write @xmath57 .
the line @xmath56 is _ tangent _ to @xmath51 at a point @xmath58 if we can say @xmath59 .
bitangent _ is a line @xmath56 such that @xmath60 or @xmath61 .
when @xmath56 and @xmath51 intersect only at one point @xmath58 we call that point a _ hyperflex_. over a field of characteristic greater than three , the number of hyperflexes is bounded by 12 @xcite .
there are only two quartic curves with exactly 12 hyperflexes , the fermat quartic defined by @xmath62 and the curve defined by @xmath63 @xcite . over @xmath38 ,
the fermat quartic has 28 hyperflexes .
the fifty - six exceptional curves on @xmath16 come from the twenty - eight bitangents to the quartic curve @xmath51 .
each bitangent to @xmath51 lifts to a pair of exceptional curves in @xmath16 that intersect transversely at two points of @xmath51 or , in the case of a hyperflex , are tangent to each other as pictured below .
since the bitangents to @xmath51 are lying in the plane @xmath6 , each bitangent is guaranteed to intersect each of the other 27 bitangents once .
similarly to the case of cubic surfaces , now up to four bitangents can intersect at once ( * ? ? ? * proof of lemma 4.1 ) .
this corresponds to four of the exceptional curves in @xmath16 intersecting at one point .
we call such a point in @xmath16 a _ generalized eckardt point_. as with cubic surfaces , a point on @xmath16 where three exceptional curves meet is called an eckardt point . in order to find a formula for @xmath64 similar to hirschfeld s formula for @xmath65 we need to establish a notation for the points on the bitangent lines to @xmath51 .
let @xmath66 be a bitangent to @xmath51 for @xmath67 and let : + @xmath68 = number of points on @xmath66 where exactly 4 bitangents meet , + @xmath69 = number of points on @xmath66 where exactly 3 bitangents meet , + @xmath70 = number of points on @xmath66 where exactly 2 bitangents meet , + @xmath71 = number of points on @xmath66 that are not on any other bitangents , + @xmath72= number of times @xmath66 intersects @xmath51 over @xmath73 .
thus @xmath74 .
+ then , @xmath75 and @xmath76 now let @xmath77 since each @xmath66 will lift to two exceptional curves in @xmath16 that intersect only over @xmath28 points of @xmath51 , we get the formula @xmath78 notice that we do not double count the @xmath79 points because those are places where the map @xmath54 is one - to - one .
also notice that in order for @xmath16 to have a chance at fullness , we need @xmath80 .
we can use equations ( [ line ] ) and ( [ int ] ) to turn ( [ pts ] ) into a formula which depends on only @xmath36 , the number of points on the quartic @xmath51 , and the number of generalized and classical eckardt points . to do this we first manipulate ( [ int ] ) to get the equality @xmath81 which leads to @xmath82 then combining @xmath83 from ( [ line ] ) with ( [ f ] ) we get the following formula for @xmath84 , @xmath85 using equations ( [ f ] ) and ( [ g ] ) to eliminate the @xmath86 and @xmath84 in ( [ full ] ) yields the desired formula @xmath87 in order for @xmath16 to be full
, the equation @xmath88 must be satisfied . or we can solve @xmath89 full degree two del pezzo surfaces can only exist over @xmath73 when @xmath37 . in order to find a lower bound for the size of the field
, we must find the smallest field in which a degree two del pezzo can split .
we discover this bound by looking at how the bitangents of a quartic plane curve intersect .
since each of the bitangents must intersect the 27 others and at most four can intersect simultaneously , each bitangent must contain at least 9 points .
the number of points on a line defined over @xmath28 is @xmath90 , hence @xmath91 the upper bound is found by analyzing equation ( [ full ] ) .
there are at most 126 generalized eckardt points on a degree two del pezzo surface since each bitangent line contains at most @xmath92 of them .
hence , @xmath93 there are also at most 242 eckardt points on a degree 2 del pezzo surface since each bitangent can contain at most 13 eckardt points and not all 28 of them can .
even so , assuming all 28 lines contain 13 eckardt points yields @xmath94 .
thus , we can see that the maximum value for the right hand side of ( [ full ] ) is 253 when @xmath95 . in order to maximize the @xmath36 on the left hand side of ( [ full ] )
, we can assume that the quartic @xmath51 does not have any @xmath28-rational points .
thus we see that in order for a degree two del pezzo surface defined over @xmath28 to have the possibility of being full , we need @xmath96 or @xmath97 . when @xmath98 , the weil bounds tell us that @xmath51 will contain at least 2 and at most 74 rational points , so the case @xmath99 does not apply . in this case
( [ full ] ) becomes @xmath100 .
one solution to this equality is @xmath101 , which we have not yet ruled out as a possibility .
the degree two del pezzo surfaces we experiment with all contain generalized eckardt points . over small finite fields ,
split surfaces always have generalized eckardt points . but
kaplan ( * ? ? ?
* prop 76 ) has found two surfaces over @xmath43 that do not contain these points . over @xmath102 and @xmath103 split degree two del pezzo surfaces
contain generalized eckardt points . over a field with @xmath36 elements ,
every bitangent to the quartic @xmath51 contains exactly @xmath90 @xmath28-rational points .
when @xmath104 , @xmath105 for each @xmath106 , so @xmath16 contains @xmath107 generalized eckardt points .
when @xmath108 , there are 12 points on each bitangent so @xmath109 for each @xmath67 .
thus @xmath16 contains at least @xmath110 generalized eckardt points .
now suppose by way of contradiction that the size of the ground field is 13 and the del pezzo does not contain any generalized eckardt points .
then , every bitangent to @xmath51 must intersect 2 lines simultaneously at 13 points , @xmath111 for all @xmath67 .
this means that the number of eckardt points on the corresponding del pezzo surface is @xmath112 .
thus , the surface must contain a generalized eckardt point .
one reason we restrict ourselves to surfaces with generalized eckardt points is that over an algebraically closed field , a quartic curve @xmath51 admitting a generalized eckardt point is equivalent to @xmath51 admitting an involution ( * ? ? ?
* theorem 5.1 ) .
thus there is nice symmetry between the lines .
kuwata gives explicit equations for the quartic curves which admit a pair of commuting involutions , and these are the surfaces we study @xcite .
let @xmath51 be a quartic plane curve with a pair of commuting involutions . by a change of coordinates , possibly defined over a field extension
, @xmath51 can be defined by an equation of the form @xmath113 kuwata uses the equation above to find a family of quartic curves whose twenty - eight bitangent lines are all defined over the base field which does not need to be algebraically closed .
we call the curves @xmath114 defined below _ kuwata curves _ @xcite .
let @xmath115 be three elements in @xmath73 satisfying @xmath116 let @xmath117 be the quartic plane curve given by @xmath118 the twenty - eight double tangent lines of @xmath119 are given by @xmath120 we use the computer algebra system magma to find all kuwata curves over our desired fields .
then we use the equations of the lines to determine the intersection form @xmath121 of each line . with this information
, we are able to find full degree two del pezzo surfaces .
the following section lists the results we were able to find while experimenting with kuwata curves in magma .
another recent result involving degree two del pezzo surfaces with generalized eckardt points is due to salgado , testa , and vrilly - alvarado @xcite .
let @xmath122 be homogeneous polynomials of degrees two and four respectively such that the polynomial @xmath123 has distinct roots .
when the characteristic of the ground field is not two , the surface @xmath16 in the weighted projective space @xmath124 with equation @xmath125 is a smooth del pezzo surface of degree two with an involution given by @xmath126 and contains at least two generalized eckardt points @xmath127 $ ] and @xmath128 $ ] .
in fact , every del pezzo surface of degree two with a point contained in four exceptional curves has an involution and is of the form described above .
in this section we present the results we found using magma and kuwata curves . for each of the finite fields we list the full degree two del pezzo surfaces and how each of the bitangents to the quartic @xmath51 intersects the other bitangents in the form @xmath129 defined above .
our first result on full degree 2 del pezzo surfaces was first found by hirschfeld when he was studying full cubic surfaces@xcite .
he found that there is only one cubic surface over @xmath38 with exactly one rational point not on the lines .
this point can be blown - up to create a full degree two del pezzo . over the field @xmath38 there exists a unique full degree two del pezzo surface up to isomorphism .
it is given by the equation @xmath130 and all of the lines on @xmath0 contain nine generalized eckardt points .
the bitangents corresponding to equation ( [ f13_1 ] ) come in two intersection types .
there are 24 bitangents of type ( 1,11,2,0,0 ) and 4 of type ( 3,9,0,0,2 ) .
the quartic curve @xmath135 contains eight @xmath40-rational points .
the bitangents corresponding to equation ( [ f13_2 ] ) also come in two intersection types .
there are 24 bitangents of type ( 1,11,2,0,0 ) and 4 of type ( 1,12,0,0,1 ) .
the quartic curve @xmath136 contains four @xmath40-rational points , all hyperflexes .
we know that these are the only full degree two del pezzos over the fields @xmath137 because nathan kaplan has found all the split degree two del pezzo surfaces over these fields ( * ? ? ?
* prop 70,71,72 ) , and our results agree . in ( * ? ? ? * prop 73 ) , kaplan gives seven isomorphism types for split degree two del pezzo surfaces defined over @xmath41 .
his first four equations correspond to the first four full @xmath16 s we list above that come from kuwata s equations .
his sixth equation is isomorphic to the kuwata curve @xmath141 which is not full .
thus , once the field has at least 17 elements , there is enough room for a degree 2 del pezzo surface to be split but not full .
his fifth and seventh equations do not correspond to kuwata curves , but are indeed full and are isomorphic to the last two equations in the theorem .
the bitangents corresponding to equation ( [ f17_1 ] ) come in two intersection types .
there are 12 bitangents of type ( 1,8,8,0,1 ) , 16 of the type ( 3,3,12,0,0 ) , and @xmath142 . for equation ( [ f17_2 ] ) there are 12 bitangents of type ( 1,9,6,0,2 ) , 12 of the type ( 1,7,10,0,0 ) , 4 of type ( 3,6,6,3,0 ) , and @xmath143 . for equation ( [ f17_3 ] )
there are 12 bitangents of type ( 1,9,6,0,2 ) , 12 of the type ( 1,8,8,1,0 ) , 4 of type ( 3,3,12,0,0 ) , and @xmath143 . for equation ( [ f17_4 ] ) there are 8 bitangents of type ( 0,9,9,0,0 ) , 8 of the type ( 1,8,8,1,0 ) , 8 of type ( 1,7,10,0,0 ) , 4 of type ( 1,9,6,0,2 ) , and @xmath144 . for equation ( [ f17_5 ] )
there are 4 bitangents of type ( 0,9,9,0,0 ) , 12 of the type ( 0,10,7,1,0 ) , 6 of type ( 1,7,10,0,0 ) , 6 of type ( 1,9,6,0,2 ) , and @xmath142 . for equation ( [ f17_6 ] )
there is 1 bitangent of type ( 1,9,6,0,2 ) , 2 of type ( 1,9,6,2,0 ) , 8 of type ( 0,9,9,0,0 ) , 10 of the type ( 0,10,7,1,0 ) , 1 of type ( 1,7,10,0,0 ) , 2 of type ( 0,10,7,0,1 ) , 4 of type ( 0 , 11 , 5 , 0 , 2 ) , and @xmath142 . in (
* prop 75 ) , kaplan gives fourteen isomorphism types for split degree two del pezzo surfaces defined over @xmath42 .
his @xmath146 and @xmath147 equations correspond to the equations ( [ f19_1 ] ) and ( [ f19_2 ] ) in the list above .
his @xmath148 , and @xmath149 equations are isomorphic to the five other kuwata curves and are not full .
@xmath150 , @xmath151 , @xmath152 and @xmath153 equations are not kuwata curves and are not full .
but his @xmath154 , @xmath155 , and @xmath156 curves define full del pezzos that are not kuwata and are isomorphic to the last three equations in the theorem above .
the bitangents corresponding to the quartic in equation ( [ f19_1 ] ) come in five different types .
there are 8 bitangents of type ( 1,7,10,2,0 ) , 4 of type ( 1,7,10,0,2 ) , 8 of the type ( 0,7,13,0,0 ) , 4 of type ( 1,8,8,1,2 ) , 4 of type ( 1,8,8,3,0 ) , and @xmath157 . for equation ( [ f19_2 ] ) there are 12 bitangents of type ( 1,6,12,1,0 ) , 12 of type ( 1,5,14,0,0 ) , 4 of the type ( 3,6,6,3,2 ) , and @xmath158 . for equation ( [ f19_3 ] )
there are 2 bitangents of type ( 0,10,7,3,0 ) , 2 of type ( 0,10,7,1,2 ) , 8 of type ( 0,9,9,2,0 ) , 4 of type ( 0,9,9,0,2 ) , 2 of type ( 0,9,9,1,1 ) , 6 of type ( 0,8,11,1,0 ) , 2 of type ( 1,6,12,1,0 ) , 1 of type ( 1,9,6,2,2 ) , 1 of type ( 1,5,14,0,0 ) , and @xmath157 . for equation ( [ f19_4 ] )
there are 2 bitangents of type ( 0,10,7,2,1 ) , 4 of type ( 0,10,7,1,2 ) , 5 of type ( 0,7,13,0,0 ) , 5 of type ( 0,9,9,2,0 ) , 8 of type ( 0,8,11,1,0 ) , 1 of type ( 1,7,10,0,2 ) , 1 of type ( 1,9,6,4,0 ) , 2 of type ( 1,7,10,2,0 ) , and @xmath159 . for equation ( [ f19_5 ] ) there are 6 bitangents of type ( 0,7,13,0,0 ) , 6 of type ( 0,8,11,1,0 ) , 6 of type ( 0,9,9,2,0 ) , 6 of type ( 1,7,10,2,0 ) , 3 of type ( 1,7,10,0,2 ) , 1 of type ( 3,3,12,0,2 ) , and @xmath158 .
the bitangents corresponding to the quartic in equation ( [ f23_1 ] ) come in three different types .
there are 12 bitangents of type ( 1,16,12,3,2 ) , 12 of type ( 1,5,14,4,0 ) , 4 of type ( 3,3,12,6,0 ) , and @xmath162 . for equation ( [ f23_2 ] )
all 28 bitangents are of type ( 3,0,18,3,0 ) and @xmath163 .
there are twelve other isomorphism classes of kuwata curves that correspond to split but not full degree two del pezzo surfaces over @xmath43 . in (
* prop 76 ) , kaplan says that there are at least 19 isomorphism classes of split degree two del pezzo surfaces over @xmath43 .
he gives two classes that are not isomorphic to kuwata curves because they have no non - trivial automorphisms .
thus , kaplan has found two split surfaces over @xmath43 that contain no generalized eckardt points .
his surfaces are not full . since kaplan only gives the equations for two classes , and we know the equations for 14 kuwata classes , there are still at least three isomorphism classes of split @xmath16 s to be found that could possibly be full .
once the field size is larger than @xmath164 , none of the del pezzo surfaces corresponding to kuwata curves are full .
thus , a full @xmath16 defined over @xmath28 with @xmath165 can not have two commuting involutions in its automorphism group .
so we are left with the open question : akikazu kuribayashi and kaname komiya .
on weierstrass points and automorphisms of curves of genus three . in _ algebraic geometry ( proc .
summer meeting , univ .
copenhagen , copenhagen , 1978 ) _ , volume 732 of _ lecture notes in math .
_ , pp . 253299 .
springer , berlin , 1979 .
j. kollr . , volume 32 of _ ergebnisse der mathematik und ihrer grenzgebiete .
3 . folge . a series of modern surveys in mathematics [ results in mathematics and related areas .
3rd series .
a series of modern surveys in mathematics]_. springer - verlag , berlin , 1996 .
andr weil . .
in _ proceedings of the international congress of mathematicians , 1954 , amsterdam , vol .
iii _ , pp .
erven p. noordhoff n.v .
, groningen ; north - holland publishing co. , amsterdam , 1956 . | hirschfeld classified split del pezzo surfaces of degree at least three whose points are all contained on the lines in the surface .
we continue his work and begin the classification of split degree two del pezzo surfaces over finite fields whose points are all on the fifty - six exceptional curves of the surfaces . |
the notorious complexity of the yang - mills vacuum manifests itself in emergent infrared phenomena which , despite vigorous theoretical efforts over the last decades , remain only fragmentarily understood .
quark confinement @xcite was early recognized as a paradigm for these infrared complexities , both because of its unprecedented nature and of its almost universal impact on the hadronic world .
the long struggle to explain the nonperturbative confinement mechanism has led to the development and often ongoing refinement of a wide variety of theoretical approaches @xcite , ranging from various vacuum models to ab - initio lattice simulations wil74etc .
the latter , in particular , have provided strong and unbiased evidence for the main signatures of quark confinement , i.e. the linear growth ( and saturation ) of the static quark - antiquark potential at large distances and the associated color flux - tube formation @xcite ( and breaking @xcite ) . beyond establishing confinement and its characteristics , intense efforts were devoted to pinning down the underlying dynamical mechanism and to identify the infrared degrees of freedom best suited to describe it in humanly fathomable terms .
vacuum model analyses have played a prominent and often pioneering role in this endeavor .
the majority of them is based on ensembles of specific constituentfields , i.e. at least partly localized gauge fields ( with their prospective monopole loop and center vortex content ) which are supposed to populate and disorder the yang - mills vacuum .
potential candidates for these building blocks can be classified according to the increasing number of spacetime dimensions in which they are localized , and thus equivalently to their increasing efficiency in disordering the vacuum .
the minimally , i.e. in just two dimensions localized candidates are center vortices @xcite . an intermediate position occupy both ( gauge - projected ) abelian che97,tho76 and non - abelian monopoles or dyons @xcite ( possibly in the guise of bps constituents @xcite of kvbll calorons @xcite , see below ) which are localized in the three spacial dimensions . among the maximally , i.e. in all four ( euclidean ) spacetime dimensions localized candidates for the building blocks of confining field configurations , finally , are regular - gauge instantons @xcite , calorons with nontrivial @xcite holonomy and merons dea76,len04,len08 .
all of these pseudoparticles carry a nontrivial topological charge , mediate vacuum tunneling events and solve the classical yang - mills equation .
the latter property , in particular , has raised hopes that at least one of these solutions may provide a basis for semiclassical and therefore potentially analytical treatments of confinement .
( this proved indeed possible in lower - dimensional models where instantons or vortices were shown to generate confinement at weak coupling @xcite . ) in classically scale - invariant , four - dimensional yang - mills theory , however , where one is inevitably faced with large couplings at large distances , the justification for semiclassical approximations depends crucially on the characteristic distance scales involved . a successful semiclassical description of confinement , in particular , would require all physically relevant confinement effects ( as , for example , the quark binding inside light hadrons ) to take place over distances which remain small enough to avoid the strong - coupling regime .
in order to overcome these potentially stymying limitations , one may resort to simulating fully interacting pseudoparticle ensembles numerically .
the first and still most extensive such calculations were based on rather dilute superpositions of instantons and anti - instantons in singular gauge @xcite .
such superpositions maintain part of the instantons semiclassical nature , are known to generate a variety of important physical effects including spontaneous chiral symmetry breaking , and describe a large amount of successful vacuum and hadron phenomenology @xcite .
however , they fail to generate a confining , i.e. linearly growing potential between static quarks @xcite . by now , numerical simulations were performed for superpositions of all the other above - mentioned pseudoparticle candidates as well .
a central result was that each of them proved capable of generating a finite string tension .
the discovery of new finite - temperature instantons @xcite has given renewed impetus to semiclassically inspired confinement models .
these generalized calorons contain @xmath1 self - dual monopole ( dyon ) constituents without classical interactions and carry a non - trivial holonomy as an explicit degree of freedom .
the ability of the calorons to separate into @xmath1 quasi - independent constituents ( without cost in action ) in the confinement phase has led to simulations @xcite where the separation appears as an additional ( internal ) degree of freedom compared to the old - fashioned instanton simulations .
encouraging results have been obtained .
later , an analytic solution for the caloron - dyon gas was proposed @xcite at the price of a non - positive weight of an overwhelming multitude of configurations with respect to the moduli metric @xcite .
an analytic as well as numerical treatment of the purely abelian non - interacting dyon gas was shown to provide a confining force bru11 .
in addition , equilibrated lattice gauge - field configurations were successfully searched for the presence of pseudoparticles ( with the notable exception of merons ) in the yang - mills vacuum .
such searches are typically performed by smoothing techniques designed to filter out infrared gluon fields and in particular solutions of the classical yang - mills equation . for examples in the most extensively studied case of instantons and their size distribution
@xcite , and for calorons with nontrivial holonomy refs .
@xcite . in the present paper ,
building upon the above developments , we are going to introduce and study a new type of pseudoparticle ensemble .
the latter is based on superpositions of dimerons and anti - dimerons ( in singular gauge ) , i.e. on those four - dimensionally localized solutions of the classical su@xmath2 yang - mills equation which contain two meron centers dea76 .
the relatively large dimeron solution family includes as one extreme members which are contracted to ( singular - gauge ) instantons and as another members which are dissociated into two far separated merons .
moreover , it contains continuous sets of dimeron fields which interpolate between these extremes .
hence our configuration space includes many of the previously studied singular - gauge instanton and single - meron configurations and is ideally suited to study transitions between instanton and meron ensembles .
this is a crucial advantage because such a transition process was conjectured to take place between the instanton , dimeron and meron components of the yang - mills vacuum as a function of the bare gauge coupling .
indeed , this transition is a central ingredient of the meron - induced confinement scenario which was put forward on the basis of qualitative arguments by callan , dashen and gross ( cdg ) @xcite ( and further explored in several promising directions , including e.g. a two - phase picture of hadron structure @xcite which recieved some support from lattice simulations @xcite ) .
( for more recent work on merons see refs .
an alternative meron confinement mechanism was suggested in refs . @xcite . )
the cdg confinement scenario was inspired by the already mentioned weak - coupling confinement mechanism in 2 + 1 dimensional yang - mills - higgs theory @xcite .
in contrast to this classic paradigm , however , even the ingenious use of semiclassical arguments turned out to be of limited use in 3 + 1 dimensional yang - mills theory .
in fact , the resulting estimates remained qualitative at best .
further progress was achieved only recently , when a numerical study of superpositions of merons and antimerons @xcite provided the first convincing support for a central assumption of the cdg scenario , namely that the meron component of the yang - mills vacuum may indeed confine , i.e. generate a static quark potential which rises linearly at large interquark separations .
the suggested origin of the meron population in the vacuum from dimeron dissociation remained obscure , on the other hand , since the field content was from the outset restricted to a purely meronic phase .
to search for the development of such a meron - dominated phase and to analyze the underlying mechanism in dimeron - antidimeron ensembles evolving under the yang - mills dynamics will therefore be important points on our agenda . for this purpose , and in contrast to previous simulations of pseudoparticle ensembles
, we will monitor pertinent vacuum properties in a wide range of gauge coupling values .
in particular , we will look for signatures of the envisioned dimeron dissociation process .
similar to the gradual break - up of molecules into their atomic constituents with increasing temperature , this dissociation is suggested to be driven by a competition between the decreasing attraction among the dimeron s regularized meron centers and the entropy gain from their increasingly distant positions .
guided by the behavior of an ideal dimeron gas , one may indeed roughly estimate the inter - meron attraction to decrease relative to the entropy with growing coupling . in view of the strong interactions anticipated especially between the meron centers of different dimerons in significantly dissociated ensembles , however , such estimates are unreliable .
moreover , in our dynamical context the coupling - dependent competition between energy and entropy is particularly subtle since both attraction and entropy are expected to depend only logarithmically on the intermeron separation .
this further impedes even qualitative analytical estimates of dimeron ensemble properties .
our fully interacting ensembles , on the other hand , are well suited to tackle and clarify these issues .
since they take all classical gauge interactions including those which may be strong , long - range or many - body between the dimerons into account , we do not have to rely on the semiclassical or any other weak - coupling or low - density approximation .
nevertheless , it will be of heuristic value and of help for future , possibly more analytic treatments to understand which aspects of the dimeron - ensemble dynamics may be described semiclassically .
in fact , one of our reasons for transforming the individual dimerons into a singular gauge was to provide additional insight into this issue by more strongly localizing them .
this allows the dimerons and their meron centers to retain their identities and their classical shapes to a larger extent , even in the presence of quantum fluctuations , and thus facilitates ( semi-)classical behavior . at several points during the course of our investigation we will therefore check for indications of such behavior
besides studying the dimeron dissociation process and its dynamics in detail , we will also survey other structural and topological properties of the ensemble configurations .
in particular , we will examine distributions and correlations of the topological charge carriers and search for ordering tendencies with respect to both dimeron and meron centers .
( a side benefit for investigating topology distributions is that dimerons , in contrast to e.g. center vortices and monopoles @xcite , carry the topological charge of the yang - mills gauge group directly . )
this will result in a more detailed understanding of the restructuring processes which accompany the transition to the strong - coupling regime .
we will pay particular attention to the behavior of the topological susceptibility and of confinement properties , which we monitor by evaluating wilson - loop expectation values and the associated string tension , as a function of the bare coupling .
the quantitative picture emerging from these investigations will reveal , in particular , at which stage of dissociation ( as quantified e.g. by the average separation of the dimerons meron partners ) dimeron ensembles can best describe the confining phase of the yang - mills vacuum . in the next section
we start by acquainting the reader with pertinent properties of the dimeron solutions , transform them into a singular gauge and discuss the regularization of their singularities .
we then set up our ensemble field configurations and write down their partition function .
section [ simul ] presents an outline of our simulation strategy and , in particular , of the measures taken to control systematic finite - size and discretization errors . the next and central sec .
[ ressec ] contains our results and their discussion .
we start with an extensive statistical analysis of the spacetime , topological - charge and color structure of the ensemble configurations as well as of the crucial dimeron dissociation process .
these investigations provide a broad spectrum of insights into the behavior of the dimeron ensemble and its phase structure as a function of the gauge coupling .
they furthermore relate this behavior to the changes in the average properties of the individual dimerons .
we then evaluate the topological susceptibility and the static quark potential ( based on the calculation of wilson - loop expectation values ) and analyze the results in relation to the coupling - dependent restructuring of the ensemble configurations .
we further discuss scale - setting issues , evaluate dimensionless observable ratios , rewrite our results in physical units and confront them with those of other approaches .
section [ sum ] , finally , contains a summary of our main findings and presents our conclusions .
in the following subsections we construct the ( continuum ) dimeron field - configuration space on which our study will be based . along the way , we will motivate our expectation that such dimeron configurations approximate crucial features of the gauge - field population in the yang - mills vacuum , and in particular that they may provide a decent description of the confinement - deconfinement transition with decreasing gauge coupling .
we start by deriving the building blocks of our fields , i.e. the individual dimeron solutions of the yang - mills equation in singular gauge , and discuss their collective coordinates .
we then superpose these solutions to define our gauge - field configuration space and write down the partition function which specifies its dynamics .
finally , we discuss the ( both gauge - dependent and -independent ) singularities of our configurations and introduce suitable regularization procedures to prepare for their numerical treatment in sec . [ simul ] . dimerons ( or meron pairs ) are those classical solutions of the euclidean yang - mills equation which contain two meron centers where their only singularities are located @xcite . in regular gauge , with the two merons located symmetrically at the distances @xmath3 from the center at @xmath4 , the su@xmath0 dimeron solution family reads @xcite@xmath5 u^{\dagger } \sigma
_ { \mu \nu } u \label{dimreg}\]]where @xmath6 with the pauli matrices @xmath7 and the t hooft symbols @xmath8 . ( the latter are defined as @xmath9 , @xmath10 for @xmath11 and @xmath12 @xcite . )
the unitary matrices @xmath13 are global su@xmath14 color rotations .
the dimerons ( [ dimreg ] ) carry the same topological charge as instantons , i.e. @xmath15 . the anti - dimeron @xmath16 with opposite topological charge @xmath17 is obtained from eq .
( [ dimreg ] ) by replacing @xmath18 .
in contrast to instantons , however , dimerons are not selfdual .
the solution class ( [ dimreg ] ) depends on eleven real and continuous collective coordinates or moduli whose values uniquely and completely specify each member .
eight of these parameters , @xmath19 and @xmath20 , determine the spacetime position and orientation of the dimeron while the remaining three determine the su@xmath0 group elements @xmath21 .
euler angles which parametrize ( constant ) su@xmath22 group elements . ] the collective coordinates correspond to those eleven independent combinations of the classical and continuous yang - mills symmetries , i.e. of euclidean spacetime translations and rotations , conformal transformations and global su@xmath0 color rotations , which transform a representative dimeron into a gauge - inequivalent solution @xcite .
it is instructive to consider several limits of the solution family ( dimreg ) . for @xmath23
the meron centers coalesce and the dimeron turns into a pointlike , regular - gauge instanton .
( a finite - size instanton is reached after suitably regularizing the meron singularities , cf .
reg . ) for @xmath24 , on the other hand , the dimeron breaks up into two merons .
although these merons maintain the fixed relative color orientation of eq .
( [ dimreg ] ) , changing it by hand will yield increasingly action - degenerate two - meron solutions when the inter - meron separation @xmath25 becomes large .
this is because the color - dependent attraction which locks the meron constituents of eq .
( [ dimreg ] ) into their rigid color orientation decreases with increasing @xmath25 .
hence both instantons and merons are contained in the solution class ( [ dimreg ] ) as limiting cases .
the fields ( [ dimreg ] ) therefore provide on - shell interpolations between instantons and isolated meron pairs , i.e. continuous paths in yang - mills solution space which connect these particular gauge fields . in our context
these paths are of particular interest since they provide preferred doorways along which the dimerons may dissociate into merons .
as already mentioned , such dissociation processes are conjectured to drive the deconfinement - confinement transition .
the solutions ( [ dimreg ] ) have the same @xmath26 behavior as an instanton in regular gauge bel75,col80,vai99 , i.e. they contain a long - distance tail @xmath27 @xmath28 which implies an exceptionally weak localization . in multi - dimeron configurations these tails generate strong overlap interactions between the individual dimerons .
building dimeron ensembles by superposing solutions of the type ( [ dimreg ] ) would thus lead to very strongly correlated systems , in some respects similar to the regular - gauge instanton ensembles studied in ref .
@xcite . in the present work
we will take a different route , however .
in contrast to merons , dimerons like instantons may be transformed to singular gauges in which they are more strongly localized , with their long - range tails decaying as @xmath29 .
a superposition of singular - gauge dimerons thus improves the vacuum description at short distances , compared to superpositions of either regular - gauge instantons or single merons .
as long as the characteristic dimeron size is small compared to the average inter - dimeron separation , furthermore , it will provide much better approximations to classical yang - mills solutions .
this is essentially because at moderate pseudoparticle densities the contribution of the overlap regions to the action is much smaller than in regular - gauge dimeron superpositions .
hence singular - gauge multi - dimeron configurations provide a privileged testing ground especially for the semiclassical features of the cdg confinement mechanism @xcite .
in addition , dimeron configurations in singular gauge may also describe less classical or even fully quantum - mechanical aspects of confinement .
this holds , in particular , for the vacuum disordering mechanism envisioned by cdg .
the latter relies on the long - distance tails of the individual merons which come into play when the separation @xmath25 between the meron centers of the dimerons becomes large . as discussed above ,
the partner merons are then practically independent and regain their long - range tail @xmath28 ( at least in isolation ) .
the ensuing long - distance color correlations among these merons were found in refs .
len04,len08 to be far from semi - classical and to generate confinement .
of course , a complete dimeron dissociation into isolated and thus far separated merons is impossible in the finite spacetime volumes in which numerical simulations are feasible .
( the same holds for the solution in which one of the merons delocalizes on the three - sphere at spacetime infinity dea76,cal78 .
) however , such isolated merons would anyhow be unphysical ( since they carry infinite action ) and incompatible with a finite dimeron density .
instead , one expects that beyond some typical separation the two meron centers of a dimeron experience stronger interactions with their individual field environment than with their partner .
the rigid link between their color orientations can then be broken , i.e. the partner merons can become effectively independent of each other . in order to construct the multi - dimeron configurations motivated above
, we first transform the dimeron solution family ( [ dimreg ] ) into a specific singular gauge,@xmath30where @xmath31 is the large , i.e. topologically active gauge group element @xmath32(@xmath33 @xmath34 ) which has a singularity at the origin of the coordinate system .
the result can be cast into the form@xmath35(@xmath36 ) .
choosing for simplicity @xmath37 and @xmath38 , the two scalar functions @xmath39 become @xmath40the antisymmetric and anti - selfdual field @xmath41 contains important parts of the su@xmath0 and spacetime tensor structures ( and their mutual couplings ) .
explicitly , @xmath42(latin indices are spacial ) .
after regularization of the short - distance singularity ( cf .
[ reg ] ) the @xmath23 limit turns the dimeron ( [ gmsing ] ) into a singular - gauge instanton with @xmath43 $ ] and @xmath44 . in the following , we will often refer with the term dimeron to both dimerons and anti - dimerons in singular gauge .
occasionally , we will use the term pseudoparticle for the same purpose .
the leading asymptotic behavior of the singular - gauge dimerons ( [ gmsing ] ) [email protected] .
the long - range tail ( [ regdimtail ] ) has as intended disappeared and the overlap between neighboring dimerons is strongly reduced .
( this is in contrast to the meron ensembles of refs . @xcite whose constituents exist only in regular gauge . )
in addition to the meron - center singularities at @xmath46 , the solution ( [ gmsing ] ) inherits another singularity at @xmath47 from the gauge transformation ( [ sgtrf ] ) .
hence the impact of the latter will disappear when forming ( topologically insensitive ) gauge - invariant quantities from eq .
( [ gmsing ] ) .
the regularization of these singularities will be discussed in sec .
[ reg ] . by construction , the singular - gauge dimerons ( [ gmsing ] ) provide only a subset of the full solution family .
the complete eleven - parameter family can be recovered by translating the solutions ( [ gmsing ] ) to @xmath48 and by gauge - rotating them with a constant matrix @xmath49 su@xmath0 , i.e.@xmath50where the orthogonal matrices @xmath51 are obtained from the condition @xmath52 .
the group elements @xmath13 depend on three real and continuous parameters which form ( local ) coordinates on the su@xmath0 group manifold @xmath53 . in the following
it will be convenient to use the quaternion representation@xmath54which embeds @xmath53 into @xmath55 with euclidean coordinates @xmath56 by imposing the unit three - sphere constraint @xmath57 . in these
coordinates @xmath58 takes the form @xmath59 for our discussion below it will be useful to keep in mind that dimerons have three more continuous and noncompact collective coordinates than instantons ( but three less than a pair of independent merons , due to the locking of the inter - meron color orientation ) .
the three additional coordinates arise from the more complex structure of extended ( i.e. @xmath60 ) dimerons which requires more degrees of freedom to locate and orient them in spacetime .
this results in a substantially larger position entropy @xcite which is instrumental in counterbalancing the ( regularized ) dimeron action which grows logarithmically with @xmath61 .
indeed , the larger entropy is a necessary ( but not sufficient ) requirement for dimerons to dissociate with increasing coupling @xmath62 and to finally split up into their meron partners .
as motivated above , we intend to study a model which drastically reduces the field content of yang - mills theory ( as integrated over in amplitudes and the partition function ) to superpositions of @xmath63 dimerons and @xmath64 @xmath63 antidimerons in singular gauge , i.e. @xmath65with @xmath66 @xmath67 .
each term in this sum is uniquely characterized by the set @xmath68 of collective coordinates of the corresponding ( anti- ) dimeron .
we recall that the configurations ( [ cf ] ) differ distinctly from those obtained by transforming a regular - gauge dimeron superposition into ( any ) singular gauge .
this is because in eq .
( [ cf ] ) the gauge of each dimeron is chosen relative to its individual position .
one may wonder , incidentally , whether @xmath69-meron solutions with @xmath70 and their anti - solutions should be added to the superposition ansatz ( [ cf ] ) .
as in the case of instantons , however , the bogomolnyi - type bound @xmath71 on the action @xcite implies that such multi-(anti)-meron contributions to the partition function are exponentially suppressed relative to the dimeron contributions .
seem to be known analytically . ] since the entropy increases only logarithmically with @xmath69 and thus can not compensate this suppression , such multi - meron contributions may be safely neglected .
nevertheless , the ansatz ( [ cf ] ) should be regarded as a rather minimal choice .
it is mainly geared towards a transparent study of the proposed ( instanton and ) dimeron dissociation mechanism and its role in the deconfinement - confinement transition @xcite .
hence there are several natural directions in which eq .
( [ cf ] ) may be extended in future studies to provide a more complete description of the yang - mills vacuum physics .
an example would be to add the meron - antimeron pair solutions @xcite of the yang - mills equations , again individually transformed into singular gauge .
this would maintain the approximately semiclassical nature of the configurations at small @xmath72 and yield a richer dynamics .
however , it would probably also lead to a less transparent interpretation of the results , and we do not expect the topologically trivial meron - antimeron pair configurations to provide qualitatively new insights into the transition behavior . indeed , their limits for @xmath73 , a pure - gauge field of zero action , and for @xmath24 , a meron and an independent antimeron , indicate that meron - antimeron pairs do not generate new pathways for the transition .
more promising improvement options would include generalizations of the dimeron superposition ansatz ( [ cf ] ) which allow for a complete transition into a meron ensemble ( e.g. by releasing the rigid color locking between the dimerons meron partners beyond a suitable intermeron separation @xmath74 ) or the admixture of a pure instanton component @xcite with a realistic size distribution @xcite .
as discussed above , we view the dimeron configurations ( [ cf ] ) as a pertinent subset of the su@xmath0 gauge fields governed by the yang - mills dynamics .
hence we define the partition function of our dimeron model as@xmath75 \right\ } \label{pf}\]]where @xmath76 is the euclidean yang - mills action@xmath77 = \frac{1}{2g^{2}}\int d^{4}xtr\left\ { f_{\mu \nu } f_{\mu \nu } \right\ } = : \int d^{4}xs\left ( x\right ) \label{sym}\]]and @xmath78 is the gauge field strength @xmath79 .
\label{fym}\]]as already indicated , we expect the gauge interactions among the dimerons in eq .
( [ pf ] ) to play an important role in generating confining long - range correlations for large @xmath25 and , in particular , to provide a disordering mechanism for the vacuum .
this is in contrast to the situation in non - confining singular - gauge instanton ensembles where for most amplitudes a random orientation of the instantons ( i.e. the neglect of the interaction term @xmath80 in eq .
( [ pf ] ) ) yields a reasonable approximation to the vacuum physics @xcite .
the integration over the collective coordinates will be performed with the measure@xmath81(@xmath82 is the su@xmath0 haar measure ) for the dimerons , and analogously for the anti - dimerons .
this type of measure if familiar from instanton vacuum models @xcite .
an improved alternative may include jacobians which arise from the transformation of the linear gauge - field measure @xmath83 into the collective - coordinate basis and thereby implement the full moduli - space metric . finally , an additional @xmath84 dependence of the measure can emerge from the trace anomaly cal78,cal279 .
the singular - gauge dimeron solutions ( [ gmsing ] ) contain two types of singularities .
the first are those located at the constituent meron centers @xmath85 which persist in any gauge . in addition , there is a gauge singularity inherited from the topologically large and thus necessarily singular gauge transformation ( [ sgtrf ] ) , chosen to sit at the origin . in isolated dimerons , such gauge singularities could be gauged away and would therefore not affect ( topologically insensitive ) observables .
this ceases to be the case for the superpositions ( [ cf ] ) , however , where the location of the gauge singularities varies with the positions of the pseudoparticles .
since both types of singularities would impede our numerical simulations , they have to be regularized in a physically acceptable manner .
( recall that the individual dimeron fields ( [ dimreg ] ) and ( [ gmsing ] ) do not solve the yang - mills equation at their singularities , so that a suitably localized regulator will only minimally affect the semiclassical properties of the configurations ( [ cf ] ) . ) a natural way to regularize the singularities at the meron centers is to add the square of a size parameter @xmath86 to the denominators @xmath87 in eqs .
( [ f1 ] ) and ( [ f2 ] ) .
such a regulator imitates the way in which the size parameter enters the instanton solutions .
hence it may arise from scale - symmetry breaking quantum fluctuations which are expected to smear the classical singularities .
a finite @xmath86 furthermore acts as a uv cutoff since it limits the gradients of the configurations ( [ cf ] ) .
of course , the regularized field configurations cease to be exact solutions of the classical yang - mills equation .
hence the value of @xmath86 should be chosen large enough to avoid sizeable discretization errors ( cf .
[ eran ] ) but also small enough to avoid unnessary deformations of the dimeron solutions from the semiclassical saddle points .
we have found @xmath88 to be a reasonable compromise between both criteria and will use this value in our numerical simulations , if not stated otherwise .
the remaining singularities are , at least for isolated dimerons , gauge artefacts .
they remain unlikely to have a physical impact in our dimeron superpositions , too , but they may nevertheless cause problems in numerical approximations ( even for single dimerons ) , e.g. when subtle cancellations in gauge - invariant quantities are upset by discretization errors . a finite - size regulator as above
would hardly help to avoid such problems since it would just spread out the unphysical action - density peaks at the singularities .
hence we treat these gauge - dependent singularities in a pragmatic way , namely by interpolating the gauge field in 4@xmath89 balls of radius @xmath90 around the singularities with the field value at a specific point on the surface of the ball . in practice
, it turns out that with @xmath88 these highly localized singularities are smoothly regularized by taking @xmath91 as small as @xmath92 .
( the values of dimensionful quantities , such as @xmath86 and @xmath93 above , are given in numerical units originating from the discretization grid to be introduced in sec .
[ box ] . as a convenient length unit we have chosen @xmath94 , i.e. a tenth of the distance between nearest grid points .
( of course , this @xmath84 should not be confused with the four - vector @xmath84 which parametrizes the dimeron solutions . ) when transforming our results into physical units starting from sec .
[ topdissec ] , we will denote quantities in these numerical units by a hat above their symbols , cf .
[ topdissec ] . ) in our simulations it typically took several hundred sweeps through configurations of @xmath95 pseudoparticles ( cf .
mcsec ) before such a singularity was first encountered .
( even with a dramatically reduced floating - point variable length of 16 bits , incidentally , we have observed no overflows ( which unregularized gauge singularities would generate ) when evaluating the action density of a regularized two - dimeron configuration . )
in the following subsections we summarize how we have generated the dimeron configuration ensembles on which all our subsequent calculations will be based .
we further discuss predominant sources of systematic errors in these ensembles .
owing to the translational invariance of the yang - mills dynamics ( [ sym ] ) , ensemble averages over the pseudoparticle configurations ( [ cf ] ) in infinite spacetime will likewise be translationally invariant . in our simulations
we are restricted to a bounded volume of numerically manageable size , on the other hand , and thus have to keep boundary artefacts in calculated amplitudes under control , ideally within the size of the statistical uncertainties .
our initial step in this direction was to adopt _ singular - gauge _ dimerons with their more rapidly decaying long - range tails as the constituents of our gauge - field configurations ( [ cf ] ) . a sufficient suppression of boundary artefacts turned out to require additional measures , however , since the overlap among the still rather moderately localized dimerons generate important ( and typically repulsive ) interactions over ranges beyond the average nearest - neighbor distance .
in fact , these interactions are expected to play an important role in the confinement mechanism and should thus be distorted as little as possible by the boundary .
simulation costs , on the other hand , with their by far largest part due to integrating the action density on a spacetime grid , should be kept minimal . in order to approximately meet these conflicting goals
, we will adopt a multi - layered description of the boundary with decreasing grid - point density towards the outer layers , as sketched in fig .
[ gridfig ] .
the innermost core volume is a rectangular spacetime box in which we intend to evaluate amplitudes and thus require the physics to best approximate the infinite - volume limit . in order to allow for an efficient evaluation of wilson loops with elongated time directions in sec .
[ wlsec ] , this volume and its grid are chosen to be asymmetric .
one dimension , singled out as the euclidean time direction , contains 40 grid points while each spacial dimension contains 25 .
this turns out to provide a sufficiently high resolution for the amplitudes to be calculated below ( cf .
[ eran ] ) .
the core volume is encompassed by an ensemblevolume which contains the ( anti- ) meron centers of all ( anti- ) dimerons .
its surface is implemented numerically by rejecting metropolis updates ( cf .
[ mcsec ] ) during which a meron center would leave this volume .
the extent of the ensemble volume in a given direction will be taken 15% larger than that of the core in the same direction . in the part of the ensemble box which surrounds the core
we reduce the grid - point density to @xmath96 of that in the core volume .
this turns out to yield a still adequate resolution while substantially reducing the computational cost of evaluating the action .
a satisfactory suppression of field distortions inside the core volume turns out to require an additional precautionary measure , however .
it consists in correcting for a particularly prominent boundary effect , namely the artificial attraction of the pseudoparticles to the surface of the ensemble volume .
the latter arises because a substantial part of the action density of dimerons near the boundary is located outside the ensemble volume and thus not accounted for while the compensating tails of outside dimerons reaching into the ensemble volume are neglected .
we approximately remove this artefact by surrounding the ensemble box with another , covering volume which extends beyond the ensemble box by 10% of the core size in each direction . to keep the additional computational costs under control ,
we decrease the grid - point density of the outermost shell ( i.e. the part of the covering volume not shared by the ensemble volume ) inversely with the distance from the ensemble boundary .
since this shell does not contain the rapidly varying fields close to the meron centers , calculating the action in the full covering volume indeed largely prevents the fake attraction to the ensemble boundary ( cf .
[ dimenssec ] ) .
alternatively , finite - size effects in pseudoparticle ensembles can be efficiently corrected by employing the ewald summation technique , as has been shown very recently in the simpler case of abelian dyon field ensembles @xcite .
we evaluate the discretized functional integral over the pseudoparticle fields , i.e. the multi - dimensional integral over their collective coordinates in the partition function ( [ pf ] ) or any other amplitude , stochastically by monte - carlo importance sampling .
hence we average over dimeron configurations chosen randomly from a gibbs distribution with boltzmann factor @xmath80 where @xmath76 is the yang - mills action ( [ sym ] ) .
more specifically , we use the metropolis algorithm to generate homogeneous markov sequences of dimeron configurations that visit fields with larger probabilities more often . after reaching equilibrium , the probability of finding a configuration in the ensemble of subsequently generated fields
follows the gibbs distribution .
the initial pseudoparticle configurations for these markov sequences are obtained by choosing their collective coordinates from a uniform random distribution .
this procedure has equal access to all members of the configuration space and practically always results in configurations far from equilibrium , with actions several orders of magnitude above their equilibrium values .
we have also tested various more ordered initial arrangements of the dimerons and convinced ourselves that those lead within errors to equivalent equilibrium ensembles ( see sec . [ eran ] ) . for the markov update of the @xmath69-th configuration @xmath97 to its sequel @xmath98 according to the metropolis rules , we first generate some candidate configuration @xmath99 by randomly choosing an individual pseudoparticle in @xmath97 and modifying alternately ( from one candidate to the next ) either its position , i.e. @xmath100 and @xmath101 , or its color orientation @xmath102 .
the increments @xmath103 @xmath104 and @xmath105 are chosen randomly and uniformly subject to the constraint that their length remains limited .
more specifically , we demand @xmath106 $ ] , @xmath107 $ ] as well as @xmath108 $ ] ) are varied independently .
the constraint among them is reinstated only afterwards by rescaling all coefficients with a common factor .
( hence @xmath109 represents the haar measure on su@xmath110 . ) ] and further restrict @xmath111 .
the candidate configuration @xmath112 is accepted as the new configuration @xmath113 with probability @xmath114 -s\left [ a^{\prime } \right ] \right ) \right\ } $ ] .
if a candidate is rejected , the unchanged configuration is taken as @xmath113 ( and included in measurements like all others ) .
the choice of the maximal modification step sizes @xmath115 and @xmath116 can be used to optimize both the thermalization rate and the decorrelation among subsequent field configurations of the generated ensembles .
this requires a compromise between too small values , which impede the progression through configuration space , and too large values which more strongly decorrelate subsequent configurations but cause an unefficiently large update rejection rate . examination of the relation between maximal step sizes and typical acceptance probabilities suggests an optimal acceptance rate of about 25% for updates in both position and color space . during the initial thermalization process we will therefore increase ( decrease ) the maximal step sizes @xmath117 and @xmath118 after every 10 consecutive candidate field configurations by 10% if the average acceptance rate for these configurations falls below ( rises above ) 25%
this dynamical step - size adaptation procedure accelerates the approach to equilibrium and avoids getting trapped into approximate would - be equilibria .
initially , i.e. far from equilibrium and near the high - action random configurations , the fields can relax in larger steps while closer to equilibrium the action fluctuations become smaller and require decreasing step sizes to maintain sufficient acceptance rates . from the time when approximate equilibration sets in , however , the step size is kept constant to preserve detailed balance .
we further note that increasing @xmath119 increases the acceptance probability since the yang - mills action ( [ sym ] ) scales as @xmath120 .
hence larger @xmath119 allow for larger step sizes , accelerate thermalization ( in less markov steps ) by decreasing the autocorrelation time and result in equilibrium ensembles with larger entropy . as a case in point , for @xmath121 ( and @xmath88 )
we find @xmath122 and @xmath123 in equilibrium , where the scale of the fluctuations is set by the competition between action and entropy .
( for @xmath124 , i.e. for more strongly localized meron centers , one instead reads off the smaller values @xmath125 and @xmath126 from fig .
[ thermhistory ] . ) for @xmath127 , on the other hand , one finds the indeed substantially larger maximal step sizes @xmath128 ( again for @xmath88 ) .
the decreasing step width during a typical thermalization history is plotted in the uppermost row of fig .
[ thermhistory ] for @xmath121 and @xmath129 .
note that different step sizes in position and color space are generally required to obtain action changes of comparable magnitude .
the efficiency of the ensemble generation process can be further improved by exploiting the during equilibration decreasing ruggedness and action of the dimeron configurations in yet another way , namely by starting on coarser grids and increasing the resolution only when a higher accuracy of the action evaluation becomes necessary .
indeed , as long as the action density is comparable to or larger than its value around the regularized meron centers , a reduced resolution is generally sufficient and can save computer time . during thermalization
the overall action decreases strongly , however , and the grid must be refined to prevent the then more prominent meron centers from artificially reducing their action density by hiding between grid points . in practice
, we implement this refinement procedure by starting the simulations with a @xmath130 times smaller grid - point density . when the values of suitable quantities ( e.g. the distance between nearest - neighbor meron centers ) begin to saturate , this factor
is increased to @xmath131 . only after again reaching approximate saturation the full grid - point density of sec .
[ box ] is activated for the final approach to equilibrium and the subsequent generation of the thermal ensembles .
the approximate saturation plateaus and subsequent grid refinements ( indicated by dotted vertical lines ) are clearly visible in the thermalization histories of fig .
[ thermhistory ] . in the following
we will denote a set of @xmath132 consecutive markov steps as a sweep ( where @xmath133 , cf .
[ fc ] ) . during such sweeps
each pseudoparticle of a configuration is on average once considered for a full update .
( the factor two arises since the individual markov steps attempt to modify either a dimeron s position or its color orientation , i.e. only one of the two subsets of its degrees of freedom . ) after four consecutive sweeps with the acceptance rate kept fixed at @xmath134 , all pseudoparticles in a configuration are therefore on average updated once . in order to reduce autocorrelations among ensemble configurations and thus to increase the statistical independence of successive measurements
, we will only select the configurations generated by every fifth sweep ( after approximate equilibration of the markov chain ) as ensemble members and employ a binning procedure to calculate amplitudes and their errors ( cf .
[ statsec ] ) . as indicated in the previous section , a reliable calculation of vacuum expectation values as ensemble averages
requires a sufficient thermalization of the markov sequences which generate the ensemble configurations .
our criteria for when their distribution approximates the equilibrium distribution closely enough are based on the measurement of several observables and correlations to be described below .
first insights into the configurations thermalization properties can be gained by following the markov evolution of the average distance @xmath135 of a fixed meron center to its nearest neighbor as a function of the number of sweeps .
two examples for such histories are depicted in the second row of fig .
[ thermhistory ] . in the left panel
the coupling is set to @xmath136 , i.e. our smallest value , which results in the slowest equilibration rates we have to deal with in this paper ( cf .
[ mcsec ] ) .
( for these measurements we have also selected a smaller meron - size regulator @xmath137 which further slows the equilibration process . ) a first and general insight to be read off from this figure is how the degree of thermalization reached after a given number of sweeps depends on the measured quantity .
while the average distance to the closest neighboring meron of equal topological charge appears to have equilibrated after about 800 sweeps , the analogous distance for oppositely charged merons may not be fully thermalized even after 1200 sweeps .
a similarly long equilibration history can be observed for the average probability @xmath138 that the nearest neighbor of a given meron center has opposite topological charge , as plotted in the last row of fig .
[ thermhistory ] .
this behavior seems to indicate that the average distance between two meron centers from the same dimeron equilibrates faster than their average separation from the meron centers of neighboring dimerons . for @xmath129 the typical relaxation times reduce to at least half of those at @xmath121 , confirming the expectation that more strongly coupled ensembles thermalize faster . for a better understanding of the global thermalization behavior , and especially of the character and uniqueness of the reached equilibria , we have also compared ensembles resulting from different initial configurations .
more specifically , we have prepared a set of initial dimeron arrangements in which the pseudoparticles were regularly positioned at maximal average distances .
their topological charges , color orientations and sizes were chosen either to approximate different types of action minima or to imitate close - to - equilibrium configurations found in previous thermalization runs .
all ensembles resulting from these different initializations turned out to generate within errors the same averages .
this provides , for once , evidence for the ergodicity of the underlying markov process , i.e. for its ability to access any possible dimeron configuration in a finite number of steps .
more importantly , this result strongly supports the conclusion that the thermalization processes indeed have come sufficiently close to equilibrium , i.e. that our measurements in sec .
[ ressec ] are performed in almost thermalized ensembles .
this conclusion is strengthened by the fact that all our runs with non - random initializations were performed at @xmath121 where equilibration is particularly slow .
we now turn to the analysis of the systematic errors which arise from finite - resolution and finite - size effects .
we then describe the steps taken to control them by accordingly refining our calculational strategy .
additional error - reduction measures , which apply to the calculation of specific amplitudes only , will be outlined in their corresponding sections below .
discretization errors arise in our context from the practical necessity to sample the field configurations ( [ cf ] ) on spacetime grids of finite resolution . to meet our accuracy requirements ,
we adapt the grid - point density according to the characteristic length and gradient scales in the pseudoparticle configurations under consideration . during thermalization , where these scales change drastically , we do so dynamically as described in sec .
[ mcsec ] .
in the ( approximately ) thermalized ensembles , on the other hand , these scales are essentially fixed by the pseudoparticle density and by the size of the regularized meron - center singularities . ) , on the other hand , that they have practically no impact on the simulations . ] as described in sec .
[ reg ] , the regulator is chosen about five times larger than the lattice unit @xmath94 , at @xmath139 .
this proves sufficient to keep the action of an isolated dimeron practically independent of its position on the grid .
although configurations with gradients larger than those around single meron centers frequently occur among the dimeron superpositions ( cf ) , their enhanced action renders them relatively unimportant when thermalization is achieved .
this explains why we did not encounter such configurations in our equilibrated ensembles ( cf .
[ dimenssec ] ) . the grid resolution s
impact on the calculated action values and on the thermalization process can be seen directly in the markov evolution histories of fig .
[ thermhistory ] .
the two vertical lines indicate the sweep numbers at which the resolution of the grid is refined ( cf .
mcsec ) . the full resolution , corresponding to @xmath94 ,
is reached only after the second refinement step . in order to provide a particularly stringent test of discretization errors ,
we have reduced our regulator @xmath140 for this simulation by a factor of four , to @xmath141 .
this size is of the order of the minimal lattice unit and about half of the value @xmath142 below which discretization errors become a concern . as a consequence
, one observes that the evolution of both plotted quantities ( i.e. the average meron - center distances @xmath143 and the probability for the nearest neighbor - meron to have opposite @xmath144 ) on the coarsest grid begins to saturate at a preliminary would - be equilibrium when further relaxation is prevented by insufficient action - density resolution . already after the first grid refinement
, however , relaxation continues until the deviation from the thermal values reduces to about 15 20% .
the equilibrium values are approached after the second refinement step . for our standard meron size @xmath145 the discretization errors
will of course be much smaller and should be well under control .
this conclusion is confirmed by checks on other observables , e.g. when calculating link elements and their concatenations into wilson loops in sec .
[ wlsec ] .
we now turn to the discussion of finite - size effects . those are artefacts of the simulation volume s boundary and turn out to be more difficult to control than the discretization errors .
our choice of the more strongly localized dimerons in singular gauge ( [ gmsing ] ) as the constituents of the configurations ( [ cf ] ) was partly motivated by reducing such boundary effects .
for the same purpose we designed the multi - layered boundary outlined in sec .
[ box ] . nevertheless , sizeable boundary artefacts of different origins , of different @xmath146 dependence and with varying impact on the calculated quantities remain to be dealt with . in order to analyze those quantitatively , we start by monitoring violations of translational invariance in the ensemble - averaged action density @xmath147 as a function of the distance from the boundary for two values of @xmath146 .
the results are of direct physical and practical importance since @xmath148 is related to the gluon condensate and since a reliable evaluation of the action is indispensable for generating trustworthy ensembles .
the action density @xmath149 , as defined in eq .
( [ sym ] ) , can be obtained analytically by inserting the dimeron superposition ( [ cf ] ) into the field strength ( [ fym ] ) .
the ensemble average @xmath150 is then evaluated as outlined in sec .
[ mcsec ] , with ensemble configurations taken from every fifth sweep after equilibration . in order to visualize the local homogeneity properties of the action density in different sectors of the simulation volume , we plot @xmath150 in figs .
[ stg1fig ] and [ sdg5fig ] as a function of the distance from the center of the box along two different directions .
the latter were chosen to provide insight into the anisotropy of the boundary effects , caused in particular by our asymmetric grid with its elongated temporal direction .
( instead of calculating the values of @xmath149 at neighboring , equidistant points throughout these directions , we surround them with adjacent hypercubic boxes of side length @xmath151 inside which we sample @xmath149 at 10 randomly chosen points . )
we start our discussion of the @xmath148 plots with those generated at our smallest coupling value , @xmath121 . in the left panel of fig .
[ stg1fig ] we show @xmath148 ( black squares , full line ) along the time direction , in the right panel along the average over all diagonals of the spacial box at the mid point of the time axis .
deviations from a translationally invariant , i.e. constant @xmath150 near the center of the box are somewhat smaller along the time direction than along the diagonal .
this is expected since the former keeps a larger distance from the spacial boundaries .
nonetheless , @xmath152 fluctuates considerably even in the temporal direction and even close to the box center .
a reasonable fit to a constant plateau remains possible inside most error bars up to distances of order @xmath153 both along the time and diagonal directions , however , as also shown in fig .
[ stg1fig ] . beyond distances of about ten lattice units
the missing field contributions from outside of the box volume ( cf .
[ box ] ) begin to reduce the action density substantially . as a consequence of these results
, we will restrict the volume in which we evaluate amplitudes to subvolumes of the core box in which the maximal distances from the center are of order one ( if not noted otherwise ) . to demonstrate the @xmath146 dependence of the boundary effects , we further show the two analogous @xmath148 profiles for @xmath129 in fig .
[ sdg5fig ] .
clearly , the boundary effects are strongly reduced and the regions close to the center now show manifest plateaus which extend up to distances @xmath154 from the center . with further increasing @xmath146
the boundary artefacts become even more restricted to the surface region until the growing dimeron dissociation ( i.e. the growing intermeron distance @xmath155 ) creates a different type of sensitivity to the boundary , as discussed in secs . [ box ] and [ dimenssec ] .
since in general both character and strength of the boundary artefacts are amplitude dependent , we have also monitored the vacuum expectation value of quadratic probe wilson loops @xmath156 as a function of their distance from the box center .
the results will guide us in reliably calculating the expectation values of rectangular wilson loops in sec .
[ wlsec ] .
since the largest among these loops play a crucial role in understanding the confinement properties of dimeron ensembles , it becomes especially important to understand the boundary s impact on them .
we survey it by evaluating the expectation values @xmath157 of quadratic wilson loops
( for details see sec .
[ wlsec ] ) of side length @xmath158 centered along the same two lines as the action density @xmath150 above .
the loop orientation turns out to have no significant impact on the value of @xmath157 and will thus be averaged over . in fig .
[ stg1fig ] we have included plots of @xmath159 along the time direction in the left panel and along the average over the spacial diagonals in the right panel for @xmath121 , and in fig .
[ sdg5fig ] we display @xmath160 for @xmath129 . in all four cases , with growing @xmath161
the distance dependence of @xmath157 becomes increasingly proportional to that of the average action density @xmath150 .
for quadratic wilson loops of minimal size this is expected since such plaquettes form the essential part of a discrete approximation to the yang - mills action density . in our case , the observed
proportionality further indicates that the configurations are sufficiently smooth over our rather large test loops , probably because their side length is comparable to the regulated meron size @xmath145 . to summarize ,
while discretization errors in our dimeron simulations can be reliably controlled , we have observed significant violations of translational invariance in both action density and wilson loops . although our efforts to reduce boundary artefacts prove effective close to the center of the core box , substantial deviations from homogeneity set in at distances of order one , in particular for our smallest coupling value @xmath136 where they are most pronounced .
these boundary effects originate from a combination of the strong intermediate - range interactions between dimerons and the collective effects emanating from the boundary ( cf .
[ box ] ) .
the lessons learned from the above analysis will guide us in keeping boundary artefacts of our results tolerably small , mainly by relying on specifically reduced volumes in which to evaluate amplitudes . the effective dimeron field theory introduced in sec .
[ sdeft ] has three adjustable parameters : the gauge coupling @xmath146 , the meron - center size @xmath162 , and the approximately equal numbers @xmath163 of dimerons and antidimerons .
we have generated all our ensembles with @xmath164 pseudoparticles , divided into @xmath165 dimerons and @xmath166 antidimerons , in the multi - layered multi - grid volume specified in sec .
the meron centers are smeared as described in sec .
[ reg ] , with a common regulator value @xmath88 .
we have generated two independent ensembles for each of the coupling values @xmath167 ( with decreasing number of members ) and a third one for @xmath121 .
the total number of sweeps per coupling in equilibrium as well as the maximal stepsizes are collected in tab . [ enstab ] .dimeron ensemble characteristics .
( for definitions see sec .
[ mcsec ] . ) [ enstab ] [ cols="^,^,^,^",options="header " , ] an important dimensionful quantity is the meron - center size regulator @xmath86 which furnishes the characteristic input scale of the dimeron ensembles ( cf . sec .
[ reg ] ) . for @xmath168 and @xmath169
we find @xmath170fm , i.e. about a quarter of the typical instanton size @xmath171 fm in ( singular - gauge ) instanton - liquid models @xcite .
this indicates that our regularization procedure deforms the classical dimeron solutions only moderately and should not impede a potentially semiclassical behavior of their superpositions .
the average dissociation @xmath172 of the dimerons at @xmath168 increases the effective dimeron size , on the other hand , and brings it closer to typical instanton sizes . finally , it is useful to note that @xmath173 gev approximately sets the renormalization scale of our ensembles .
( recall that @xmath86 acts as an uv regulator e.g. for the action density which diverges logarithmically when @xmath174 . )
these @xmath175 values are of the order of typical lattice scales , which may help to explain why our results for the scale - dependent action density are compatible with lattice values .
we now turn to the dimeron density of our ensembles . recalling the grid - point distribution from secs .
[ box ] and [ statsec ] , the four - volume of our ensemble box is given by @xmath176 .
the corresponding values of the dimeron density @xmath177 in physical units are listed in table [ respu ] . in the preferred coupling region
@xmath168 they are about 2080% larger than the typical instanton densities @xmath178 @xmath179 of instanton liquid models @xcite .
the physical significance of this result will be discussed below .
moreover , it is interesting to note that the density @xmath180 @xmath179 of merons in the @xmath181 ensemble of ref .
@xcite is similar to the density of our meron constituents , i.e. approximately twice as large as our dimeron densities @xmath182 .
this adds to our earlier evidence for an increasingly dynamical role of the largely independent meron constituents from the farthest dissociated dimerons .
it remains to put our results for the topological susceptibility in the last column of table [ respu ] into a quantitative perspective .
the dimeron ensemble predictions for @xmath183 in the coupling region @xmath168 are about 50% smaller than the physical value @xmath184mev ( i.e. the su@xmath2 lattice value with the scale set by @xmath185 @xmath186 , cf .
[ tssec ] ) .
similarly small values @xmath187mev were encountered in the meron ensembles of ref . @xcite while regular - gauge instanton ensembles yield @xmath188mev .
hence our relatively small @xmath183 values are consistent with previous indications for the meron centers , which carry only half of the dimerons topological charge , to increasingly determine the topological - charge fluctuations for @xmath189 .
( one should also keep in mind , however , that we consider our values for @xmath190 as lower bounds , cf .
another suggestive pattern in our data is the weak bare - coupling dependence of the @xmath183 values for @xmath168 .
as already alluded to , this behavior may reflect the well - known temperature independence of the topological susceptibility in the confined yang - mills phase @xcite which in turn indicates that @xmath183 is relatively insensitive to temperature - induced variations of the coupling @xmath191 up to the critical temperature @xmath192 .
( for @xmath193 , on the other hand , @xmath194 drops very rapidly . ) the unusually large value of @xmath183 in our @xmath195 ensemble was already noted above in the raw data ( cf .
table [ tstab ] ) .
of course this result raises suspicion . in a random distribution of merons , into which the dimerons dissociate for @xmath196 as far as our limited field configuration space and the boundary allows
, one would expect the topological susceptibility to be an incoherent sum of single - pseudoparticle contributions and thus to have a value of the order of the pseudoparticle density .
this expectation was verified explicitly in random singular - gauge instanton @xcite , regular - gauge instanton and single - meron @xcite ensembles .
the fact that our for @xmath196 strongly increasing @xmath194 values fail to scale with the density is therefore probably another manifestation of the residual correlations between the dimerons meron partners and the related @xmath196 boundary artefacts .
we recall that in secs .
[ dissocsec ] and [ tssec ] these artefacts were found to impede the development of a genuine random meron ensemble and to strongly contaminate the topology distributions .
fortunately , in the physically relevant coupling region @xmath168 these effects are much weaker and under far better control .
our above results suggest a remarkably comprehensive role for dimerons in the yang - mills vacuum .
a key feature of the emerging picture is the rather distinct division of labour between almost fully contracted , instanton - like dimerons and their far dissociated , meron - like counterparts . in dynamical equilibrium at intermediate ,
i.e. physical couplings both of these components coexist at fairly high densities . in particular
, our overall dimeron densities significantly exceed the instanton density in instanton liquid models .
hence the instanton - like dimeron component should be sufficiently populated to perform essentially all established tasks of instantons in the yang - mills vacuum @xcite .
the strongly quadratic rise of our heavy - quark potentials at small interquark separations provides additional evidence in this direction ( cf .
[ wlsec ] ) .
the meron - like component , on the other hand , seems to be mainly responsible for the longest - distance vacuum physics including , most importantly , linear quark confinement as established in sec .
[ wlsec ] and in single - meron ensembles . on a structural level ,
the above reinterpretation of the instanton component in the yang - mills vacuum as mainly consisting of hardly dissociated dimerons is compatible with the existing instanton phenomenology as well . indeed , due to the destructive interference between their meron - center tails the instanton - like dimerons experience much weaker long - range interactions than the almost liberated merons of the meron - like component ( cf .
dimenssec [ colcorsec ] ) .
similar to singular - gauge instantons in sufficiently dilute superpositions they overlap less with each other , furthermore , and thus retain more of the classical solutions shapes and semiclassical behavior .
hence the instanton - like dimeron component indeed resembles a rather weakly interacting liquid . on the other hand , this raises the question why cooling studies , designed to detect semiclassical solutions in equilibrated lattice configurations , have reported clear evidence for instantons or calorons with non - trivial holonomy but none for dimerons cal - latt , instlat .
closer inspection reveals , however , that these findings are not in conflict with our above scenario .
in fact , the finite - resolution algorithms used to identify these solutions by their shapes may have counted small dimerons as instantons . moreover , under cooling the logarithmic attraction between the dimerons ( smoothed ) meron centers can no longer be counterbalanced by their decreasing entropy . hence the dimerons will shrink and at least partially coalesce into instantons @xcite as which they are identified afterwards . as established in sec .
[ wlsec ] , the effectively released meron centers of the meron - like dimeron component develop the strong long - range correlations required to generate confinement .
although these color correlations produce a more complex long - distance structure than encountered in fully disordered random ensembles , they still generate enough entropy for a string tension of the physically expected magnitude to develop ( cf .
[ wlsec ] ) .
moreover , it seems likely that our above scenario can be extended to an at least qualitatively realistic description of the qcd vacuum .
the instanton - like component of dimeron ensembles , in particular , should adapt to the presence of light quarks in essentially the same manner as in instanton liquid models .
this is because most instanton effects in the light - quark sector , including the induced quark interactions which spontaneously break chiral symmetry , originate from the characteristic quark zero - modes arising in the instanton background .
since these zero modes are of topological origin , their existence and bulk properties @xcite are little affected by moderate deformations of the instantons into dimerons .
hence the ensuing phenomenology should be similarly robust under such deformations .
we have studied the extent to which the yang - mills vacuum can be described by restricting its gauge - field content to superpositions of regularized su@xmath2 ( anti- ) dimeron solutions .
our set - up and analysis of the corresponding effective theory focused mainly on structural changes in dimeron ensembles as a function of the bare gauge coupling and traced their impact on topological and confinement properties .
the underlying quantum superpositions of dimerons , with their dynamics governed by the yang - mills action , were generated without recourse to weak - coupling or low - density approximations by monte - carlo simulations in a wide range of bare coupling values .
the localization of the individual dimerons was enhanced by transforming them into a singular gauge , which simultaneously improves numerical efficiency , the vacuum description at shorter distances and the compatibility with semiclassically motivated instanton liquid models .
our initial survey of dimeron ensemble properties and their coupling dependence concentrated on the spacial distribution of the dimerons , their local degree of selfduality and their meron substructure . to understand relevant aspects of the underlying inter - dimeron dynamics , we further studied short- and intermediate - range correlations between positions , topological charges and color orientations of neighboring dimerons and of their meron centers .
the resulting structural patterns turned out to be multifaceted and revealing . a strong repulsion up to distances of the meron center sizes
governs the short - range interactions between dimerons of both topological charges and at all couplings . in our
minimally coupled and thus most ordered ensemble , we additionally find a striking ( about sevenfold ) preference for nearest - neighbor dimerons to carry opposite topological charges , indicating a strong intermediate - range attraction ( repulsion ) between dimerons of opposite ( equal ) topological charge . a less pronounced but still significant majority of next - to - nearest neighbors is found to carry identical topological charges .
the resulting short- to mid - range order among the topological charges is reminiscent of debye screening clouds in a plasma .
these ordering patterns can be traced to the dependence of the underlying interactions on the relative color orientations between neighboring dimerons .
ensemble and configuration averages of these color orientations indicate , on the other hand , that in the mean the intermediate - range attraction between neighboring dimerons and anti - dimerons is almost completely compensated by their shorter- and longer - range repulsion and by the ensemble entropy . in any case , with increasing coupling and hence entropy the observed topological order weakens rather rapidly , together with the underlying color correlations .
the most far - reaching impact of the increasing bare coupling , however , manifests itself in the structural changes which it induces in the individual dimerons .
in particular , we found the first robust evidence , established in fully interacting ensembles , for the dimerons to graduallydissociate into their meron constituents when the coupling grows .
this process is driven by a competition between the coupling - dependent interactions and the entropy .
several complementary pieces of evidence further indicate that the meron constituents continuously replace the dimerons as the dynamically relevant degrees of freedom . with growing coupling * * * * the interactions between merons belonging to different dimerons therefore increasingly determine the ensemble properties .
this renders the dimerons meron partners more and more independent ( except for their frozen relative color orientation ) and allows our field configurations to approximate more closely those of meron ensembles .
in particular , however , the changing nature of the dynamically active topological - charge carriers and of their coupling - dependent correlations leaves various revealing imprints on bulk ensemble properties . among the latter ,
the topological susceptibility is particularly relevant in our context .
when calculated in a sequence of growing evaluation volumes up to the largest accessible ones , it is found to level off towards the end of the sequence .
this indicates or at least foreshadows the expected saturation of the topological - charge fluctuations in the thermodynamic limit .
moreover , the resulting values of the susceptibility in the intermediate , physical coupling range remain within errors practically constant .
this behavior may be related to the similarly weak temperature dependence of the topological susceptibility in yang - mills theory below the deconfinement temperature .
our intermediate - coupling results reproduce those of comparable single - meron ensembles rather closely , furthermore , which seems to indicate that ( on average relatively loosely bound ) meron centers of dimerons and single merons produce topological charge fluctuations of roughly equal strength . on the other hand ,
our results for the topological susceptibility underestimate lattice and instanton ensemble values by about 50% , perhaps due to insufficient saturation in our evaluation volumes and the underlying boundary effects . in any case
, the dimeron ensemble predictions for the susceptibility could be straightforwardly adjusted by increasing the dimeron density .
since the long - range color tails of less bound meron centers disorder the vacuum more strongly , we have payed particular attention to the confinement properties of the dimeron ensembles .
the latter were monitored by evaluating ensemble averages of rectangular wilson loops and by extracting the associated static quark - antiquark potentials for a wide range of coupling values .
initially , the resulting potentials are found to rise quadratically with the interquark separation .
this behavior is typical for ( sufficiently dilute ) instanton ensembles and indicates that our short - distance potentials are dominated by contributions from moderately dissociated dimerons which act essentially as singular - gauge instantons . for quark - antiquark separations beyond about half a fermi ,
on the other hand , the wilson loops indeed develop an area - law behavior and the potentials consequently show the linear distance dependence characteristic of quark confinement .
the associated string tension is found to grow with the coupling and to reach values of the experimentally expected order of the magnitude in the physical coupling range .
hence we have shown without reliance on weak - coupling or low - density approximations that sufficiently strongly coupled and therefore on average far enough dissociated dimeron ensembles indeed produce confining long - range correlations of about the physically required strength . since the latter originate from the meron tails of the more strongly dissociated dimerons , these results also provide complementary evidence for single - meron ensembles to confine ( even though our finite simulation volume and our limited field configuration space prevent the dimerons from breaking up completely ) . moreover ,
both dimeron and single - meron ensembles are found to generate confining static - quark potentials of comparable strengths at our largest accessible interquark distances .
since the linear rise of the meron - induced potential turns out to continue beyond these distances , furthermore , our above reasoning indicates that the same should happen in dimeron ensembles .
hence we expect an ongoing linear rise of the dimeron - induced potentials into the distance region where the underlying color flux tubes are fully developed .
while dimeron ensembles thus share with single - meron ensembles the main features of the confinement mechanism , they additionally provide a more complete description of the yang - mills vacuum .
the first evidence is that our results establish how a confining meron component in the vacuum can robustly emerge from instanton and dimeron dissociation .
this provides , in particular , the first quantitative confirmation for a key ingredient of the meron - induced confinement scenario envisioned more than three decades ago on the basis of qualitative arguments .
in addition , our results on the coupling dependence of dimeron ensemble properties should contain useful information on their temperature dependence and , in particular , on the impact of temperature - induced variations of the gauge coupling ( although we did not pursue this issue in the present paper ) . furthermore , we found evidence for the physical dimeron density to be about twice as large as typical instanton densities .
this allows confining dimeron ensembles to retain a phenomenologically relevant fraction of contracted and thus more weakly interacting , instanton - like dimerons .
an independent indication for their considerable density is the strong quadratic rise of our heavy - quark potentials at short interquark distances .
the instanton - like component should adapt to the presence of light quarks in essentially the same fashion as in phenomenologically successful instanton vacuum models , furthermore , since the underlying quark zero - modes in the instanton background are of topological origin and therefore little affected by the moderate deformations of instantons into contracted dimerons .
hence , in addition to confining color , dimeron ensembles should be able to reproduce much of the successful vacuum and hadron phenomenology predicted by ( non - confining ) singular - gauge instanton ensembles .
we would like to thank e .-
ilgenfritz and m. wagner for their help and numerous intense discussions , and we are grateful to m. wagner for providing us with his pseudoparticle simulation code . h.f . and
acknowledge financial support from the deutsche forschungsgemeinschaft ( dfg ) under project mu932/6 - 1 .
99 a. jaffe and e. witten , _ quantum yang - mills theory_,available at http://www.claymath.org/millennium/yang-mills_theory/ ; j. greensite , prog . part .
. phys . * 51 * , 1 ( 2003 ) ; a. di giacomo , arxiv : hep - lat/0510065 ; m. shifman , arxiv:1007.0531 ; r. alkofer and j. greensite , j. phys .
g : nucl . part .
* 34 * ( 2007 ) s3 ; a collection of useful reviews can be found in _ at the frontier of particle physics handbook of qcd _ , vol 1 - 4 , m. shifman ( ed . ) , world scientific , singapore ( 2002 ) and _ confinement , duality and nonperturbative aspects of qcd _ , p. van baal ( ed . ) , proc .
nato asi series b , vol .
368 , kluwer , new york ( 2002 ) .
a. di giacomo , m. maggiore , and s. olejnik , nucl .
* b 347 * , 441 ( 1990 ) ; v. singh , d. a. browne , and r. w. haymaker , phys .
* b 306 * , 115 ( 1993 ) ; g. s. bali , c. schlichter , and k. schilling , phys . rev .
d * 51 * , 5165 ( 1995 ) ; p. cea and l. cosmai , phys .
d * 52 , * 5152 ( 1995 ) ; m. caselle , f. gliozzi , u. magnea , and s. vinti , nucl . phys . *
b 460 * , 397 ( 1996 ) ; p.
pennanen , a. m. green and c. michael , phys .
d * 56 * , 3903 ( 1997 ) ; m. zach , m. faber , and p. skala , phys . rev .
d * 57 * , 123 ( 1998 ) ; y. koma , m. koma , and p. majumdar , nucl . phys . * b 692 * , 209 ( 2004 ) ; n.d .
hari dass and p. majumdar , jhep * 10 * , 020 ( 2006 ) ; m. n. chernodub and f.v .
gubarev , phys .
d * 76 * , 016003 ( 2007 ) ; p. giudice , f. gliozzi , and s. lottini , j. high energy phys . * 01 * ( 2007 ) 084 ; r.w .
haymaker and t. matsuki , phys .
d * 75 * , 014501 ( 2007 ) ; a. dalessandro , m. delia , and l. tagliacozzo , nucl .
b * * 77**4 , 168 ( 2007 ) ; m.s .
cardaci , p. cea , l. cosmai , r. falcone and a. papa , phys .
d * 83 * , 014502 ( 2011 ) .
m. faber , j. greensite and s. olejnik , phys .
d * 57 * , 2603 ( 1998 ) ; l. del debbio , m. faber , j. giedt , j. greensite and s. olejnik , phys .
d * 58 * , 094501 ( 1998 ) ; m. engelhardt and h. reinhardt , nucl . phys .
* b 585 * , 591 ( 2000 ) ; m. engelhardt , m. quandt and h. reinhardt , nucl .
* b 685 * , 227 ( 2004 ) ; s. rafibakhsh and s. deldar , aip conf . proc . *
892 * , 527 ( 2007 ) .
g. t hooft , nucl . phys .
* b 79 * , 276 ( 1974 ) ; a. m. polyakov , jetp lett . * 20 * , 194 ( 1974 ) ; g. t hooft , in high energy physics , edited by a. zichichi , editrice compositori , bologna , ( 1976 ) ; s. mandelstam , phys
. reports * * 23**c , 245 ( 1976 ) ; g. t hooft , nucl . phys . * b190 * , 455 ( 1981 ) ; r. haymaker , phys . rep . * 315 * , 153 ( 1999 ) ; a. kronfeld , g. schierholz and u.j .
wiese , nucl .
b293 * , 461 ( 1987 ) ; g. ripka , lecture notes in physics 639 , springer , berlin ( 2004 ) .
chernodub and m.i .
polikarpov , lectures given at _
nato advanced study institute on confinement , duality and nonperturbative aspects of qcd _ , cambridge , england , 23 jun - 4 jul 1997 , arxiv : hep - th/9710205 ; g. bali , talk given at _
3rd international conference on quark confinement and the hadron spectrum _
( confinement iii ) , newport news , va , jun 1998 , hep - ph/9809351 ; r. haymaker , phys . rep . * 315 * , 153 ( 1999 ) .
d. tong , arxiv : hep - th/0509216 ; k. konishi , lect .
notes phys .
* 737 * , 471 ( 2008 ) ; d. tong , annals phys .
* 324 * , 30 ( 2009 ) ; m. shifman and a. yung , _ supersymmetric solitons _ , cambridge university press , cambridge ( uk ) , ( 2009 ) ; m. shifman , int . j. mod
a * 25 * , 4015 ( 2010 ) . c. gattringer , phys .
d * 67 * 034507 ( 2003 ) ; c. gattringer and s. schaefer , nucl .
* b 654 * , 30 ( 2003 ) ; e .-
ilgenfritz , m. mller - preussker , b.v .
martemyanov and a.i .
veselov , phys .
d * 69 * , 114505 ( 2004 ) ; e .-
ilgenfritz , m. mller - preussker and d. peschka , phys .
d * 71 * , 116003 ( 2005 ) ; e .-
ilgenfritz , b.v .
martemyanov , m. mller - preussker and a.i .
veselov , phys .
d * 71 * 034505 ( 2005 ) ; phys .
d * 73 * , 094509 ( 2006 ) .
i. kogan and a. kovner , _ monopoles , vortices and strings : confinement and deconfinement in 2 + 1 dimensions at weak coupling _ , in at the frontier of particle physics handbook of qcd , m. shifman ( ed . ) , world scientific , singapore ( 2001 ) .
t. schfer and e.v .
shuryak , rev .
phys . * 70 * , 323 ( 1998 ) ; d.i .
diakonov , prog .
. phys . * 51 * , * * * * 173 ( 2003 ) . for an elementary introduction to instantons
see h. forkel , _ a primer on instantons in qcd _ , hep - ph/0009136 .
ilgenfritz , m.l .
laursen , g. schierholz , m. mller - preussker , h. schiller , nucl . phys .
* b 268 * , 693 ( 1986 ) ;
chu and s. huang , phys .
d * 45 * , 2446 ( 1992 ) ; m .- c .
chu , j.m .
grandy , s. huang and j. negele , phys .
d * 49 * , 6039 ( 1993 ) ; a. hasenfratz , t. degrand , and d. zhu , nucl . phys . * b 478 * , 349 ( 1996 ) ; p.
de forcrand , m. g. perez , i .- o . stamatescu , nucl .
* b 499 * , 409 ( 1997 ) ; d.a .
smith and m.j .
teper , phys .
d * 58 * , 014505 ( 1998 ) ; p.
van baal , nucl .
. suppl . * 63 * , 126 ( 1998 ) ; a. hasenfratz and c. nieter , nucl .
. suppl . * 73 * , 503 ( 1999 ) ; m.g .
perez , o. philipsen , i .- o . stamatescu , nucl . phys .
* b 551 * , 293 ( 1999 ) ; t. degrand , a. hasenfratz , phys .
d * 64 * , 034512 ( 2001 ) ; e .-
ilgenfritz et al , nucl .
. suppl . * 119 * , 754 ( 2003 ) ; c. gattringer et al .
* 129 * , 653 ( 2004 ) , f. bruckmann et al . , contribution to lattice 2004 , batavia , illinois ( 2004 ) , hep - lat/0408036 .
n. drukker , d.j .
gross and n. itzhaki , phys .
d * 62 * , 086007 ( 2000 ) ; j.v .
steele and j.w .
negele , phys .
* 85 * , 4207 ( 2000 ) ; h. reinhardt and t. tok , phys
* b 505 * , 131 ( 2001 ) ; a. montero and j.w .
negele , phys . lett .
* b 533 * , 322(2002 ) ; k. kondo , n. fukui , a. shibata and t. shinohara , phys .
d * 78 * , 065033 ( 2008 ) .
de forcrand and m. pepe , nucl .
* b 598 * , 557 ( 2001 ) ; ph .
de forcrand and m. pepe , nucl .
* b * ( proc . suppl . ) * 94 * , 498 ( 2001 ) ; f. gutbrod , eur . phys . j. c * 8 * , 1 ( 2001 ) ; nucl . phys .
* b 720 * , 116 ( 2005 ) .
shuryak and j.j.m .
verbaarschot , phys .
d * 52 * , 295 ( 1995 ) ; h. forkel , braz .
* 34 * , 875 ( 2004 ) ; _ qcd glueball sum rules and vacuum topology _ , proceedings of continuous advances in qcd , minneapolis , 2006 , p. 283
, world scientific , singapore ( 2007 ) , arxiv : hep - ph/0608071 .
kronfeld , m.l .
laursen , g. schierholz , and u .- j .
wiese , nucl .
* b 292 * , 330 ( 1987 ) ; a.s .
kronfeld , m.l .
laursen , g. schierholz , c. schleidermacher and u .- j .
wiese , nucl .
* b 305 * , 661 ( 1988 ) ; t.a .
degrand and a. hasenfratz , nucl . phys .
* b 520 * , 301 ( 1998 ) ; e. vicari and h. panagopoulos , phys .
rep . * 470 * , 93 ( 2009 ) .
b. alls , m. delia and a. di giacomo , nucl . phys .
* b 494 * , 281 ( 1997 ) [ erratum ibid . *
b 679 * ( 2004 ) 397 ] ; c. gattringer , r. hoffmann and s. schaefer , phys .
b * 535 * , 358 ( 2002 ) ; b. lucini , m. teper and u. wenger , arxiv : hep - lat/0401028 .
r. akhoury and v.i .
zakharov , phys .
b * 438 * , 165 ( 1998 ) [ arxiv : hep - ph/9710487 ] ; k.g .
chetyrkin , s. narison and v.i .
zakharov , nucl .
* b 550 * , 353 ( 1999 ) ; v.i .
zakharov , nucl .
b ( proc . suppl . ) * 164 * , 240 ( 2007 ) . | we numerically simulate ensembles of su@xmath0 yang - mills dimeron solutions with a statistical weight determined by the classical action and perform a comprehensive analysis of their properties as a function of the bare coupling .
in particular , we examine the extent to which these ensembles and their classical gauge interactions capture topological and confinement properties of the yang - mills vacuum .
this also allows us to put the classic picture of meron - induced quark confinement , with the confinement - deconfinement transition triggered by dimeron dissociation , to stringent tests . in the first part of our analysis we study spacial , topological - charge and color correlations at the level of both the dimerons and their meron constituents . at small to moderate couplings ,
the dependence of the interactions between the dimerons on their relative color orientations is found to generate a strong attraction ( repulsion ) between nearest neighbors of opposite ( equal ) topological charge .
hence the emerging short- to mid - range order in the gauge - field configurations screens topological charges . with increasing coupling this order weakens rapidly , however , in part because the dimerons gradually dissociate into their less localized meron constituents . monitoring confinement properties by evaluating wilson - loop expectation values
, we find the growing disorder due to the long - range tails of these progressively liberated merons to generate a finite and ( with the coupling ) increasing string tension .
the short - distance behavior of the static quark - antiquark potential , on the other hand , is dominated by small , instanton - like dimerons .
string tension , action density and topological susceptibility of the dimeron ensembles in the physical coupling region turn out to be of the order of standard values .
hence the above results demonstrate without reliance on weak - coupling or low - density approximations that the dissociating dimeron component in the yang - mills vacuum can indeed produce a meron - populated confining phase .
the density of coexisting , hardly dissociated and thus instanton - like dimerons seems to remain large enough , on the other hand , to reproduce much of the additional phenomenology successfully accounted for by non - confining instanton vacuum models .
hence dimeron ensembles should provide an efficient basis for a rather complete description of the yang - mills vacuum . |
in this section we sketch the algorithm proposed in the main text of the letter , give a proof on the validity of the measure used to quantify the accuracy of approximate eigenstates , and present some further results supporting the accuracy of the proposed method .
in table [ table : algorithm ] we provide a practical algorithm for the variational optimization of the nrg states .
let us assume we are interested in the @xmath54 lowest energy states of a system on sites @xmath92 which we write as @xcite @xmath93 s_1 } \cdots \mathbf{a}^{[n ] s_n } { { \vert\mu\rangle } } \bigotimes_{j=1}^{n } { { \verts_j\rangle}}. \label{eq : nrgmps}\ ] ] here , the left boundary vector is a trivial vector @xmath94 and @xmath95 where the one is at the position @xmath96 .
we will only consider the case without employing symmetries as the generalization is straight forward .
we will use a mpo representation for the hamiltonian operator , @xmath97 s'_1 , s_1 } \cdots \mathbf{o}^{[n ] s'_n , s_n } { { \vert\mathtt{i}\rangle } } \bigotimes_{i=1}^{n } { { \verts'_j\rangle } } { { \langles_j\vert}}.\ ] ] we assume real arithmetics , the generalization to complex arithmetics is trivial . * variational nrg algorithm * 1 . perform @xmath3 nrg steps to obtain @xmath98 + ( or just choose some random states ) .
2 . choose a site @xmath99 , initialize @xmath100}$ ] and @xmath101}$ ] .
3 . calculate / retrieve tensors @xmath102 and @xmath103}_{r , q , r'}$ ] .
4 . define @xmath104 , @xmath105 and @xmath106 and minimize @xmath107 5 .
@xmath20 s}_{l , r } : = [ \mathbf{x}]_{(l , s),r}$ ] 6 .
if converged , then stop . 7 . if ( @xmath108 ) , update @xmath109 and recalculate @xmath101}$ ] .
@xmath110 or @xmath111 and go to @xmath112 . in the algorithm
, we employ the fundamental concepts of the matrix product state formalism , such as calculating the blocks in a recursive way , @xmath113}_{r , q , r ' } = \sum_{l , l',s , s',p } l^{[j-1]}_{l , p , l ' } a^{[j-1 ] s}_{l , r } a^{[j-1 ] s'}_{l',r ' } o^{[j-1 ] s',s}_{p , q}\ ] ] and similarly for the right block @xmath103}_{r , q , r'}$ ] .
the starting block is trivial , @xmath100 } \equiv \mathbf{i}_{1\times 1 \times 1}$ ] .
the only step which differs from the mps formalism is initialization of the right block @xmath101 } \in \mathbb{r}^{m \times 1 \times m}$ ] which now contains the sum over @xmath54 energies and can be written as @xmath114}_{r , q , r ' } \equiv \delta_{q,0 } \delta_{r , r ' } w_r.\ ] ] where @xmath115 is some chosen weight function , e.g. @xmath116 ( no weighting ) or @xmath117 ( position weighting ) or @xmath118 ( energy weighting ) . in case of energy weighting
, we use the initial approximation for the energies and re - calculate the boundary block after each sweep . for the unitary optimization we formally reshape tensors @xmath119 }
\cdot \underline{r}^{[j]}$ ] , @xmath120}$ ] , and @xmath26}$ ] to matrices @xmath104 , @xmath105 and @xmath106 as @xmath121_{(l , s),(l',s ' ) } = \sum_{p } l^{[j]}_{l , p , l ' } o^{[j ] s ' s}_{p , q}$ ] , @xmath122_{i , j } \equiv r_{i , q , j}$ ] and @xmath123_{(l , s),r } \equiv a^{[j ] s}_{l , r}$ ] .
the idea behind this matrix reshaping is that we can now use some black - box routine which minimizes the following cost function @xmath124 under the constraint that @xmath125 .
the algorithm which gives the best results according to our experience was proposed in ref .
@xcite .
_ relation to the nrg .
_ the mps description of our description ( [ eq : nrgmps ] ) is tightly connected to the nrg algorithm where one diagonalizes an effective hamiltonian @xmath126 ( eq .
42 in @xcite ) and obtains eigenvalues @xmath127 and eigenvectors @xmath128 which are related to the eigenvectors of the smaller system ( @xmath129 sites ) as ( eq .
43 in @xcite ) @xmath130 the coefficients @xmath131 in the above relation are identical to the tensor coefficients @xmath132 s}_{r , w}$ ] used in ( [ eq : nrgmps ] )
. _ relation to dmrg with targeting_. the main difference between the dmrg with targeting ( dmrgt ) and the vnrg is that the former optimizes the set of states by moving the external bond along the chain whereas in the latter the external bond is fixed to a chosen site . in the dmrgt ,
the energies ( and thus our cost function ) do not monotonically decrease when moving the bond along the chain but there exists a site in the chain where the energies are minimal .
typically , in models with translation invariance such as the ising model or the heisenberg model , the highest accuracy is reached when at the center of the chain . in impurity models with falling hopping terms , however , the optimal site to put the external leg is ( for long chains ) the end of the chain - in which case the cheaper variant ( scaling as @xmath133 ) of the vnrg is preferable .
we have described the cheaper variant of the vnrg in the manuscript but we have skipped the more expensive one where the external leg is associated with some inner site .
in fact , to reduce computational complexity , we do not actually associate the external leg to a site but rather introduce an additional 3-dim tensor on the bond connecting two sites : the third leg ( the external one , similarly to the `` physical '' leg in the matrix product state ) plays the role of the external leg .
this way , the optimization of the more expensive variant of the vnrg is as follows .
all tensors associated to physical sites are optimized as in the cheaper vnrg , that is minimizing the cost function ( [ eq : cost ] ) under a unitary constraint .
this in total costs @xmath134 for one sweep along the chain .
the additional tensor which selects the eigenstates is optimized in the same way as all tensors in the dmrg with targeting .
the dmrg with targeting involves finding @xmath54 lowest eigenstates of a sparse @xmath135 matrix and similarly in our case where we seek @xmath54 lowest eigenstates of a sparse @xmath136 matrix .
the cost of the dmrgt is thus @xmath137 for one sweep whereas the equivalent step contributes @xmath138 to the vnrg , just once in a sweep .
_ statement . _
let @xmath139 be a real symmetric linear operator with eigenvectors @xmath140 such that @xmath141 , and let @xmath142 be an approximation to @xmath143 with fidelity @xmath144 . if @xmath145 , then @xmath146 where @xmath147 is the spectral gap to the nearest eigenlevel .
we have assumed that all considered states are normalized with respect to the euclid norm .
we assume real algebra ( generalization to complex algebra is trivial ) .
_ let us write the approximation @xmath148 as @xmath149 where @xmath150 , @xmath151 and @xmath152 .
it is easy to show that the measure @xmath153 can be written as @xmath154 or as an expression for @xmath155 as @xmath156 the second term on r.h.s is upper bounded by @xmath157 , @xmath158 the denominator is lower bounded by @xmath159 which , assuming @xmath160 and thus @xmath161 , finally gives us the upper bound on @xmath155 and the lower bound on fidelity @xmath162 ( for @xmath163 ) as @xmath164 the only assumption we made was @xmath160 and @xmath165 .
approximate eigenstates of the siam on @xmath81 ( @xmath85 ) sites obtained by nrg and vnrg and total number of kept states @xmath166 .
we have set @xmath167 , @xmath168 , and @xmath82 and used ( sub - sector ) bond dimension @xmath169 . the basis for comparison are the nrg results with @xmath170 kept states ( and @xmath171 ) . ] on figure [ fig : fidelity ] we present the fidelity for the interacting single impurity anderson model ( siam ) with respect to very precise results obtained with the total number of states @xmath170 and a bond dimension @xmath171 in individual subsectors , such that the actual number of states in individual subsectors after truncation to in total @xmath54 states is always smaller than @xmath0 and thus no additional truncation is made in subsectors .
we compare the results obtained with @xmath172 and @xmath173 , in both cases again using a sufficiently large @xmath0 .
let us stress that we compare the nrg and vnrg states to the very precise states obtained by the nrg which were not further optimized in order to avoid possibly unfair comparison with preference for vnrg ( no change in the fidelity is observed even if we do optimize the results for @xmath170 ) .
we observe that the accuracy of the approximate eigenstates is always improved by the variational method proposed in the manuscript .
the improvement is most significant for those states with a low energy , since the higher excited states are more entangled(see e.g. v. alba , m. fagotti , p. calabrese , j.stat.mech.0910 , p10020 , * 2009 * for an analysis on integrable spin systems ) and require a larger bond dimension in the description with matrix product states .
+ + a single approximate eigenvalue in general does not tell anything about the quality of the corresponding approximate eigenvector except for the ground state ( or the state on the opposite side of the spectrum ) .
we consider the first @xmath54 lowest lying states , so the situation might be similar also in our case but we are not able to prove it . on the other hand
, good eigenvectors automatically also mean good eigenvalues , seen from the fact that the eigenvalue error is bounded from above by @xmath174 we observe that the results for the difference in energies , @xmath175 , and the measure @xmath176 are qualitatively very similar , as shown on fig .
[ fig : energydiff ] where we plot the error of energies for the same data used as in figs . 3 and 4 in the letter .
we also test the variational nrg method on an example of quantum ising chain in critical transverse magnetic field . similar to the tilted ising presented in the letter
, we first calculate @xmath0 excited states with a bond dimension @xmath0 by means of the dmrg with targeting and then improve the results using the variational scheme .
the quantum ising chain in transverse field is a quadratic model and as such exactly solvable which allows us to compare the obtained energies ( eigenvalues ) to the exact values .
we observe ( fig .
[ fig : criticalising ] ) a very good qualitative agreement between the measure @xmath177 which quantifies the accuracy of the eigenstates , and the absolute errors of the energies .
note that in this case , the improvement on the dmrg is more significant , since the model is critical and the excited states carry more entanglement - and thus demand a higher bond dimension in the mps description . for the quantum ising chain of @xmath51 sites in critical magnetic field.,title="fig : " ] + for the quantum ising chain of @xmath51 sites in critical magnetic field.,title="fig : " ]
the accuracy of the expectation values is connected to the fidelity .
the zero temperature green function is determined by matrix elements @xmath178 and the corresponding energy differences @xmath179 where @xmath180 is the energy of the ground state , and also @xmath181 and the corresponding energy differences , see @xcite for details . to obtain the complete green function
, one would have to merge contributions from all nrg patches ( i.e. for all system sizes ) which is beyond the scope of this manuscript .
we instead only calculate the ten largest matrix elements mentioned before and compare the accuracy before and after the variational optimization and show the results in table [ fig : table1 ] .
we observe that by the variational optimization we gain roughly one digit of accuracy in the matrix elements . as seen from the matrix elements , the only states that contribute to the green functions are those which have one particles less or more than the ground state , that is the states which live in the subsectors @xmath183 and @xmath184 , respectively
. we may therefore restrict the optimization to only these two sector which enhances the accuracy and reduces the computational costs .
the ground state can be optimized independently from that by the dmrg or mps techniques . in fig .
[ fig : subsector ] we show that improvement of the energies in the subsector containing the state @xmath185 where the basis of comparison are the nrg results with @xmath170 . in this case
we use sufficiently large sub - sector bond dimension @xmath171 such that no additional truncation is done in the sub - sectors and we take the nrg results where in total @xmath166 states were kept . for finite - temperature green functions we also need excited states in other symmetry subsectors with respect to the total particle number and total @xmath186 .
however , they can all be optimized independently and in parallel . | the numerical renormalization group ( nrg ) is rephrased as a variational method with the cost function given by the sum of all the energies of the effective low - energy hamiltonian .
this allows to systematically improve the spectrum obtained by nrg through sweeping .
the ensuing algorithm has a lot of similarities to the density matrix renormalization group ( dmrg ) when targeting many states , and this synergy of nrg and dmrg combines the best of both worlds and extends their applicability .
we illustrate this approach with simulations of a quantum spin chain and a single impurity anderson model ( siam ) where the accuracy of the effective eigenstates is greatly enhanced as compared to the nrg , especially in the transition to the continuum limit .
the density matrix renormalization group @xcite ( dmrg ) , devised to improve on the wilson s numerical renormalization group ( nrg ) @xcite , has become the method of choice to simulate one - dimensional quantum many - body systems at zero temperature and has found a large number of applications in the fields of the condensed matter physics , quantum chemistry and quantum information theory where it turned out that dmrg is essentially equivalent to simulating quantum systems in terms of matrix product states ( mps ) @xcite . in all these methods ,
a quantum many - body state is represented by associating matrices to local configurations at sites and the coefficients in the expansion over the configuration states are given as products of the corresponding matrices .
the dmrg can be used to calculate not only the ground state but also the excited states through the concept of _ targeting _ where these matrices are chosen such that they well represent many states at the same time . in the nrg ,
on the other hand , the low energy states of a system are expressed in terms of the low energy states of a smaller system which is a suitable description of impurity systems with energy scale separation . in this context
, the nrg gives remarkably good results and has retained the position as a widely used impurity solver @xcite whereas the dmrg is rarely used to calculate the excited states @xcite except for the spectral gap .
one of the reasons is that the nrg is less costly as it provides @xmath0 excited states at the cost of @xmath1 as compared to dmrg with targeting with the cost @xmath2 .
presently , the dmrg is mostly used in its time dependent form ( see e.g. @xcite ) which is also true for impurity systems @xcite where also other density - matrix related concepts are used @xcite .
still , dmrg s remarkable ability to calculate and optimize many excited states has not been used in this context .
even as an impurity solver , the nrg has certain limitations : the hopping terms should fall off sufficiently fast which means lower resolution of the spectral densities at higher energies e.g. hubbard bands .
otherwise it becomes inaccurate for longer chains which is seen as a violation of the friedel sum rule @xcite .
it also becomes highly expensive for many - band impurity problems whereas an additional self - consistency constraint in the context of dynamical mean - field theory , see e.g. @xcite , calls for more accurate impurity solvers .
what makes the nrg really different from the dmrg is that it does not provide any feedback mechanism to optimize the matrices by sweeping along the chain and a method along this line has already been used with quantum fields @xcite . in this letter
, we identify the cost function in the nrg algorithm and introduce a feedback mechanism by which the states can be optimized in a variational way under the original nrg cost function .
we relate the nrg and the dmrg by identifying the common cost function in both approaches show that the proposed scheme improves the results of both methods while retains the lower , nrg - like , scaling of computational costs .
_ numerical renormalization group .
_ in the context of the nrg , the lowest energy eigenstates of a quantum many - body system on @xmath3 sites , @xmath4 , are approximated by orthonormal states @xmath5 } \equiv \ { \psi^{[n]}_1 , \ldots , \psi^{[n]}_{d_{n}}\}$ ] defining an effective low energy hamiltonian @xmath6 } = \sum_{\alpha=1}^{d_n } e^{[n]}_\alpha { { \vert\psi^{[n]}_\alpha\rangle } } { { \langle\psi^{[n]}_\alpha\vert } } \label{eq : heff}\ ] ] with effective energies @xmath7}_\alpha = { { \langle\psi^{[n]}_\alpha \vert } } \hat{h } { { \vert\psi^{[n]}_\alpha\rangle}}$ ] .
the set of effective states @xmath5}$ ] is obtained recursively from @xmath8}$ ] , the effective description of a system on @xmath9 sites , extended by an extra site of a local dimension @xmath10 , resulting in an extended hamiltonian @xmath11 } \in \mathbb{c}^ { ( d_{n-1 } d_n ) \times ( d_{n-1 } d_n ) } $ ] that , in order to prevent exponential growth , has to be projected to a subspace of a smaller dimension @xmath12 . in the nrg , this subspace is spanned by the lowest energy eigenvectors as @xmath13 } = \mathbf{u}^h \mathbf{h}_{\rm ext}^{[n ] } \mathbf{u}$ ] where @xmath13 } \in \mathbb{c}^ { d_n \times d_n } $ ] is a new effective hamiltonian and @xmath14 is an isometry matrix @xmath15 .
the set of effective states @xmath16}_\alpha \}$ ] is given as a mps with an external @xmath17-index @xcite @xmath18}\rangle } } = \sum_{\{s_j\ } } \mathbf{e}_0 \cdot \mathbf{a}^{[1 ] s_1 } \mathbf{a}^{[2 ] s_2 } \cdots \mathbf{a}^{[n ] s_n } \mathbf{e}_\alpha { { \verts_1,\ldots s_n\rangle } } \label{eq : psin}\ ] ] with the right boundary vector @xmath19 enumerating basis states .
the set @xmath5}$ ] is orthonormal due to the isometry constraint for matrices @xmath20}$ ] defined as @xmath21 } ] _ { ( l s ) r } \equiv a^{[j ] s}_{l r}$ ] .
the nrg projection is equivalent to requiring that the _ sum _ of the new effective energies @xmath22}$ ] is minimal which , according to ( [ eq : heff ] ) , corresponds to the trace of @xmath23}$ ] and leads to a cost function @xmath24 } \mathbf{a } ) \quad\textrm{where}\quad \mathbf{a}^h \mathbf{a } = \mathbf{1 } \label{eq : cost1}\ ] ] which is minimized exactly by the nrg isometry @xmath25 .
this cost function is different from other dmrg - based approaches ( e.g. @xcite ) where the cost is determined from the ground state alone .
_ variational optimization . _ the identification of the nrg cost function allows us to optimize not only the tensor associated with the last site in the nrg - mps ( [ eq : psin ] ) but an arbitrary tensor in the chain . ) in the optimization of a tensor @xmath26}$ ] ( circles ) under the hamiltonian matrix product operator ( squares ) . ] the cost @xmath27 } ) = \sum_{\alpha } { { \langle\psi^{[n]}_\alpha\vert } } \hat{h } { { \vert\psi^{[n]}_\alpha\rangle}}$ ] can easily be understood as a tensor network depicted on fig .
[ fig : trh ] where the @xmath17-index enumerating states @xmath28}_{\alpha}\rangle}}$ ] is contracted with the @xmath17-index in the conjugate mps @xmath29}_\alpha\vert}}$ ] .
the cost can thus be minimized by varying an arbitrary isometry @xmath30}$ ] in the nrg - mps ( [ eq : psin ] ) which we will illustrate by expressing the hamiltonian as a matrix product operator ( any 1d operator can be written as a mpo ) @xmath31 \underline{s}_1 } \cdots \mathbf{h}^{[n ] \underline{s}_n } \mathbf{e}_0\ , { { \verts_1',\ldots , s_n'\rangle } } { { \langles_1,\ldots , s_n\vert } } $ ] where tensors @xmath32}$ ] are represented by squares on fig .
[ fig : trh ] .
an arbitrary isometry @xmath20}$ ] is extracted out of the cost function ( [ eq : cost1 ] ) which now takes a form @xmath33 } \mathbf{a } \mathbf{r}_\gamma^{[j]\ , t } \mathbf{a}^{h } ) \quad\textrm{for}\quad \mathbf{a}^{h } \mathbf{a } = \mathbf{1 } \label{eq : gencost}\ ] ] where @xmath34}$ ] and @xmath35}$ ] correspond to the contractions of the tensor networks on the left and right side of the chosen site @xmath36 , respectively ( fig .
[ fig : trh ] ) , explicitly defined as @xmath37}]_{(\alpha ' s')(\alpha s ) } = l_{\alpha s \gamma s ' \alpha'}$ ] and @xmath38}]_{\beta ' \beta } = r_{\beta\gamma\beta'}$ ] . furthermore
, a modified weighted cost function may be considered where the effective states are weighted according the their importance , e.g. by a boltzmann factor @xmath39}_{\alpha}}$ ] ( see fig .
[ fig : trh ] ) .
the cost ( [ eq : gencost ] ) is minimized using the conjugate gradient method with a unitary constraint @xcite in a variational way ( i.e. the sum of energies can only decrease ) and the computational costs scale as @xmath1 , same as in the original nrg .
the number of optimization steps depends on the quality of the initial state and the desired accuracy .
_ density matrix renormalization group .
_ a special property of the nrg - mps states is a two - fold nature of the external @xmath17-index which acts both as an enumerating index as well as a virtual bond in extending the system for a site . for a fixed system size
, the external index can be associated with any site in the chain and moved along the chain by means of the singular value decomposition ( svd ) .
interestingly , the concept of a moveable external index is intrinsic to the finite size dmrg algorithm _ with targeting _ @xcite .
the basic concept of the finite size dmrg is as follows : split the system as @xmath40 , find the optimal tensor at site @xmath36 and move to the next site @xmath41 .
the way one finds the optimal tensor at site @xmath36 is by considering the reduced density matrix ( dm ) for @xmath42 , obtained by tracing the dm of the complete system ( `` universe '' ) over the environment @xmath43 whereas the universe can be either in a pure state ( ground state ) or in a mixed state of lowest energy states as is the case with the targeting .
formally , the eigenstates of the universe can be written as @xmath44 where the index @xmath17 enumerates the states and the optimal tensor @xmath26}$ ] is obtained by the svd as @xmath45 s_j } _ { l c } \sigma_c v_{c s_{j+1 } \alpha r}$ ] .
identifying @xmath46 s_{j+1 } \alpha}_{l , r } \equiv \sigma_l v_{l s_{j+1 } \alpha r}$ ] , we recover a nrg - mps with the index @xmath17 associated with the site @xmath47 ( instead of site @xmath3 ) .
however , this part is ignored in the dmrg since the tensor at site @xmath47 is obtained in the subsequent step .
therefore , the index @xmath17 is intrinsic to but hidden in the dmrg and is carried back and forth along the chain . in the dmrg ,
only the central sites are optimized ( the boundary parts are treated exactly ) .
we can however sweep to the very end of the chain and end up exactly with the nrg - mps ( [ eq : psin ] ) where the last tensor @xmath48}$ ] carries the @xmath17 index which now allows us to enlarge the chain for an extra site by a nrg step .
looking back , we realize that the optimal tensor @xmath49 is an isometry matrix minimizing the same cost function ( [ eq : cost1 ] ) as in the nrg context , just that it represents an arbitrary pair of neighboring sites instead of the last site only .
hence , the dmrg minimizes the same cost function as the nrg .
we also realize that the truncation in the svd decreases the accuracy of states and it is not guaranteed that the optimization at the next site will recover the same accuracy ( in fact the highest accuracy is reached in the center of the chain )
. therefore , the set of excited states obtained by the dmrg can be further optimized by means of the variational principles proposed earlier .
_ results . _
the variational optimization algorithm for a nrg - mps set of states is put to the test on two qualitatively distinct models .
first , we consider a quantum ising chain in a tilted magnetic field @xmath50 which is a simple example of non - integrable quantum spin chains . ) on @xmath51 sites with @xmath52 , obtained by the dmrg ( red ) and optimized by the variational method ( blue ) , for various bond dimensions @xmath0 . ]
the accuracy of the approximate eigenstates of ( [ eq : tilted ] ) is quantified by a measure @xmath53 which puts a lower bound to the fidelity ( see supplementary material .
we consider a chain of @xmath51 sites , obtain the excited states by means of the dmrg and optimize them using the variational approach .
the dmrg provides provides highly accurate results as seen from the fig .
[ fig : tilted1 ] , leaving little space for improvements . however , the results are not optimal and can be improved by the variational nrg .
higher improvement is observed for smaller bond dimensions @xmath0 which suggests a possibility of improving the excited states in the regime where sufficiently large @xmath0 are not reachable due to higher entanglement , such as in two - dimensional or critical ( see supp .
material ) systems .
it is also worth noting that the computational costs of one dmrg sweep with targeting @xmath54 states scale as @xmath55 , compared to either @xmath56 or @xmath57 for the variational optimization where the @xmath17-index is associated with an inner or a boundary site , respectively .
the pre - factors to @xmath58 are similar ( related to the number of lanczos steps ) . as the second example we consider the single impurity anderson model ( siam ) which , after a logarithmic discretization @xcite of the conductance band , is described by a semi - infinite linear chain where the impurity ( @xmath59 ) is coupled to a chain of fermions ( @xmath60 ) as @xmath61 with @xmath62 , @xmath63 , @xmath64 , and @xmath65 ( see @xcite for details ) .
the nrg hamiltonian @xmath66 presents an effective description of the anderson problem in the energy interval @xmath67 $ ] and only the complete sequence @xmath68 gives the complete spectrum .
as mentioned in the introduction , the nrg works best for strong scale separation , i.e. @xmath69 which however take one further away from the continuum limit @xmath70 and yield lower resolution at higher frequencies .
contrarily , the dmrg does not rely on energy scale separation and thus quickly falling hopping terms but is numerically more costly , scaling as @xmath2 ( here @xmath71 ) .
this suggests a great potential in optimizing the _ existing _ nrg set of states using the variational method proposed in this letter where the costs only scale as @xmath1 , as in the nrg , but , like in the dmrg , no scale separation is required .
let us first consider the exactly solvable non - interacting siam hamiltonian with @xmath72 with a discretization parameter @xmath73 . ) on @xmath74 sites ( @xmath75 ) for nrg , dmrg , and the variational nrg , for @xmath76 .
we have set @xmath77 , @xmath78 , @xmath79 . ]
we observe from fig .
[ fig : nisiam ] that for a fixed bond dimension of a nrg - mps , the accuracy of eigenstates obtained by the dmrg and the variational optimization ( vnrg ) is several orders of magnitude better than the nrg results .
qualitatively the same results are obtained for the absolute errors of energies ( see supplementary material ) .
we now consider the interacting siam hamiltonian ( [ eq : siam ] ) with @xmath80 on @xmath81 sites where , unlike the non - interacting case , we employ the symmetries ( total particle number of either spin ) .
we analyze the improvement of the first one thousand nrg states after variational optimization ( fig . [ fig : siam ] ) in the regime where nrg already produces accurate results due to scale separation ( @xmath82 ) . due to the symmetries
, the bond dimension @xmath83 ( which determines the computational complexity ) now refers to the maximal number of states in individual subsectors whereas the total number of kept states is denoted by @xmath84 .
states for the interacting siam chain ( [ eq : siam ] ) on @xmath81 sites ( @xmath85 ) , for nrg and the variational nrg _ with symmetries_. sub - sector bond dimensions are taken @xmath86 , truncated to @xmath87 states of lowest energy .
we have set @xmath82 , @xmath88 , @xmath89 and @xmath90 . ] as shown on fig .
[ fig : siam ] , the accuracy of eigenstates is greatly enhanced by the optimization . while the cost of an optimization step is comparable to the cost of a nrg step , a smaller bond dimension is needed to represent states using the vnrg as compared to the nrg .
this effect is more pronounced for small values of @xmath91 while for larger values of @xmath91 the nrg already gives highly accurate results , only requiring little improvement .
the results are consistent also for the fidelity of excited states and the absolute errors of energies ( see supplementary material ) .
we stress that the logarithmic discretization is not needed for the variational scheme proposed in this letter . _ conclusion .
_ we have proposed a variational method to simulate or optimize the effective low energy description of one dimensional quantum systems , introducing a feedback mechanism to the nrg and identifying the cost function .
furthermore , we have expressed the dmrg method with targeting as a nrg method with a moveable external index enumerating states and shown that the dmrg results can be further improved by the variational optimization .
we have tested the methods on the quantum ising chain in a tilted magnetic field and the single impurity anderson model where we observed significant improvement of the approximate low energy eigenstates .
the proposed technique can be directly applied to optimize existing nrg results @xcite and could be a beneficial improvement of impurity solvers in the context of dynamical mean - field theory , see e.g. @xcite . _ acknowledgments .
_ we acknowledge fruitful discussions with h. g. evertz , a. gendiar , k. held , t. pruschke , g. sangiovanni , a. toschi , and r. itko , and financial support by the eu project quevadis and the fwf sfb project vicom .
the computational results were in part achieved using vienna scientific cluster .
10 s. r. white , phys .
rev .
lett . * 69 * , 2863 ( 1992 ) .
k. g. wilson , rev .
mod . phys . * 47 * , 773 ( 1975 ) .
s. stlund and s. rommer , phys .
rev .
lett .
* 75 * , 3537 ( 1995 ) ; g. vidal , phys .
rev .
lett . * 93 * , 040502 ( 2004 ) ; f. verstraete , d. porras , and j. i. cirac , phys .
rev .
lett . * 93 * , 227205 ( 2004 ) .
r. bulla , t. a. costi , and t. pruschke , rev .
mod .
phys . *
80 * , 395 ( 2008 ) .
i. schneider , a. struck , m. bortz , and s. eggert , phys .
rev .
lett .
* 101 * , 206401 ( 2008 ) .
u. schollwoeck , rev .
mod .
phys .
* 77 * , 259 ( 2005 ) .
s. nishimoto , f. gebhard , and f. jeckelmann , j. phys . : condens .
matter * 16 * , 7063 ( 2004 ) .
a. weichselbaum , f. verstraete , u. schollwck , j. i. cirac , and j. von delft , phys .
rev .
b * 80 * , 165117 ( 2009 ) .
m. karski , c. raas , and g. s. uhrig , phys .
rev .
b * 77 * , 075116 ( 2008 ) .
e. jeckelmann , phys .
rev .
b * 66 * , 045114 ( 2002 ) . c. raas and g. s. uhrig , eur . phys .
j. b * 45 * , 293 ( 2005 ) .
h. saberi , a. weichselbaum , and j. von delft , phys .
rev .
b * 78 * , 035124 ( 2008 )
. w. hofstetter , phys . rev . lett . * 85 * , 1508 ( 2000 ) . f. b. anders and a. schiller , phys .
rev .
lett .
* 95 * , 196801 ( 2005 ) .
r. itko and t. pruschke , phys .
rev .
b * 79 * , 085106 ( 2009 ) .
a. georges , g. kotliar , w. krauth , and m. j. rozenberg , rev .
mod .
phys . * 68 * , 13 ( 1996 ) .
r. m. konik and y. adamov , phys .
rev . lett . * 98 * , 147205 ( 2007 ) ; g. p. brandino , r. m. konik , and g. mussardo , j. stat . mech . * ( 2010 ) * p07013 .
t. abrudan , j. eriksson , and v. koivunen , signal processing * 89 * , 1704 ( 2009 ) .
h. g. evertz , m. ganahl , k. held , i. piorn , t. pruschke , and f. verstraete , _ in preparation_. |
network routing based on content identifiers has recently become a topic of extensive discussion , due to the benefits that could be provided by a location - independent data distribution network @xcite , more commonly referred to as an information - centric network ( icn ) . for instance , the icn _ request - response _ mode of operation alleviates client mobility issues @xcite and natively supports interdomain multicast @xcite .
furthermore , content security ( as opposed to channel security ) is inherently supported by transmitting signed copies of content @xcite .
this in turn allows for in - network caching , which can transform the internet into a native content distribution network @xcite .
finally , as shown recently @xcite , the icn paradigm can bring benefits also at the transport layer , where caches can be exploited to alleviate congestion .
on the other hand , enormous effort has been spent to de - ossify the end - to - end internet transmission model to enable new functionalities .
examples include ip multicast and anycast and supporting ip mobility at the network layer @xcite , @xcite , @xcite . however , the difficulties of deploying those solutions at large scale led to the design of application - layer solutions such as overlay caching instead of native , in - network caching , overlay indirection techniques @xcite , @xcite , @xcite , dnssec and ipsec to enhance security , just to name a few . even though these solutions have the potential to enable new services ( or applications ) , they appear inferior compared to an icn mode of operation , as they can not natively support security , mobility , in - network caching and multicast : in all cases the in - network forwarding entities are forced to operate on the five - tuple ` < sourceip , destinationip , sourceport , destinationport , protocol > ` being , therefore , completely content - agnostic . arguably , the icn paradigm has the potential to deal with the internet s most daunting problems in a native manner . to reach this point ,
however , a new architecture based on core icn principles will have to be deployed over the current ip internet architecture , clearly , a rather challenging task . in this paper , we identify two main obstacles that hinder the deployment of icn on top of the current internet .
these are : _
i ) _ scalability of name resolution , a core networking problem @xcite and _ ii ) _ content provider - controlled access to content , a business model problem , which however , is deeply integrated into the core networking principles of today s internet and therefore , affects the design of any new architecture .
content access control here is linked to content access logging and the transmission of content transparently to the content provider from in - network caches .
we discuss each of these two challenges in more detail next .
based on these considerations , in this paper , we propose a fully backward - compatible and incrementally - deployable icn - oriented architecture that meets scalability concerns , but at the same time takes into account the business requirements of the main internet market players .
two main schools of thought have emerged in the icn - related literature regarding name resolution and name - to - location mapping .
the first one , mainly adopted by the original ccn / ndn proposal @xcite , advocates the hop - by - hop resolution of _ requests _ or _ interests _ at the data plane .
effectively , name resolution is coupled with name - based forwarding with each interest packet being locally resolved to the next ( router ) hop .
this approach has the advantage of locally making forwarding decisions , but on the downside , huge volumes of state need to be maintained in ( manually - set ) fib tables @xcite .
ccn / ndn routers effectively have to keep state _ per packet _ ,
an issue traditionally considered as an implementation challenge @xcite . to deal with
the scalability problems @xcite of the original proposal and the huge state that needs to be kept at all routers , recent developments in the ndn space have proposed an ndn - based dns system , dubbed ndns @xcite , as well as the involvement of content providers to help in the name - resolution process @xcite .
the second school of thought decouples name - resolution from name - based routing by using a separate name - resolution system , similar in nature to dns ( _ e.g. _ , @xcite , @xcite , @xcite , @xcite ) .
although this approach avoids pushing excessive state to router forwarding tables , it requires the deployment of new infrastructure by operators .
for instance , as shown in @xcite , the support of the dona @xcite architecture at tier-1 autonomous systems ( ases ) requires the deployment of small - to - medium size data centres to support name resolution . such , extra infrastructure built in from scratch
has the obvious downsides of huge investment requirements , as well as the shift challenge to this new mode of operation .
moreover , focusing on the practical deployment of icn , the full cycle of the name resolution process still remains unclear .
name resolution and data delivery mechanisms often build on the implicit assumption that content names or identifiers are already available to the end users , prior the aforementioned coupled or decoupled resolution steps . obviously , developing a mechanism for the retrieval and delivery of content names to the end users raises concerns regarding both scalability aspects related to the enormous size of the namespace , and compatibility issues with respect to both application and network layer interfaces .
we also note that the requirements of today s dynamic and interactive applications would not be served adequately by fully transparent in - network caching driven solely from search engine content name results .
we discuss and evaluate these concerns later .
content providers ( cps ) and cdns require , for commercial and regulatory reasons , full control over the content requested and transferred .
this has been largely overlooked by research efforts in the icn area , which have mainly focused on naming schemes and name resolution systems to address scalability issues .
for instance , the consensus around opaque and permanent content names ignores the fact that content can be served from isp - operated in - network caches , transparently to the cp or cdn .
_ pay per click " _ business models , however , would face significant limitations from this design choice in an icn setting , that would practically prevent cps and cdns from billing their customers .
alternative approaches based on isp - cdn collaborations to log content requests can not but be unrealistic : dnss can keep track of requested content and could possibly report back to the relevant cps / cdns .
this , however , would mean that slas should be in place between _ all isps and all cps / cdns at a global scale _ , a rather unrealistic assumption .
at the same time , transparent in - network caching mechanisms would typically allow only limited control over the content delivered to clients .
that is , coarse grained ttl - based mechanisms would be the only means for cps / cdns to manipulate updated content , leading either to the delivery of stale content , or the unnecessary delivery from the cp .
that said , active cache purging is another requirement that calls for control of content from cps and cdns .
although content access control might sound as a trivial implementation or a business model issue , we argue that it might well hinder the engagement of cps and cdns from the adoption of icn .
summarising , we argue that these concerns of : _
i ) _ scalability and incremental deployment support of a name - oriented architecture , and _
ii ) _ exclusive content access control at the cp side with simultaneous support for transparent in - network caching have been overlooked by the community so far . as a consequence
, the full potential of an icn mode of operation has not been exploited in full yet , making the adoption and deployment of the icn paradigm an unrealistic target .
although clean slate research has revealed many of the benefits that icn can bring , we argue that deployability has to be put at the forefront of any icn design , rather than being treated as an afterthought .
we address the deployability concerns discussed above by introducing a novel information - focused network architecture , which overcomes scalability concerns and is fully backward compatible with the current ip architecture .
our proposed architecture first introduces a name resolution process tailored to carefully manage information exposure _ e.g. , _ enabling content access logging ( section [ cc::cpbnm ] ) .
this name resolution process is combined with a new naming scheme , which builds on the notion of _ ephemeral names _ ( section [ cc : en ] ) .
_ name resolution is controlled by content providers _ based on a fully backward - compatible mechanism that supports in - network caching and the direct control of ephemeral names lifetime , thus facilitating content access logging and active purging of stale cached data .
the proposed mechanism completes the full cycle of the name resolution process , delivering content names to clients , without imposing any requirement for additional mechanisms . in - network caching , name - based routing and support for network - layer multicast are all integrated in the _ location - independent routing layer _
( lira ) , an extra layer in the protocol stack placed at level 3.5 " of the protocol stack , above the ip and below the transport - layer ( section [ cc : cl ] ) .
lira absorbs " the location - independence nature of icn , leaving the network layer to operate based on ip addresses .
resolution of content names does not rely on large volumes of fib table entries , and routing takes place based on a hybrid of ip addresses ( at the ip layer ) and location - independent transient content names ( at the lira layer ) ( section [ rsn - details ] ) .
our design does not require blanket adoption in order to realise the benefits of icn . instead , isps
can incrementally deploy lira nodes with little investment .
furthermore , the fact that routing is ( in the worst case ) based on ip addresses guarantees full backward compatibility with the current internet architecture .
our results show that even with a subset of nodes upgraded to support lira functionality , our design achieves considerable performance gains ( section [ eval ] ) .
in order to deal with the scalability concerns raised above , we design a name resolution scheme which involves the content provider and does not require extra name - based resolution machinery ( _ e.g. , _
@xcite , @xcite , @xcite , @xcite ) , or manually - set , bloated fib tables ( _ e.g. , _ @xcite , @xcite ) .
in particular , any user will have to consult the cp ( or cdn ) and ask " for the name/`contentid ` before any content transfer can start ( see next section for details on the ` contentid ` ) .
users reach the content provider based on the standard procedure of the current internet , that is , based on urls , dns resolution and ip addresses .
this first part of the resolution ( _ i.e. , _ reaching the cp to get the ` contentid ` ) is based on ip addresses and is location - dependent .
we note that users do not get the whole of the chunk from the cp ( but only the ` contentid ` ) , which can be served from any other cache in the network . in this way , we realise _ semi - transparent _ in - network content caching , which we argue is in the best interests of both cp / cdns and isps alike . as discussed later on in this section ,
the second part of the name resolution , which also leads to the content transfer itself is location - independent , according to the philosophy of icn .
summarising , the _ content provider - controlled name resolution procedure " _ introduced here is fully backward compatible and does not require extra investment from isps , or cps / cdns . to provide full content access control to cps ,
we introduce the concept of _ ephemeral names _ , which are used for location - independent content delivery .
our primary motivation behind the introduction of _ ephemeral names _ is to avoid dissemination of the name/`cid ` of a content to other users , as this could potentially lead to accessing the content from in - network caches , transparently to the cp / cdn .
this section explains the structure , usage and design principles of these names .
the lira architecture uses flat names composed of two parts ( see fig .
[ ephemeral - names ] ) .
the main part of the naming structure , the ` contentid ` , or ` cid ` reflects the name of the content itself and is based on the premise of _ ephemeral or transient names_. according to this concept , content providers choose arbitrary strings and assign them to the content they host .
the names are flat , in the sense that they bear no structure related to routing ( _ e.g. , _ aggregation ) ; however , cps may impose structures related to the internal organisation of their content .
ephemeral names should be unique to guarantee collision - free name resolution , which can be easily achieved with the use of arbitrary hashes .
the names are self - certifying and expire " after some time interval .
this transitioning interval should be coarser than the time needed to support in - network caching and multicast ( _ e.g. , _
names should not change on a per - request basis ) - see section [ transitioning - interval ] for details .
@xmath0 the second part of the _ ephemeral name _
, the ` serviceoptions ` , can be used to realise preferential treatment of content .
although the use of this part of the name is not necessary in our architecture , and is not necessarily of ephemeral nature , we believe that it can help in the caching and scheduling process .
for instance , the ` serviceoptions ` part can be used to flag content that should or should not be cached .
we leave such investigations for future work . in case of permanent names , search engines would operate based on names ( similarly to today s operation based on urls ) .
this operation is clearly not in the best interests of cps / cdns given the _ pay per click " _ models in use today and transparent in - network caches used in icn .
transient names dis - incentivise search engines from disseminating ` cids ` , but at the same time allow for both access logging at the cp / cdn and transparent in - network caching .
one might claim that search engines would prefer to provide the ` cid ` directly to users , as this would lead to faster content access ( _ i.e. , _ users would not need the extra rtt to travel to the cp / cdn to get the ` cid ` ) .
however , given ( i ) the transient character of names , and ( ii ) the delivery of bundles of ` cid`s by cps ( see section [ transitioning - interval ] ) , this would require search engines to devise mechanisms for retrieving and disseminating ` cid`s each time they change , only to save a single rtt in each bundle .
this limits the incentives of search engines to provide ` cid`s without the consent of cps / cdns .
moreover , and most importantly , ephemeral names allow cps / cdns to actively control the cached content served to their clients _
e.g. , _ by changing the ` cid`s of content chunks existing cached copies get practically invalidated .
this is an important feature of the proposed approach , which can not be supported in alternative proposals ( _ e.g. , _
@xcite ) .
the combination of the name resolution at the cp , together with the ephemeral nature of content names supports a number of desirable features .
first and foremost , name resolution is under the control of the cp , enabling access logging .
secondly , versioning of updated content and purging of old content from in - network caches is also under the control of the cp .
although ttl - like techniques , such as the ccn staleness option , can support content updating , it is not easy to set such values given today s interactive applications . setting ttl values for individual content items ( _ e.g. , _
@xcite ) would always face the tradeoff of short ttls resulting in unnecessary delivery from the content provider , while longer ttls would result in delivering outdated content .
using ephemeral names , cached content can instead be actively invalidated when needed . along the same lines ,
the transitioning interval of ephemeral content names is an issue that requires further attention and is related , among others , to the popularity of the content as well as the size of content chunks .
frequent change of the name can result in suboptimal performance , since each change purges the content in caches .
we deal with this tradeoff by setting the transitioning interval of content names to a value inversely proportional to the popularity of the content itself .
popularity is measured by the cp and can be based on the number of requests for the content in question , per some time interval . although more sophisticated settings can be found , with this simple setting for the transitioning interval we avoid changing the ` cid ` of rarely accessed content too frequently , and we also avoid leaving the ` cid ` of popular content the same for too long .
finally , to alleviate the need to travel to the cp for every chunk request , we assume that upon each request for a content item , the cp sends back to the client the up - to - date " ephemeral names of the next few subsequent chunks , that is , not only the name of the immediately following one .
the number of subsequent ` cid`s sent by the cp to the client is left for future investigation .
adding extra functionality , or altering completely the operation of _ existing _ core network protocols can prove difficult to be done incrementally ( _ e.g. , _
ipv6 ) and flag - days " are not an option for incorporating new components at a global scale . for these reasons ,
we propose _ addition _ instead of _ replacement _ of an extra layer to the protocol stack , which we call _ location - independent routing layer _
lira sits on top of the network ( ip ) layer and below the transport layer .
it operates based on _ ephemeral names _ and integrates all the required functionality to realise _ location independence _ , taking advantage of _ information centricity _ and its well - known gains @xcite .
although recent studies have proposed http as the layer that can integrate information or content centricity @xcite , here we argue that in order for in - network caching and multicast to be smoothly incorporated in the new ecosystem , any information - centric operation needs to be _ below the transport layer_. otherwise , the transport protocol can merely connect two specific endpoints cancelling any notion of location - independent content transfer . instead , breaking the end - to - end transmission model below the transport layer allows to leverage ( icn enabled ) in - network caching , both in terms of native multi - source routing and localised congestion control @xcite , going far beyond traditional ip multicast or anycast mechanisms .
lira is implemented in just a small subset of nodes ( see section [ lira - nodes ] ) , which can be transparently planted in the network , and it manages incoming and outgoing content based on their names .
the main name management functionality is implemented in a routing table , which we call _ content forwarding information base _ ( c - fib ) ( section [ rsn - table ] ) .
a similar notion to the lira layer has been proposed in the past in @xcite , but in a totally different context , addressing the exhaustion of ipv4 addresses .
the evolution of nat boxes ( together with the painfully slow incremental deployment of ipv6 ) has dealt with this problem and hence , the related efforts became obsolete .
the content forwarding information base ( c - fib ) table keeps track of recently requested and served content ( in terms of ` cid`s ) and maintains forwarding information used for the delivery of those content items , providing also support for in - network caching and multicast . upon subsequent request(s ) for a content already in the c - fib table , lira is redirecting requests towards the direction where the content has been sent , or served from , similarly in principle to breadcrumbs routing @xcite .
we note that _ the c - fib table essentially acts as a cache _ for ` cid`s served recently through this router ( somewhat similarly to @xcite and @xcite )
. however , c - fib table entries are not permanent , as in ccn s fib , but rather are assisting in location - independent content delivery from neighbouring nodes ( see section [ rsn - details ] for details on the c - fib structure ) .
the typical structure of the c - fib table is illustrated in table [ table : r1routing - table ] .
the table maintains one entry per content chunk .
the following information is maintained for each entry : _
i ) _ ` cid ` , the content identifier of the chunk , _ ii ) _
@xmath1 , the _ incoming interface _ _
i.e. _ , the index of the interface from which the content is received , that is , the content source indicated by the dns , _ iii ) _
@xmath2 , the _ outgoing interface _ _
i.e. , _ the index(es ) of the interface(s ) towards which the content is currently being forwarded , _ iv ) _
@xmath3 , the _ temporary interface _ , _
i.e. , _ the index(es ) of the interface(s ) where the content has been forwarded , _
@xmath4 , the _ multicast ip _ field that holds ip addresses of clients participating in a multicast session .
note that interface entries in the c - fib table denote real interfaces ( _ i.e. , _ directions towards which requests / content should be forwarded ) and not ip addresses of sources / destinations ( apart from the multicast ip field ) . by doing
so we realise the _ location independence _
property of icn in lira .
the lira node structure is the main component of the proposed architecture , which integrates information centricity .
lira nodes implement the lira layer with its c - fib table discussed above in order to realise named content management and subsequently location independence .
lira nodes also include caches that temporarily store named content chunks ( _ i.e. , _ in - network caching ) .
although by default all lira nodes include both the c - fib table and content caches , we also evaluate ( in section [ eval ] ) the case of lighter " lira nodes , where , based on node centrality metrics and to facilitate incremental deployment , some nodes implement the c - fib table and some others implement caches .
our design does not require all nodes of a domain to become lira nodes and it is operational regardless of this .
being always based on ip , nodes fall back to normal ip operation and route towards the direction indicated by location - based addresses .
note that all routers maintain the default ip - based fib table . therefore , incompatibility issues or requirements for simultaneous shift to icn operation do not exist . as we show later on in the evaluation section , an average of 50% of nodes within a domain can provide considerable performance gain .
careful network planning ( _ e.g. , _ depending on topological issues ) and incremental upgrade of normal routers to lira nodes gives a major advantage to the proposed architecture in terms of deployability compared to other icn architectures .
we proceed with the description of the name resolution and content delivery process , illustrated in fig . [
fig : summary ] .
we then give details of the entries of the _ content forwarding information base _ ( c - fib ) table during the content delivery process .
for this purpose we use the network topology presented in fig .
[ example ] .
tables [ table : r1routing - table ] and [ table : r2routing - table ] are also used to present the entries of the c - fib table(s ) for a sequence of important events taking place in our example scenarios ( denoted with timestamp @xmath5 ) .
the name resolution process is initiated through existing protocols ( _ i.e. , _ dns and http ) to guarantee backwards compatibility and facilitate adoption of icn . as a first step ( fig .
[ fig : summary ] ) and identically to what is happening today , users resolve urls through a request to the dns .
the dns responds with the ip address of the content provider .
the user generates an http head request @xcite at the application layer . at this stage , routing is location - dependent and is based on the ip address indicated by the dns . at the content layer , the request is asking for the ` cid ` .
the cp sends back an http response packet containing the up - to - date name , _
i.e. , _ ` cid ` , of the requested content in the etag field of the http response header @xcite .
the destination ip address of that packet is that of the requesting client .
this packet can be piggybacked with data to avoid an extra rtt between the client and the cp . in this case , however , given that requests are sent per chunk , we can not take advantage of in - network caching .
this option can be considered in special cases ( _ e.g. , _ when a client is close to the cp and chances of finding the content cached are slim ) .
the client issues a request for the first chunk of the content object ( _ e.g. , _ client a in the example of fig .
[ example ] ) .
the request includes the ip address of the cp at the ip layer and the ` cid ` of the chunk at the lira layer . [ !
t ] [ ! t ] l |*5c @xmath6 & @xmath7 & @xmath8 & @xmath9 & @xmath3 & @xmath4 + @xmath10 & @xmath11 & 1 & 3 & - & - + @xmath12 & @xmath11 & 1 & - & 3 & - + @xmath13 & @xmath11 & 1 & 2 & 3 & - + @xmath14 & @xmath11 & 1 & - & 2 , 3 & - + @xmath15 & @xmath16 & 1 & 3 & - & - + @xmath17 & @xmath16 & 1 & 2 , 3 & - & b s ip + @xmath18 & @xmath16 & 1 & - & 2 , 3 & - + l |*5c @xmath19 & @xmath7 & @xmath8 & @xmath9 & @xmath3 & @xmath4 + @xmath20 & @xmath11 & 1 & 2 & - & - + @xmath21 & @xmath11 & 1 & - & 2 & - + @xmath22 & @xmath16 & 1 & 2 & - & - + @xmath18 & @xmath16 & 1 & - & 2 & - + lira nodes along the path check the ` cid ` included in the request against the entries of their c - fib table .
if an entry for the ` cid ` exists , then they forward according to this entry . if not , they forward according to the ip address .
the ip address points to the cp , hence , content can always be resolved according to that in the worst case , _ e.g. , _ in case of lira - incompatible nodes or domains . at this point , assuming the content is not locally cached ( see section [ caching ] for details on in - network caching ) , the request is forwarded towards the cp .
the index of the network interface used to forward the request is marked as the @xmath8 for this content chunk ( _ i.e. , _ interface 1 - see time @xmath10 in table [ table : r1routing - table ] ) . at the same time , the index of network interface from which the request was received is marked as an output interface ( interface 3 in our example ) .
the content chunk is then sent back from the cp ( or any other cache further down the path ) . during the data transfer
no change is made in the c - fib table entries of intermediate lira nodes ( time @xmath10 ) .
when the chunk transfer completes , which is denoted by an end of chunk ( eoc ) field , the intermediate lira nodes change their c - fib entries for this ` cid ` by marking the interfaces through which they forwarded the data ( _ i.e. , _
@xmath2 ) as @xmath3 ( _ temporary interface _ ) - interface 3 is moved to @xmath3 at @xmath12 in table [ table : r1routing - table ] .
this is done since the content can possibly be delivered from there too ( _ i.e. , _ the content has possibly been cached towards this direction ) .
when the client sees the eoc field / bit set , it forwards the next request towards the original cp ( similarly to the initial request - step above ) in order to obtain the ` cid ` of the next chunk .
lira nodes by default support in - network caching . in the simplest case , on - path in - network
caching is supported by simply performing a lookup of the ` cid ` of a request message , at the local cache index . in case
the requested content chunk is cached locally , the corresponding data is returned through the network interface the request was received from ( @xmath9 ) . in our example scenario , client b issues a request for content @xmath11 . once the request for @xmath11 reaches @xmath19 , the c - fib table of @xmath19 is updated to include @xmath23 and @xmath24 ( @xmath20 in table [ table : r2routing - table ] ) . then , at time @xmath13 , the request for @xmath11 reaches @xmath6 .
content chunk @xmath11 is found cached at @xmath6 whose interface 2 is marked as @xmath9 and the content is sent towards client b. by introducing the @xmath3 field in the c - fib table we further realise off - path in - network caching @xcite , as well as user - assisted in - network caching @xcite , @xcite . when a content chuck is not found in the local cache
, the lira node sends the received request towards both the ( permanent ) incoming interface @xmath8 ( as indicated by the name resolution process ) and the temporary interface(s ) @xmath3 . in our example , @xmath6 sends two requests for @xmath11 towards both the ( permanent ) @xmath8 1 and the @xmath3 3 ( @xmath13 in table [ table : r1routing - table ] ) . whichever of the two interfaces ( 1 or 3 ) starts receiving the requested data first is marked as the incoming interface for this content and the remaining ( temporary ) interfaces are pruned down .
pruning here can be realised through a negative ack ( nack ) packet which travels towards the source of the content .
if @xmath8 answers first , the @xmath3 is removed from the corresponding c - fib table entry .
alternative strategies can be applied here , by selectively forwarding a request to one or more of the available interfaces _
e.g. , _ always forwarding only towards an off - path cache , since requests are always routable to the cp at the ip layer . finally , at time @xmath14 when @xmath11 transfer completes ( from either the local , or a remote cache ) , interface 2 is added to the list of temporary incoming interfaces ( @xmath3 ) at @xmath6 , since @xmath11 can now be found this way too ( similarly to @xmath12 ) .
the c - fib table of @xmath19 is also updated to include interface 2 as @xmath3 ( step @xmath21 ) .
we note that in order to avoid routing loops in case no other device towards @xmath3 ( client a in our case ) has the content cached , we discard requests ( for items in the c - fib table ) that come in through its marked @xmath3 .
this is done because any lira node towards the @xmath3 ( client a in this case ) will forward the request based on its ip address ( carried at the ip layer and always pointing towards the permanent content source , hence through @xmath25 in our example ) if it finds no entry in its c - fib table for the requested content . in turn ,
upon receipt of the request , @xmath25 will send the request back towards the same direction ( towards client a here ) , since it still has got the related entry in its c - fib table .
this will result in the request travelling back and forth creating an endless routing loop .
multicast support is enabled through the use of the @xmath9 and @xmath4 fields of the c - fib table .
as described above , during the chunk transfer , the network interface of the lira node where the incoming data is forwarded towards is marked in the @xmath9 field .
this @xmath9 entry enables the lira node to suppress any subsequent request for the same content chunk by adding an extra outgoing interface to its c - fib .
this is similar to the pit functionality in ccn @xcite .
note that in all above steps the ip address ( at the ip layer ) of request packets has been pointing to the cp and of content chunks to the corresponding clients .
however , in order to realise multicast transmission in this case ( _ i.e. , _ avoid sending a second request for the same chunk towards the same direction ) , the lira node that suppresses subsequent requests needs to keep the extra ip address of the clients that generated the requests .
we deal with this situation through the multicast ip " ( @xmath4 ) field in the c - fib table .
when data arrives at the branching lira node , it gets forwarded to all @xmath9 interfaces .
the @xmath4 entries are used at the ip layer to allow for the delivery of the duplicated data to the requesting recipients .
note however that multicast forks further down the path are handled locally . in our example , if an additional client c attaches to @xmath19 and requests for @xmath16 during the multicast session , its request will be suppressed by @xmath19 which will also store client c s ip address in the corresponding @xmath4 field .
@xmath6 will not be aware of client c s existence and @xmath19 is responsible for duplicating data for this client .
thus , the @xmath4 state load is distributed to the participating lira nodes avoiding the overloading of nodes closer to the root of the multicast tree . in our example network ,
client a issues a request for content @xmath16 . the c - fib table at @xmath6 marks @xmath26 and @xmath27 for ` cid ` @xmath16 ( step @xmath15 ) .
before the transfer of @xmath16 towards a completes through @xmath6 client b issues a request for @xmath16 , which goes through @xmath19 and reaches @xmath6 .
@xmath19 updates its c - fib table by putting @xmath26 and @xmath28 ( step @xmath22 ) .
@xmath6 does not forward this request further ; instead it adds interface 2 to the @xmath9 field of @xmath16 and also stores the ip address of client b ( taken from the corresponding ip layer field ) in the @xmath4 field ( step @xmath17 ) .
when @xmath16 arrives at @xmath6 ( step @xmath17 ) it is forwarded towards client a through @xmath27 , but it is also replicated and forwarded towards client b , through @xmath28 , using @xmath4 as the destination ip address .
when the chunk @xmath16 transfer completes , router @xmath6 moves interfaces 2 and 3 and @xmath19 moves interface 2 to the @xmath3 field ( step @xmath18 - table [ table : r1routing - table ] and [ table : r2routing - table ] ) .
+ note that the c - fib table introduced here , incorporates the functionality of both the pit and the fib tables of ccn .
for as long as the chunk transfer goes on and hence , the @xmath9 field is filled ( and the @xmath3 field is empty - @xmath29 and @xmath17 in @xmath6 s c - fib , see table [ table : r1routing - table ] ) , the c - fib table represents the pit table of ccn / ndn .
that is , based on this state , lira nodes are able to collapse / suppress subsequent requests for content already requested ( or under transmission ) and realise multicast . when the chunk transfer completes and the entry in @xmath9 is moved to the @xmath3 field ( @xmath30 and @xmath18 in table [ table : r1routing - table ] ) , then the c - fib table reflects the fib table of ccn / ndn . as mentioned above
, however , the c - fib table acts as a cache for recently served content and hence , it does not need to keep huge amounts of state information in the fib part of the c - fib .
we discuss and evaluate both parts of the c - fib table later in section [ eval ] .
it is generally not common to evaluate a network architecture merely in quantitative terms , given that the contribution of such studies comes mainly at a conceptual level . in our case ,
the contribution of the lira architecture comes mainly in terms of incremental deployment with backward compatibility guarantees . at the same time , however , lira can achieve all the quantifiable benefits of an icn mode of operation . to provide a thorough performance evaluation ,
we analyse conceptual and qualitative gains in sec .
[ qualitative - eval ] as well as quantitative gains in sec .
[ quantitative - eval ] .
the quantitative evaluation focuses on the deployment of the lira concept from the operators perspective .
in particular , given a fixed monetary budget that the operator is prepared to spend in order to deploy lira , we assess the best strategies of investing the capital in terms of extra equipment , which in our case translates to cache memory and c - fib tables .
we also demonstrate and quantify the benefits brought by lira to cps , with a particular focus on cache purging and the control over the freshness of the cached content .
* name resolution : * by handing control of the name resolution process to cps , lira avoids the need for either the deployment of a costly name resolution infrastructure , or the investment on in - network resources for the support of line - speed name resolution .
the operation of the c - fib as a cache for names/`cid`s is similar to @xcite .
however , lira does not necessitate the use of an explicit off - path name resolution mechanism , as it rather falls back to ip , in a backward compatible manner .
at the same time , by following a backwards compatible http - supported name resolution mechanism , lira presents a complete interface for the interaction of end - hosts with an information - centric network . to the best of our knowledge
, no exact mechanism has been proposed for the discovery ( _ i.e. , _ not only resolution ) of content names in alternative icn architectural proposals . * control of content access : * lira enables cps to directly monitor and control the access of end users to their content .
route - by - name approaches such as @xcite fail to provide such support .
cps would be reluctant to accept transparent access to their content , thus dis - incentivising the adoption of such an approach to icn by isps .
lookup - by - name approaches , on the other hand , such as @xcite and @xcite , enable this type of control , by decoupling name resolution from forwarding .
however , this comes at the cost of additional name resolution infrastructure and directly places the content access information in the hands of isps ; in turn , this introduces the burden of new ( business and technical ) interfaces between all cps and all isps at a global scale .
* mobility : * although the issue of mobility in case of lira requires further investigation and at first sight it might seem that lira can not deal with mobility efficiently , due to its dependence on ip , we note the following : upon a content request , the cp or cdn is sending back to the client the ` cid`s of the next few chunks , _
i.e. , _ not just the next one . that said , the clients operate based on ip - agnostic ` cid`s .
therefore , client mobility can be natively supported , as clients request for content based on identifiers ( in combination to the ip address at the ip layer ) .
source mobility , on the other hand , is an issue that requires further investigation as is the case with all icn architectural proposals . *
security : * by supporting self - certifying ` cids ` , lira secures the content itself rather than the communication channel , similarly to other icn architectures _
e.g. , _ @xcite .
* implementation : * the proposed lira functionalities can be deployed on nodes with only firmware updates without the need for hardware replacement or upgrade .
in fact , by relying on ip forwarding as a fallback in case of c - fib misses , lira will never result in un - routable requests / content even if deployed on just a few nodes and with minimal memory .
this is in stark contrast with previous icn proposals like ccn and ndn which require well - dimensioned fib and pit structures to operate correctly and at line speed .
c - fib can be loaded in dram , which has been shown to be able to support line - speed per - packet lookups @xcite , @xcite , is inexpensive and abundant on modern routers based on either network processors or general purpose processors .
it is in fact common to have at least a few gbs of spare dram on modern routers . since the binary code implementing lira functionalities is likely to require negligible space ,
all available dram can be used for c - fib and caching space .
c - fib entries in particular have very low memory requirements .
in fact , even assuming that ( i ) the c - fib is implemented using a hash - table with a load factor of @xmath31 and with a circular queue for replacement and ( ii ) the unfavourable case that lira chunks are named using sha-512 hashes and next hops information are coded on 2b , it is still possible to store over 15 million c - fib entries per gb of dram .
this makes c - fibs and more generally the lira node architecture easy to incrementally deploy on today s routers . as mentioned earlier
, the objective of this section is to evaluate the best possible way to invest in deploying the lira concept , from the operator s perspective .
that said , we initially evaluate the main concepts of our proposal with regard to their projected gains in terms of cache hits .
although lira is far from a caching - specific architecture , caching is : _
i ) _ the only straightforward quantitatively measurable aspect of an icn architecture , and most importantly , _
ii ) _ the main feature that requires investment from network operators . for these reasons and without by any means underestimating the gains from the above - mentioned qualitative benefits of the lira architecture , in this section , we focus on the evaluation of the main concepts included in lira as seen from an in - network caching perspective .
we use icarus @xcite to evaluate the performance of various aspects of our proposed framework based on real isp topologies from the rocketfuel dataset @xcite and synthetic workloads @xcite . due to space limitations , we omit evaluation of the multicast functionality offered by lira , since the related performance benefit is straightforward
. moreover , we only show results for the telstra and abovenet topologies , though we report that we obtain similar results with other topologies as well .
we make the code , documentation and data required to reproduce our results publicly available .
lira nodes can have a content cache or a c - fib table , or both . given a fixed total cache and c - fib capacity budget , in this section
, we identify the best possible combination of cache and c - fib deployment along two dimensions : i ) deployment strategies and ii ) caching strategies .
we attempt to capture the interaction dynamics between nodes that cache content and nodes that can route to this content , in a location - independent manner , _
i.e. , _ through c - fib table entries .
_ our first objective is to investigate the effectiveness of c - fib table entries in mapping the content cached in neighbour nodes .
our second objective is to see how c - fib table entries eventually translate to cache hits . _
modelling the performance of a network of caches is known to be complex @xcite , @xcite . as a result , it is extremely difficult to formulate optimal cache placement algorithms which are also robust to realistic traffic variations . arguably , the complexity of the optimal cache placement problem is another obstacle hindering icn deployments .
therefore , motivated by practical reasons , we propose four simple content cache and c - fib placement algorithms and show that they are sufficient to provide tangible performance gains even with partial deployments . to deploy caches and c - fibs , we rank nodes according to their betweenness centrality ( _ i.e. , _ the amount of traffic traversing them following shortest path routing @xcite ) and deploy lira functionality using the following strategies : + _ ( i ) _ cache in top 50% high centrality nodes , c - fib table in all nodes : @xmath32 . + _
( ii ) _ cache in top 50% high centrality nodes , c - fib table in top 50% high centrality nodes : @xmath33 . + _
( iii ) _ cache in all nodes , c - fib table in all nodes : @xmath34 . + _
( iv ) _ cache in all nodes , c - fib table in top 50% high centrality nodes : @xmath35 .
we run simulations and measure the mean _ c - fib freshness _ , which we define as the ratio of entries stored in c - fib tables which can correctly route to a copy of a content stored in a nearby cache .
this metric captures how well the entries of the c - fib tables deployed in the network reflect the current state of nearby caches .
we further characterise the correct c - fib entries by the hop distance to the lira node that caches the corresponding content .
note that in all cases , and regardless of the deployment strategy , the ratio of c - fib table to cache entries is fixed ( see next subsection for the evaluation of this ratio ) .
as a result , c - fib tables in fewer nodes ( than those that deploy caches ) keep more entries to match the number of cache slots ( and vice versa ) .
we also analyse the results under different caching strategies : _ leave copy everywhere _
( lce ) , according to which a copy of a content is stored in every cache traversed and _ random choice _ , according to which a content is stored only in one randomly selected caching node along the delivery path .
the rationale behind our choice is to evaluate deployment and caching performance under varying _ caching redundancy _ @xcite .
our results are shown in figs .
[ c - fib - freshness ] and [ deployment - strategies ] . +
* c - fib efficiency .
* first of all , it is important to highlight the fact that the c - fib table entries depict precisely the state of neighbour caches .
this is proved by the fact that the _ freshness ratio _ in fig .
[ c - fib - freshness ] directly translates to off - path cache gain in fig .
[ deployment - strategies ] : for instance , the freshness result in case of @xmath32 in fig .
[ freshness-1221 ] indicates that 5% of entries in the c - fib table can correctly route to the content in neighbour caches . in turn , in fig .
[ cachehits-1221 ] , the gain from off - path caching ( red , top part of bar ) is 4.5% .
this is an important result that highlights the effectiveness of the c - fib table in keeping an accurate record of the state of nearby caches ( _ i.e. , _ up to 3 hops away in our evaluation ) . +
* deployment strategy . * in terms of c - fib freshness , deploying smaller caches over more / all nodes , _
@xmath36 , seems to be more effective in capturing the state of caches from the c - fib tables ( _ i.e. , _ higher freshness in fig .
[ c - fib - freshness ] ) .
this is explained by the fact that the monitoring and mapping " mechanism provided by the c - fib table has got a wider view of the neighbourhood and can therefore , find more content items locally .
this also translates to more off - path cache hits in fig .
[ deployment - strategies ] for @xmath36 . out of the four deployment strategies under consideration here ,
@xmath33 and @xmath32 consistently perform best in terms of cache hits ( in fig . [ deployment - strategies ] ) .
this is irrespective of the freshness result , which shows that freshness improves when caches are deployed over all nodes ( _ i.e. , _
@xmath36 ) . in other words , it is better to have fewer but bigger caches placed in high centrality nodes ( as also shown in @xcite ) , rather than having smaller caches deployed in all nodes of the network . + * caching strategy . * as expected , in terms of cache hits , _
_ always performs best , for all topologies and for all deployment strategies , as a result of its reduced caching redundancy .
similar results have been reported before in @xcite .
lce on the other hand , performs roughly the same across all deployment strategies .
note that the lce result in fig .
[ deployment - strategies ] effectively reveals the performance of the ccn / ndn architecture . due to space limitations
, we do not present a full - fledged comparison between the architectures , but fig .
[ deployment - strategies ] reveals very well the cache - related performance of ccn / ndn .
we next quantify the performance benefit of off - path , c - fib - routed , caching for various values of the c - fib - to - cache size ratio ( expressed in number of entries ) and for the @xmath32 strategy ( fig .
[ hits - vs - c - fib ] ) .
considering the overall cache hit ratio ( both on- and off - path ) , we see a considerable increase when moving from a ratio value of 0.25 to a ratio value of 16 , due to c - fib routing redirections .
the results are similar for a ratio equal to 32 , but the gain in this case is marginal .
therefore , given that larger memory is required in order to deploy c - fib tables 32 times bigger than the entries in the respective caches , we conclude that a value of 16 is optimal .
although in absolute values , off - path , c - fib - based , caching contributes less than on - path caching , the gain is still far from negligible ( _ i.e. , _ it can reach up to 50% in fig .
[ hits - vs - c - fib ] ) .
we report that in our simulations , the gain from off - path caching can reach 100% , effectively doubling the gain from on - path caching .
finally , it is interesting to note the slight decrease of on - path cache hit ratio as the c - fib - to - cache size ratio increases .
this is attributed to cases where a content request encounters a c - fib table entry and gets diverted to an off - path cache , before it actually hits an existing on - path cache . in this case , and given that the c - fib table entry is found earlier in the path , we report that the delay to deliver the content back to the user is even shorter than finding the requested item in an on - path cache .
this is especially so in case of lce caching , where due to increased caching redundancy , a copy of a content has good chances of being found along the shortest path .
one of the departing points in the design of the lira architecture is the direct control of content by the cps / cdns , as discussed earlier .
we identify two main features that give direct control of the content to the cp or cdn .
the first one is the control of access logging . in lira
this is accomplished by the content provider - controlled name resolution , where clients need to get the up - to - date ` cid ` from the content provider .
this requires an extra rtt to get to the cp or cdn .
we remind that according to our discussion in section [ transitioning - interval ] , cps / cdns send more than one ` cid ` to the client , therefore , the journey to the cp / cdn happens rarely during the data transfer , or even only once in case of small files ( _ e.g. , _ web ) .
we assume this extra rtt to incur only a tiny performance penalty compared to alternative proposals that do not necessarily require this extra roundtrip .
the second feature that provides control of published content to cps is the ability to actively perform cache purging . as described in section [ cc : en ] , when cps change the ` cid ` of a content item , previously cached items no longer get hits from new requests and eventually get evicted ( denoted as _
lira w / o replacement _ ) .
taking a step further , we consider an extended version of this mechanism , where data packets explicitly indicate the ` cid ` values of the content items that should be immediately evicted from encountered caches ( denoted as _
lira w/ replacement _ ) .
figure [ fig : purging - cache - hit - ratio ] shows the cache hit ratio of the above mechanisms along with that of a simple ttl - based mechanism , where any cache hit returns the content to the client , even if this content is stale ( _ ttl - based ( all hits ) _ ) for various ttl values . in fig .
[ fig : purging - cache - hit - ratio ] , we see that the cache hit ratio in the ( _ ttl - based ( all hits ) _ ) case increases with the value of ttl , since content remains longer in the cache .
however , this also means that the corresponding cache hits result in the reception of stale content .
[ fig : purging - cache - hit - ratio ] also shows the cache hit ratio for fresh only content ( _ ttl - based ( fresh only ) _ ) , which initially increases , but then steadily drops as a result of high ttl values that increase stale cached content .
the _ lira w/ replacement _ mechanism performs considerably better than _ lira w / o replacement _ , as it immediately frees the caching space from unnecessary stale content , and better than its _ ttl - based ( fresh only ) _ counterpart .
it must be noted that the _ ttl - based ( fresh only ) _ ratio is only provided as a benchmark , as ttl - based mechanisms can not avoid serving stale content . on the other hand
, lira provides a precise mechanism to avoid serving stale cached content altogether .
we proceed to evaluate the last of the design targets behind lira , that of incremental deployability . to assess the performance gain of incrementally deploying the lira architecture , we begin by progressively adding c - fib tables starting from the highest centrality nodes .
we evaluate the performance in terms of cache hits in case of caches deployed in 25% , 50% , 75% and 100% of the nodes , starting from the highest centrality ones .
we observe in fig .
[ fig : incremental - deployment ] that performance stabilises with the c - fib table present in 20 - 30% of the nodes .
c - fib in less than 20% results in suboptimal performance , but performance does not increase considerably if we continue adding c - fib to more nodes . in terms of caches , a 25% deployment rate results in poor performance , while the performance does not improve considerably when caches are deployed in more than 50% of nodes . the difference in performance between the 50% and 100% of nodes is in the area of 1% improvement in terms of cache hit ratio for the two topologies shown here ( telstra and abovenet ) .
we conclude that adding c - fib to the top 20 - 30% highest centrality nodes and caches to 50%-75% of highest centrality nodes achieves the full performance gain of the lira architecture .
although here we present results for telstra and abovenet topologies , our results are consistent along all six evaluated topologies of the rocketfuel dataset .
there is a constant trend towards extra flexibility " in communication networks , which started with the shift from ( rigid ) circuit - switching to ( queuing - based ) packet - switching @xcite .
we see location - independent , information - centric networking as the natural next step towards _ content switching"_. to move towards this direction , however , the research community needs to take into account the interests of the main internet market players , as well as those of users .
we argue that icn research so far has focused on designing conceptually sound and scalable name - based routing architectures , but largely ignored any incentives ( provided through those architectures ) to adopt the icn technology .
the interests of content providers and cdns are largely different to those of isps and the shift to an icn environment environment makes this difference even more pronounced .
that said , unless a shift to an icn environment takes into account the interests of both cps / cdns and isps , the incentives to adopt this technology will be limited . in this paper
we have taken these concerns into consideration and have designed an incrementally - deployable icn architecture .
the proposed architecture is based on the location - independent routing layer ( lira ) and directly involves the content provider in the name resolution process .
furthermore , ephemeral names give more power to the cps / cdns over the content they publish .
our evaluation shows that even with a limited number of nodes implementing the lira architecture , isps achieve a clear performance gain , while at the same time cps / cdns have full control of their content .
this work was supported by epsrc uk ( comit project ) grant no .
ep / k019589/1 and eu fp7/nict ( greenicn project ) grant no . ( eu ) 608518/(nict)167 .
the research leading to these results was funded by the eu - japan initiative under european commission fp7 grant agreement no .
608518 and nict contract no .
167 ( the greenicn project ) and by the uk engineering and physical sciences research council ( epsrc ) under grant no .
ep / k019589/1 ( comit ) .
m. dehghan , a. seetharam , b. jiang , t. he , t. salonidis , j. kurose , d. towsley , and r. sitaraman . on the complexity of optimal routing and content caching in heterogeneous networks . in _
infocom , 2015 proceedings ieee _ , april 2015 . | we identify the obstacles hindering the deployment of information centric networking ( icn ) and the shift from the current ip architecture .
in particular , we argue that scalability of name resolution and the lack of control of content access from content providers are two important barriers that keep icn away from deployment .
we design solutions to incentivise icn deployment and present a new network architecture that incorporates an extra layer in the protocol stack ( the _ location independent routing layer _ , lira ) to integrate location - independent content delivery . according to our design
, content names need not ( and should not ) be _ permanent _ , but rather should be _ ephemeral_. resolution of non - permanent names requires the involvement of content providers , enabling desirable features such as request logging and cache purging , while avoiding the need for the deployment of a new name resolution infrastructure .
our results show that with half of the network s nodes operating under the lira framework , we can get the full gain of the icn mode of operation .
= 10000 = 10000 [ distributed networks ] |
graph - based codes , such as low - density parity - check ( ldpc ) , turbo , and repeat - accumulate codes , together with belief propagation ( bp ) decoding have shown to perform extremely close to the shannon limit with reasonable decoding complexity @xcite .
these graph - based codes can be represented by a bipartite tanner graph in which the variable and check nodes respectively correspond to the codewords symbols and the parity check constraints @xcite .
the error - correcting performance of a code is mainly characterized by the connectivity among the nodes in the tanner graph where the node degree plays an important role . to specify the node degree distribution in the tanner graph ,
the concept of degree distribution in either node perspective or edge perspective is introduced @xcite .
a code ensemble is then defined as the set of all codes with a particular degree distribution . as a unifying framework for graph - based codes , richardson and urbanke @xcite proposed multi - edge type low density parity - check ( met - ldpc ) codes .
the benefit of the met generalization is greater flexibility in the code structure , which can improve decoding performances .
this generalization is particularly useful under traditionally difficult requirements , such as high - rate codes with very low error floors or low - rate codes @xcite . a numerical technique , referred to as density evolution ( de ) ,
was formulated to analyze the convergence behavior of the bp decoder ( i.e. , the code threshold ) for a given ldpc @xcite or met - ldpc @xcite code ensemble under bp decoding , where the code threshold is defined as the maximum channel noise level at which the decoding error probability converges to zero as the code length goes to infinity .
de determines the performance of bp decoding for a given code ensemble by tracking the probability density function ( pdf ) of messages passed along the edges in the corresponding tanner graph through the iterative decoding process .
then , it is possible to test whether , for a given channel condition and a given degree distribution , the decoder can successfully decode the transmitted message ( with the decoding error probability tends to zero as the iterations progress ) .
this allows us to design and optimize ldpc and met - ldpc degree distributions using the de threshold ( i.e. , the code threshold found using de ) as the cost function . calculating thresholds and designing ldpc and met - ldpc degree distributions using de are computationally intensive as they require numerical evaluations of the pdfs of the messages passed along the tanner graph edges in each decoding iteration . because of this , for ldpc codes on the binary input additive white gaussian noise ( bi - awgn ) channel , chung _ et al . _
@xcite and lehmann and maggio @xcite approximated the message pdfs by gaussian pdfs , each using a single parameter , to simplify the analysis .
existing work concerning gaussian approximations has relied on four different parameters in order to obtain a single - parameter model of the message pdfs , including mean value of the pdf @xcite , bit - error rate ( ber ) @xcite , reciprocal - channel approximation ( rca ) @xcite and mutual information ( e.g. , exit charts ) @xcite .
several papers @xcite have investigated the accuracy of the gaussian approximation for bp decoding of standard ldpc codes and shown that it is accurate for medium - to - high rates .
however in most of the literature regarding de for met - ldpc codes , only the full density evolution ( full - de ) has been studied @xcite . in full - de
, the quantized pdfs of the messages are passed along the edges without any approximation .
typically , for full - de , thousands of points are used to accurately describe one message pdf .
schmalen and brink @xcite have used the gaussian approximation based on the mean of the message pdf @xcite to evaluate the behavior of protograph based ldpc codes , which is a subset of met - ldpc codes .
the contributions of this paper are as follows : 1 ) we investigate the accuracy of gaussian approximations for bp decoding .
we follow the approximation techniques suggested for ldpc codes @xcite , which describe each de - message pdf using a single parameter . based on our observations of the evolution of pdfs in the met - ldpc codes
, we found that those gaussian approximations are not accurate for the scenarios where met - ldpc codes are useful , i.e. , at low rate and with punctured variable nodes .
2 ) in light of this , we propose a hybrid - de method , which combines the full - de and a gaussian approximation .
our proposed hybrid - de allows us to evaluate the code threshold ( i.e. , the cost function in the code optimization ) of met - ldpc and ldpc code ensembles significantly faster than the full - de and with accuracy better than gaussian approximations .
3 ) we design good met - ldpc codes using the proposed hybrid - de and show that the hybrid - de well approximates the full - de for code design .
this paper is organized as follows .
section [ background ] briefly reviews the basic concepts of met - ldpc codes . in section [ gaussian app ]
we extend gaussian approximations for ldpc codes to met - ldpc codes , and in section [ validity of ga ] , we discuss the accuracy of the gaussian approximations under the conditions where met - ldpc codes are more beneficial .
section [ hybrid de ] presents the proposed hybrid - de method , and section [ code design ga ] demonstrates the benefits of code design using the proposed hybrid - de method over existing gaussian approximations .
finally , section [ conclusion ] concludes the paper .
unlike standard ldpc code ensembles where the graph connectivity is constrained only by the node degrees , in the multi - edge setting , several edge - types can be defined , and every node is characterized by the number of connections to edges of each edge - type . within this framework
, the degree distribution of met - ldpc code ensemble can be specified through two node - perspective multinomials related to the variable nodes and check nodes respectively @xcite : @xmath0 where @xmath1 and @xmath2 are vectors defined as follows .
let @xmath3 denote the number of edge - types corresponding to the graph and @xmath4 denote the number of different channels over which codeword bits may be transmitted .
a vector @xmath5 $ ] is defined for each node in the graph , where @xmath6 is the number of edges of the @xmath7 edge - type connected to that node , and we use @xmath8 to denote @xmath9 .
as the variable nodes receive information from the channel over which the codeword bits are transmitted , there is an additional vector @xmath10 associated with each variable node where @xmath11 designates the type of the message ( i.e. , the message pdf ) it receives from the channel .
typically , @xmath12 $ ] has only two entries since in a bi - awgn channel , a codeword bit is either punctured ( the codeword bits not transmitted : @xmath13 $ ] ) or transmitted through a single channel ( @xmath14 $ ] ) . finally , @xmath15 and @xmath16 are non - negative real numbers corresponding to the fraction of variable nodes of type ( @xmath17 ) and the fraction of check nodes of type @xmath18 in the ensemble , respectively .
( resp . , @xmath19 ) represents the unpunctured ( resp . ,
punctured ) variable nodes and @xmath20 represents the check nodes .
the number of nodes for different edge - types are shown as fractions of the code length @xmath21 , where @xmath21 is the number of transmitted ( i.e. unpunctured ) codeword bits . ]
node - perspective degree distributions can be converted to edge - perspective via the following multinomials , where @xmath22 and @xmath23 are related to the variable nodes and check nodes , respectively @xcite : @xmath24 @xmath25 where @xmath26 the rate of a met - ldpc code is given by @xmath27 where @xmath28 denotes a vector of all @xmath29 s with the length determined by the context .
a rate @xmath30 met - ldpc code ensemble is shown in fig .
[ fig : met_tanner ] , where the node - perspective degree distributions are given by @xmath31 and @xmath32 . here
@xmath33 denotes punctured nodes and @xmath34 denotes unpunctured nodes . in the bp decoding algorithm ,
messages are passed along the edges of the tanner graph from variable nodes to their incident check nodes and vice versa until a valid codeword is found or a predefined maximum number of decoding iterations has been reached .
each bp decoding iteration involves two steps . 1 . a variable node processes the messages it receives from its neighboring check nodes and from its corresponding channel and outputs messages to its neighboring check nodes .
a check node processes inputs from its neighboring variable nodes and passes messages back to its neighboring variable nodes . in most cases of binary codes transmitted on bi - awgn channel , these
bp decoding messages are expressed as log - likelihood ratios ( llrs ) @xcite .
this is used to reduce the complexity of the bp decoder , as multiplication of message probabilities corresponds to the summation of corresponding llrs .
let @xmath35 denote the message llr sent by a variable node to a check node along edge @xmath36 , ( i.e , variable - to - check ) at the @xmath37 iteration of the bp decoding , and @xmath38 denote the message llr sent by a check node to a variable node along edge @xmath36 , ( i , e , check - to - variable ) at the @xmath37 iteration of the bp decoding .
at the variable node , @xmath35 is computed based on the observed channel llr ( @xmath39 ) and message llrs received from the neighboring check nodes except for the incoming message on the current edge for which the output message is being calculated .
thus the variable - to - check message on edge @xmath36 at the @xmath40th decoding iteration is as follows : @xmath41 the message outputs on edge @xmath36 of a check node at the @xmath40th decoding iteration can be obtained from the `` tanh rule '' : @xmath42 for more details we refer readers to ryan and lin @xcite .
de is the main tool for analyzing the average asymptotic behavior of the bp decoder for met - ldpc code ensembles , when the block length goes to infinity . for bp decoding on a bi - awgn channel , these llr values ( i.e. , @xmath43 )
are continuous random variables , thus can be described by pdfs for analysis using de . to analyze the evolution of theses pdfs in the bp decoder
, we define @xmath44 which denote the pdf of the variable - to - check message , check - to - variable message and channel llr , respectively . unlike standard ldpc codes , in the met framework , because of the existence of multiple edge - types , only the incoming messages from same edge - type are assumed to be identically distributed .
however , all the incoming messages are assumed to be mutually independent .
recall that in met - ldpc codes , a variable node is identified by its type , ( @xmath45 ) , and a check node by its type , @xmath18 .
thus from ( [ eq.vn update ] ) the pdf of the variable - to - check message from a variable node type , ( @xmath45 ) , along edge - type @xmath46 at the @xmath37 decoding iteration can be written as follows @xcite : @xmath47^{\bigotimes ( d_i-1 ) } \\
\bigotimes_{k=1 , k\neq i}^{m_e}\left [ f\big({u}^{(\ell-1)}(k)\big)\right]^{\bigotimes d_k},\end{gathered}\ ] ] where @xmath48 denotes convolution .
the @xmath6-fold and @xmath49-fold convolutions follow from the assumption that the incoming messages from a check node along edge - type @xmath46 are independent and identically distributed and @xmath50 is used to denote this common pdf .
@xmath51 is the pdf of the channel llr . from ( [ eq.cn update ] ) the pdf of the check - to - variable message from a check node type , @xmath18 , along edge - type @xmath46 at the @xmath37 decoding iteration
can be calculated as follows @xcite : @xmath52^{\boxtimes ( d_i-1 ) } \stackrel[k=1 , k\neq i]{m_e}{\boxtimes } \left [ f\big({v}^{(\ell)}(k)\big)\right]^{\boxtimes d_k } , \end{aligned}\ ] ] where @xmath53 is the pdf of the message from a variable node along edge - type @xmath46 at the @xmath37 decoding iteration .
the computation of @xmath54 is not as straightforward as that for @xmath55 and requires the transformation of @xmath56 and @xmath57 .
so we use @xmath58 to denote the convolution when computing the pdf of @xmath54 for check - to - variable messages . for more details
we refer readers to richardson and urbanke @xcite .
in this section , we consider met - ldpc codes over bi - awgn channels with gaussian approximations to de .
as already shown for ldpc codes @xcite , the pdfs of variable - to - check and check - to - variable messages can be close to a gaussian distribution in certain cases , such as when check node degrees are small and variable node degrees are relatively large .
since a gaussian pdf can be completely specified by its mean ( @xmath59 ) and variance ( @xmath60 ) , we need to track only these two parameters during the bp decoding algorithm .
furthermore , it was shown by richardson _ et al . _
@xcite that the pdfs of variable - to - check and check - to - variable messages and channel inputs satisfy the symmetry condition : @xmath61 where @xmath62 is the pdf of variable @xmath63 .
this condition greatly simplifies the analysis because it implies @xmath64 and reduces the description of the pdf to a single parameter .
thus , by tracking the changes of the mean ( @xmath59 ) during iterations , we can determine the evolution of the pdf of the check node message , @xmath65 and the variable node message , @xmath66 where @xmath67 is the gaussian pdf with mean @xmath59 and variance @xmath60 .
@xmath68 and @xmath69 are the mean of the variable - to - check and check - to - variable messages , respectively . in this subsection
, we will extend the gaussian approximation method proposed by chung _ et al . _
@xcite for the threshold estimation of standard ldpc codes to that of met - ldpc codes .
this method is based on approximating message pdfs using a single parameter , i.e. , the mean of the message pdf .
recall that a variable node is identified by its type , ( @xmath45 ) , and a check node by its type , @xmath18 .
since the pdfs of the messages sent by the variable node are approximated as gaussian , the mean of the variable - to - check message from a variable node type , ( @xmath45 ) , along edge - type @xmath46 at the @xmath37 decoding iteration is given by @xmath70 where @xmath71 is the mean of the message from the channel and @xmath72 is the mean of the check - to - variable message along edge - type @xmath46 at the @xmath73 decoding iteration .
the updated mean of the check - to - variable message from check node type of @xmath18 along edge - type @xmath46 at the @xmath37 decoding iteration can be written as @xmath74^{d_i-1 } \\
\prod_{k=1,k\neq i}^{m_e } \left [ 1- \phi(m_v^{(\ell)}(k))\right]^{d_k } \big),\end{gathered}\ ] ] where @xmath75 is the mean of the variable - to - check message along edge - type @xmath46 at the @xmath37 decoding iteration .
the mean of the variable - to - check and the check - to - variable messages along edge - type @xmath46 at the @xmath37 decoding iteration is given by @xmath76 where @xmath77 and @xmath78 are the variable and check node edge - degree distributions with respect to edge - type @xmath46 , respectively and @xmath79 it is important to note that @xmath80 is continuous and monotonically decreasing over @xmath81 with @xmath82 and @xmath83 @xcite .
in this subsection , we will extend a gaussian approximation method proposed by lehmann _ et al .
_ @xcite that estimates thresholds of standard ldpc codes to that of met - ldpc codes .
this method is based on a closed - form expression in terms of error probabilities ( i.e. , the probability that a variable node is sending an incorrect message ) .
consider a check node of type , @xmath18 .
the error probability of a check - to - variable message from a check node type , @xmath18 along edge - type @xmath46 at the @xmath37 decoding iteration is given by @xmath84,\end{gathered}\ ] ] where @xmath85 is the average error probability of the variable - to - check message along edge - type @xmath46 at the @xmath73 decoding iteration .
since we suppose that the all - zero codeword is sent , the error probability of a variable node at the @xmath37 decoding iteration is simply the average probability that the variable - to - check messages are negative .
we also assume that the pdf of variable - to - check message is symmetric gaussian ; therefore the error probability of a variable - to - check message from a variable node type , @xmath86 along edge - type @xmath46 at the @xmath37 decoding iteration is given by @xmath87 where @xmath88 is the mean of the variable - to - check message from a variable node type , ( @xmath45 ) , along edge - type @xmath46 at the @xmath37 decoding iteration , and @xmath89 @xmath88 can be calculated using ( [ eq : vn to cn met maen ] ) by substituting @xmath90 for each @xmath91 , where @xmath92 is the average error probability of the check - to - variable message along edge - type @xmath46 at the @xmath37 decoding iteration .
the average error probability of the variable - to - check and the check - to - variable messages along edge - type @xmath46 at the @xmath37 decoding iteration is given by @xmath93 where @xmath77 and @xmath78 are the variable and check node edge - degree distributions with respect to edge - type @xmath46 , respectively .
in this subsection , we will extend another gaussian approximation method , proposed by chung @xcite to estimates thresholds of regular ldpc codes , to that of met - ldpc codes .
this method is called reciprocal - channel approximation ( rca ) , which is based on reciprocal - channel mapping and mean ( @xmath59 ) of the node message is used as the one - dimensional tracking parameter for the bi - awgn channel . with the rca technique in de , @xmath59 is additive at the variable nodes similar to approximation 1 ( see ( [ eq : vn to cn met maen ] ) ) . the difference between approximation 1 and approximation 3
is how the check nodes calculate their output messages .
instead of evaluating the @xmath94 function in approximation 1 , approximation 3 uses the reciprocal - channel mapping , @xmath95 , which is additive at the check nodes .
@xmath95 is defined as follows @xcite : @xmath96 where @xmath97 is the capacity of the bi - awgn channel as a function of the mean of the channel message , and @xmath98 then the mean of check - to - variable message from a check node type , @xmath18 along edge - type @xmath46 at the @xmath37 decoding iteration is given by @xmath99 @xmath75 and @xmath100 can be calculated from ( [ avg_mv ] ) and ( [ avg_mu ] ) , respectively .
as we discussed in section [ background ] , in the bp decoder , there are three types of messages : the channel message , the variable - to - check message , and the check - to - variable message .
we analyze the pdf of these messages on the bi - awgn channel to evaluate the gaussian assumption for de message pdfs .
let @xmath101 be a binary codeword ( @xmath102 ) on a bi - awgn channel . a codeword bit
, @xmath103 can be mapped to the transmitted symbol @xmath104 if @xmath105 and @xmath106 otherwise
. then , the @xmath46th received symbol at the output of the awgn channel is @xmath107 where @xmath108 is the energy per transmitted symbol and @xmath109 is the awgn , @xmath110 .
the llr ( @xmath111 ) for the received signal , @xmath112 is given by @xmath113 assuming that the all - zero codeword is sent and that @xmath114 is 1 , @xmath115 which is a gaussian random variable with @xmath116 = \frac{2}{\sigma_n^2}$ ] and @xmath117 = \frac{4}{\sigma_n^2}$ ] .
since the variance is twice the mean , the channel message has a symmetric gaussian distribution @xcite .
consider the variable node update in ( [ eq.vn update ] ) .
in the first iteration of the bp decoding , each variable node receives only a non - zero message from the channel .
hence the first set of messages passed from the variable nodes to the check nodes follow a symmetric gaussian pdf .
the following theorem describes the variable - to - check message exchanges in the @xmath40th , @xmath118 , iteration of the bp decoder .
[ variable node update ] the pdf of the variable - to - check message at the @xmath40th decoding iteration ( @xmath119 ) , is a gaussian distribution if all check - to - variable messages ( @xmath120 ) are gaussian .
if @xmath120s are not gaussian then the pdf of @xmath119 converges to a gaussian distribution as the variable node degree tends to infinity . the update rule at a variable node in ( [ eq.vn update ] )
is the summation of the channel message and incoming messages from check nodes ( @xmath120 ) .
since the channel is bi - awgn , @xmath39 follows a symmetric gaussian distribution . if all @xmath120s ( which are mutually independent ) are gaussian , then @xmath119 is also gaussian , because it is the sum of independent gaussian random variables@xcite
. if @xmath120s are not gaussian then the pdf of @xmath119 converges to a gaussian distribution as variable node degree tends to infinity , which directly follows from the central limit theorem @xcite .
if @xmath39 is non - zero and has a reasonably large mean compared to @xmath120 then @xmath39 minimizes the effect of non - gaussian pdfs coming from check nodes and tends to sway the variable - to - check message ( @xmath119 ) to be more gaussian .
moreover , @xmath119 can be well approximated by a gaussian distribution if the variable node degree is large enough . before analyzing the check - to - variable messages ,
let us first state a few useful lemmas and definitions , upon which our analysis is based .
[ lognormal ] if the random variable @xmath121 is gaussian distributed and @xmath122 , then random variable @xmath123 is said to be lognormally distributed .
[ sum of gaussain ] if @xmath124 are independent gaussian random variables with means @xmath125 and variances @xmath126 , and @xmath127 is a set of arbitrary non - zero constants , then the linear combination , @xmath128 follows a gaussian distribution with mean @xmath129 and variance @xmath130 .
[ lognormal sum_scaled ] let @xmath123 be a lognormal random variable .
then @xmath131 follows a lognormal distribution , where @xmath132 . since @xmath123 is a lognormal random variable then from definition [ lognormal ] , @xmath133 where @xmath121 is gaussian random variable .
according to lemma [ sum of gaussain ] , @xmath134 also follows a gaussian distribution .
thus from definition [ lognormal ] , @xmath135 follows a lognormal distribution .
[ remark_lognormal sum ] the assumption for the gaussian approximation is that the sum of @xmath136 independent lognormal random variables can be well approximated by another lognormal random variable .
this has been shown to be true for @xmath137 @xcite .
now consider the check node update in ( [ eq.cn update ] ) at the @xmath40th decoding iteration .
[ check node update ] the pdf of the check - to - variable message at the @xmath40th decoding iteration ( @xmath138 ) is a gaussian distribution provided that the variable - to - check messages are approximately gaussian and reasonably reliable , and the degrees of the check nodes are small . consider the check node with degree @xmath139 .
we can rewrite ( [ eq.cn update ] ) as follows : @xmath140 where we define @xmath141 and note that @xmath142 suppose @xmath143s are reasonably reliable . using taylor series expansion
, @xmath144 can be expressed as @xmath145 for simplicity , we omitted the indices @xmath146 , @xmath40 .
since @xmath147 follows an approximate gaussian distribution , according to definition [ lognormal ] , @xmath148 follows an approximate lognormal distribution and from lemma [ lognormal sum_scaled ] @xmath149 , @xmath150 follows an approximate lognormal distribution . using the assumption that , the sum of a set of independent lognormal random variables is approximately lognormal
when the set size is small ( see remark [ remark_lognormal sum ] ) , @xmath151 ( in step 1 ) follows an approximate lognormal distribution .
this is because , when @xmath147 is large with high probability , the higher - order terms in the taylor series expansion of @xmath144 are insignificant compared to the first few terms .
next , the @xmath152s are mutually independent .
thus , according to remark [ remark_lognormal sum ] , @xmath153 ( in step 2 ) follows an approximate lognormal distribution when the check node degree is small .
finally , from definition [ lognormal ] , @xmath154 ( in step 3 ) will follow an approximate gaussian distribution if the result of step 2 is lognormally distributed .
[ remark_lognormal sum_not_correct ] the assumption that , the sum of @xmath136 independent lognormal random variables can be well approximated by another lognormal random variable , is not true when @xmath136 is large .
using the above results , we investigate the accuracy of gaussian approximations to full - de of low rate met - ldpc codes , with punctured and degree - one variable nodes .
these are the cases where met - ldpc codes are most beneficial .
we also evaluate full - de simulations for these codes to measure how close the actual message pdf is ( under the full - de ) to a gaussian pdf using the kullback - leibler ( kl ) divergence @xcite as our measure .
a small value of kl divergence indicates that actual pdf is close to a gaussian pdf .
we calculate the kl divergence between 1 ) the actual message pdf ( under the full - de ) and a gaussian pdf with the same mean and variance to check whether it follows a gaussian distribution , 2 ) the actual message pdf ( under the full - de ) and a symmetric gaussian pdf with the same mean to check whether it follows a symmetric gaussian distribution .
+ in the case of standard ldpc codes , it has been observed that @xcite the check - to - variable messages significantly deviate from a symmetric gaussian distribution at low signal - to - noise ratios ( snr ) , even if the variable - to - check messages are close to a gaussian distribution .
thus gaussian approximations based on single - parameter models do not perform well for the codes at low snrs . here
we explain the reason behind this , based on the assumptions required for gaussian approximations to be accurate .
met - ldpc code with @xmath155 is compared with the symmetric gaussian pdf of the same mean . ] met - ldpc code with @xmath155 . ] according to remark [ check node update ] , the pdf of the check - to - variable messages ( @xmath138 ) can be well approximated by a gaussian pdf , if the variable - to - check messages ( @xmath143s ) are reasonably reliable given that @xmath143s are approximately gaussian and check node degrees are small . at low snr , the initial @xmath143s are not reasonably reliable .
thus @xmath138 may not follow a gaussian distribution in early decoding iterations . however , if the snr is above the code threshold , the decoder converges to zero error probability as decoding iterations proceed , thus the pdf of @xmath143 moves to right and the @xmath143s become more reliable .
hence @xmath138 may follow a gaussian distribution at later decoding iterations . to illustrate this via an example ,
we plot the actual message pdfs in the bp decoding and the kl divergence between the pdf of check - to - variable message from edge - type two and the corresponding symmetric gaussian pdf for a rate @xmath156 met - ldpc code in figs . [
fig : pdf_r ] and [ fig : kl_low rate ] , respectively .
it is clear from fig .
[ fig : kl_low rate ] that the kl divergence at a low snr has a larger value than that at high snrs .
this shows that @xmath138 significantly deviates from a gaussian distribution when the snr is low .
we can also see from figs .
[ fig : pdf_r ] and [ fig : kl_low rate ] that @xmath138 follows a gaussian distribution at later decoding iterations , and the lower the snr the more decoding iterations are required for this to happen .
based on these observations we claim that single - parameter gaussian approximations may not be a good approximation to de at low snrs .
+ met - ldpc code with @xmath157 at the first decoding iteration . ]
it has been observed @xcite that the check - to - variable messages significantly deviate from a symmetric gaussian distribution when the check node degree is large , even if the variable - to - check messages are close to a gaussian distribution .
thus single - parameter gaussian approximation models do not perform well for the standard ldpc codes with large check node degrees . here
we explain the reason behind this , based on the assumptions required for gaussian approximations to be accurate .
according to remark [ check node update ] , the pdf of the check - to - variable messages ( @xmath138s ) can be well approximated by a gaussian pdf , when check node degrees are small given that @xmath143s are approximately gaussian and reasonably reliable .
the assumption in step 2 ( see remark [ remark_lognormal sum ] ) , that is the sum of @xmath136 independent lognormal random variables can be well approximated by another lognormal random variable , is clearly not true if the check node degree is large ( see remark [ remark_lognormal sum_not_correct ] ) .
thus @xmath138 may not follow a gaussian distribution for larger the check node degrees . to evaluate the combined effect of the snr and the check node degree , we plot the kl divergence of check - to - variable message pdfs to the corresponding symmetric gaussian pdfs for a rate @xmath156 met - ldpc code at the first decoding iteration for different snrs and different check node degrees in fig .
[ fig : snrandcheckdegree ] .
our simulations show that with a large check node degree of 15 , the kl divergence is large . based on this
we claim that single - parameter gaussian approximations may not be a good approximation to de for codes with large check node degrees .
+ one of the modifications of met - ldpc codes over standard ldpc codes is the addition of punctured variable nodes to improve the code threshold ( a different use of puncturing than its typical use to increase the rate ) .
we observe that these punctured nodes have a significant impact on the accuracy of the gaussian approximation of both variable - to - check messages and check - to - variable messages .
according to theorem [ variable node update ] , if @xmath120s are not gaussian , the pdf of the variable - to - check messages ( @xmath119 ) converges to a gaussian distribution as the variable node degree tends to infinity . in the case of punctured nodes
, @xmath39 equals zeros as punctured bits are not transmitted through a channel .
hence in punctured variable nodes , @xmath119 is equivalent to the sum of @xmath120s only , which are heavily non gaussian at early decoding iterations .
thus if the variable node degree is not large enough , then @xmath119 from punctured variable nodes may not follow a gaussian distribution at early decoding iterations .
the punctured variable nodes adversely affect the check - to - variable messages as well . according to remark [ check node update ]
, the pdf of the check - to - variable messages ( @xmath138 ) can be well approximated by a gaussian pdf , if the variable - to - check messages ( @xmath143s ) are well approximated by a gaussian pdf .
this is because , @xmath144 in step 1 of ( [ check update 2 ] ) follows an approximate lognormal distribution only if @xmath119 is following a gaussian distribution . since @xmath119 from a punctured variable node
does not follow a gaussian distribution , @xmath138 also may not follow a gaussian distribution .
the end result is that punctured nodes reduce the validity of the gaussian approximation for @xmath119 and @xmath138 . to illustrate the effect of punctured nodes via an example , we plot the kl divergence of variable - to - check and check - to - variable messages to the corresponding symmetric gaussian pdfs of rate a @xmath30 met - ldpc code with punctured nodes in fig . [
fig : kl_p ] .
it is clear from fig .
[ fig : kl_varibale_p ] that the kl divergence of the variable - to - check message to the corresponding symmetric gaussian pdf from punctured nodes has a larger value than that from the unpunctured nodes in the same code .
[ fig : kl_check_p ] shows the corresponding effect on the kl divergence of the check - to - variable messages .
furthermore , the decrease of the kl divergence with decoding iterations in fig .
[ fig : kl_p ] implies that @xmath119 and @xmath138 are following a gaussian distribution at later decoding iterations .
however in general , to become gaussian it takes more decoding iterations than typical for a code without punctured nodes .
one of the advantages of the met - ldpc codes is the addition of degree - one variable nodes to improve the code threshold .
however we observe that degree - one variable nodes can affect the gaussian approximation for the check - to - variable messages . here
we explain the reason behind this , based on the assumptions required for gaussian approximations to be accurate . according to remark [ check node update ]
, the pdf of the check - to - variable messages ( @xmath138 ) can be well approximated by a gaussian pdf , if the variable - to - check messages ( @xmath143s ) are reasonably reliable given that @xmath143 are gaussian .
even though @xmath143s received from edges connected to degree - one variable nodes are gaussian , they are not reasonably reliable .
this is because variable nodes of degree - one never update their @xmath143s as they do not receive information from more than one neighboring check node .
so the @xmath138s may not follow gaussian distribution .
consequently , this may reduce the validity of the gaussian approximation to de for met - ldpc codes .
similarly any check node that receives input messages from a degree - one variable node never outputs a distribution with an infinitely large mean in any of its edge messages updated with information from degree - one variable nodes .
this is because , if @xmath158 are a set of independent random variables , then @xmath159 and @xmath160 as @xmath161 \rightarrow \infty$ ] @xcite , as is the case with degree - one variable nodes .
we observed through the simulations that the degree - one variable nodes have a small impact on the accuracy of the gaussian approximation of both variable - to - check messages and check - to - variable messages .
all the gaussian approximations we discussed in section [ gaussian app ] are based on the assumption that the pdfs of the variable - to - check and the check - to - variable messages can be well approximated by symmetric gaussian distributions .
this assumption is quite accurate at the later decoding iterations , but least accurate in the early decoding iterations particularly at low snrs or with punctured variable nodes or with large check node degrees as we have observed in section [ validity of ga ] .
making the assumption of symmetric gaussian distributions at the beginning of the de calculation produces large errors between the estimated and true distributions .
even when the true distributions do become gaussian , the approximations give incorrect gaussian distributions due to the earlier errors .
these errors propagate throughout the de calculation and cause significant errors in the final code threshold result for met - ldpc codes as we will see in section [ code design ga ] . through simulations , we observed that , when the channel snr is above the code threshold , the pdfs of the node messages ( i.e. , variable - to - check and the check - to - variable messages ) eventually do become symmetric gaussian distributions as decoding iterations proceeds .
this implies that making the assumption of symmetric gaussian distributions in the later decoding iterations of the de calculation is reasonable .
this motivates us to propose a hybrid density evolution ( hybrid - de ) algorithm for met - ldpc codes which is a combination of the full - de and the mean - based gaussian approximation ( approximation 1 ) .
the key idea in hybrid - de is that we do not assume that the node messages are symmetric gaussian at the beginning of the de calculation , i.e. , hybrid - de method initializes the node message pdfs using the full - de and then switches to the gaussian approximation .
there are two options for switching from the full - de to the gaussian approximation . as option one
, we can impose a limitation for the number of maximum full - de iterations in hybrid - de , in which we do few full - de iterations and then switch to the gaussian approximation .
although this is the simplest option , it gives a nice trade - off between accuracy and efficiency of threshold computation as shown in fig .
[ fig : met_threshold_it ] .
the second option is that we can switch from full - de to the gaussian approximation after that the pdfs for the node messages become nearly symmetric gaussian .
the kl divergence @xcite can be used to check whether a message pdf is close to a symmetric gaussian distribution .
thus , as the second option , we can impose a limitation for the kl divergence between the actual node message pdf and a symmetric gaussian pdf .
this is a more accurate way of switching than option one . because the value of the kl divergence depends on the shape of the pdf of the node messages , thus the switching point is changing appropriately with the condition ( such as snr , code rate ) we are looking at .
each option has its own pros and cons .
for instance , if the channel snr is well above the code threshold , the node messages can be close to a symmetric gaussian distribution before the imposed limit in option one for the full - de iterations .
thus we are doing extra full - de iterations that are not necessary .
this reduces the benefit of hybrid - de by adding extra run - time . in such a situation
, we can introduce the second option ( i.e. , kl divergence limit ) in addition to reduce run - time by halting the full - de iterations once the pdf is sufficiently gaussian . on the other hand ,
if the channel snr is below the code threshold , the decoder never converge to a zero - error probability as decoding iterations proceed , thus node messages may not ever follow a symmetric gaussian distribution .
this makes the option two hybrid - de always remain at full - de , as it never meets the target kl divergence limit . thus forcing a limitation for the number of full - de iterations ( i.e. , option one ) is required in order to improve the run - time of hybrid - de . because of these reasons
, we can introduce both options to the hybrid - de where option one acts as a hard limit and option two acts as a soft limit .
this is a particularly beneficial way to do the trade - off between accuracy and efficiency when computing the code threshold .
we found that it is possible to impose both options in the hybrid - de to significantly improve computational time without significantly reducing the accuracy of the threshold calculation .
t. j. richardson , m. a. shokrollahi , and r. l. urbanke , `` design of capacity - approaching irregular low - density parity - check codes , '' _ ieee trans .
inform . theory _
47 , no . 2 ,
619637 , feb .
2001 .
chung , t. j. richardson , and r. l. urbanke , `` analysis of sum - product decoding of low - density parity - check codes using a gaussian approximation , '' _ ieee trans .
inform . theory _
47 , no . 2 ,
pp . 657670 , 2001 .
l. schmalen and s. t. brink , `` combining spatially coupled ldpc codes with modulation and detection , '' in _ proc .
itg conf . on systems ,
commun . and coding ( scc)_.1em plus 0.5em minus 0.4emvde , 2013 , pp . 16 . | this paper considers density evolution for low - density parity - check ( ldpc ) and multi - edge type low - density parity - check ( met - ldpc ) codes over the binary input additive white gaussian noise channel .
we first analyze three single - parameter gaussian approximations for density evolution and discuss their accuracy under several conditions , namely at low rates , with punctured and degree - one variable nodes .
we observe that the assumption of symmetric gaussian distribution for the density - evolution messages is not accurate in the early decoding iterations , particularly at low rates and with punctured variable nodes .
thus single - parameter gaussian approximation methods produce very poor results in these cases . based on these observations ,
we then introduce a new density evolution approximation algorithm for ldpc and met - ldpc codes .
our method is a combination of full density evolution and a single - parameter gaussian approximation , where we assume a symmetric gaussian distribution only after density - evolution messages closely follow a symmetric gaussian distribution .
our method significantly improves the accuracy of the code threshold estimation .
additionally , the proposed method significantly reduces the computational time of evaluating the code threshold compared to full density evolution thereby making it more suitable for code design . belief - propagation , density evolution , gaussian approximation , low - density parity check ( ldpc ) codes , multi - edge type lpdc codes . |
in the uv , as in the optical , the light which we receive from a nebula should in great part be the light of the illuminating star scattered by dust grains in the nebula . if no other process than scattering of starlight by the average population of interstellar grains is involved , the ratio of the nebula to the star spectrum depends on the optical depth of the nebula , on the angle @xmath1 of scattering , and on the albedo and phase function of interstellar grains .
the original idea of this work was to study the behavior of the spectrum of nebulae in the 2200 region . in this spectral region ,
the spectrum of the stars is known to have a characteristic depression ( the 2200 bump ) when there is interstellar matter between the star and the observer .
the possible causes of the bump are reviewed by bless & savage @xcite . admittedly , the bump is an extinction feature ( eventually a purely absorbing one , witt et al .
@xcite ) which arises due to a particular type of interstellar grain .
since a nebula placed in front of a star produces a bump in the spectrum of the star , what will the bump spectral region of a nebula illuminated by a star at close angular distance be ?
if the particles which extinguish starlight at 2200 are present in the nebula , its spectrum should have an evident feature at 2200 .
this feature will be an absorption one if the bump carriers are purely absorbing particles .
it will be an excess of emission if the carriers scatter starlight .
section [ data ] presents the iue observations used in this work .
it consists of the spectrums of a few well known nebulae , all illuminated by close bright stars .
section [ analyse ] to section [ analyse2 ] , is an analysis of the data .
it is a remarkable fact , and the most significant result of this analysis , that all the nebulae have a spectrum which is exactly the product of the illuminating star spectrum and a linear function of @xmath0 ( section [ analyse ] ) .
some nebulae do not have a bump , and when they do have one it is always proportional to the bump of the illuminating star .
no bump is created in the nebulae . in section [ int ] ,
the relation between the spectrum of a nebula and the spectrum of the illuminating star is interpreted as scattering of starlight by a medium with an optical depth which varies as @xmath0 . a second result , the power law dependence of the surface brightness on the angular distance to the star , is arrived at and discussed in sections [ analyse2 ] and [ rdis ] .
the large variations of the surface brightness level of a nebula with the distance to the star may be an important argument in favor of a strong forward scattering phase function in the uv ( section [ rdis ] ) .
it implies that a considerable amount of starlight is scattered in directions close to the star .
different authors ( e.g. a.n . witt and collaborators ) have found a strong forward scattering phase function of the grains in the uv .
these papers are associated with models which also predict changes of the albedo and/or the phase function with the wavelength . from analysis of the present data
, i can not agree with these conclusions .
contrary to these studies , i find constant albedo and phase function of the grains in all the uv , and perhaps up to the near infrared , in all directions of space . in section [ grains ] , a few of these papers are briefly discussed along with the consequences the present paper have on the properties of interstellar grains .
it is suggested that there is no excess of extinction in the spectrum of the stars at 2200 .
an appendix will review elementary properties of scattering and the notations used in the paper .
all the observations used in this paper were made by the international ultraviolet explorer .
the iue satellite and the detector system are described by boggess et al .
recent and complete information on the project and data reduction are available at the iue website : http://iuearc.vilspa.esa.es .
the camera has two entrance apertures and two dispersion modes .
observations can be made through a large aperture ( @xmath2 ) , @xmath3 wide , or through a small one , ( @xmath4 ) , @xmath5 large in diameter .
@xmath4-aperture observations of a star generally need to be multiplied by a factor of @xmath6 to @xmath7 to match @xmath2-aperture observations of the same star .
the difference is attributed to the difficulty of holding the star within the beam when using the small aperture .
observations can be made in a low resolution mode of @xmath87 or with a higher spectral resolution of 0.10.2 .
three cameras : the lwr , the lwp , and the swp , have been used during the 20 year - long existence of the satellite .
the swp camera covers the wavelength range [ 1150 , 1980 ] , while the lwr and the lwp cameras cover the [ 1850 , 3350 ] range .
the lwr camera was the ` standard ' long wavelength camera from the beginning of the mission up until october 1983 . a problem with the camera s electronic led to a switch to the lwp .
the total range in wavelengths is from 1150 to 3350 .
the data at both boundaries should be carefully considered .
an iue spectrum gives the power per unit surface and per wavelength received by the satellite in the u.v . from the direction of a celestial object .
a typical extracted spectrum has the @xmath9-coordinate in absolute flux units ( erg/@xmath10/s / ) and the @xmath11-coordinate in .
the sensitivity of the observations is limited to a few @xmath12 erg / s/@xmath10/ .
the iue data can not be used straight away .
a transfer function , which is to be applied to the raw data , and which considerably modifies the original data files , is provided by the iue data reduction team .
i have used the .mlxo and .mxhi files which were up until recently available at the iue website .
those files are the final products of the processing from raw data to final spectrums .
following iue recommendations , the files were read by the idl software readmx.pro procedure provided in the iue package newsips for idl .
@xmath13 [ table-1 ] i have used the spectrums of nebulae ( iue class 73 ) and of stars . for each object ,
the spectrum presented in the paper is an average of the best observations of the object .
i sometimes used high dispersion spectrums and decreased the resolution by using a median filter .
observations of nebulae with clearly identified illuminating sources are not so frequent .
most were studied although they are not all presented here .
the results of the paper hold for the other nebulae as well .
the main properties of the nebulae are summarized in table [ tbl : neb ] .
i found it unnecessary to specify the iue observations name of the objects listed , since they can easily be looked at or retrieved from the iue website .
a reference list of the iue images is available upon request .
all nebulae were observed at different distances to the star .
different observations of a nebula with the same camera are often proportional when the position of the observations are close .
this proportion is justified by the proximity of the observations and indicates equal optical depths and similar mediums .
they may differ when the positions are too far apart due to modifications of the nature of the interstellar medium . when different positions to the star were exactly proportional , i chose to present either the best spectrum or the average one , so as to reduce the noise .
regions illuminated by the same star , with spectrums which can not be superimposed , were considered as separate nebulae .
the spectrums of the same object taken with the swp camera or with one of the long wavelength range cameras usually join in the common spectral region ( @xmath819002000 ) .
a slight gap between the long and the short wavelength spectrums sometimes happens and has been attributed either to a calibration problem or to a problem of precision in the pointing . in this case a correcting factor was applied to the long wavelengths range spectrum to ensure continuity from one range of wavelengths to another .
this scaling has no effect on the relations which will be obtained between the spectrums of a nebula and that of its illuminating star since these relations apply individually to each camera .
fig . [ 17tau ] , left , plots the spectrum of a nebula and its illuminating star , ( ) , as a function of @xmath14 .
the two bottom plots are two positions observed by the iue satellite in the . in both cases ,
the illumination is due to , a few arcminutes apart .
an intermediate position , 2 , not represented here , was also observed .
the spectrum of the 2 matches the 1 spectrum when multiplied by 3 .
stars and nebulae spectrums are not proportional , but fig .
[ 17tau ] shows that the spectrum of a nebula is exactly overlaid by the spectrum of the illuminating star , after multiplication by a linear function of @xmath0 .
it is worth noticing that no bump is created in the nebulae .
the preceding fit of the emission of a nebula by the product of the illuminating star flux and a linear function of @xmath0 can successfully be repeated for all the nebulae observed by the iue satellite , whether they have a bump or not .
[ nebfig ] shows the results for 8 nebulae and gives 3 coefficients , @xmath15 .
@xmath16 is the coefficient of the zero order fit , restricted to the long wavelength s range , of the spectrum of the star to the nebula spectrum .
the star spectrum , after smoothing by a median filter for sake of clarity , is multiplied by @xmath16 and represented by the curve in dashes .
the nebula spectrum is in dots .
it is fitted by the star spectrum multiplied by a linear function of @xmath0 , the overlaid solid line curve .
the relation between the spectrums can always be written as : @xmath17 where @xmath18 and @xmath19 are the nebula and the star fluxes measured by the iue satellite .
@xmath20 is a coefficient which must depend on the albedo and the phase function of interstellar grains and on the relative position of the star , the nebula and the observer .
@xmath21 can be 0 , 1 or @xmath221 .
the relation which is found between the uv emission of a nebula and its illuminating star should still apply to their uv fluxes corrected for foreground extinction : the star and the nebula are close enough to be equally ( in proportion of their flux ) affected by interstellar material situated between the star and the nebula on the one hand and the observer on the other hand . since they are proportional , the bumps of a nebula and its illuminating star should be attributed to dust in front of both the nebula and the star . [ figure-1 ] .
[ nebfig ] [ figure-2 ]
the proportion of starlight which is scattered by a nebula is evaluated by the @xmath23 parameter .
@xmath24 can theoretically be defined as the ratio of the nebula surface brightness to the star flux , both measured at the same position and corrected for foreground extinction .
slight or no reddening of the star s light between the star and the scattering volume should also be assumed .
the maximum proportion of starlight which can be scattered in the direction @xmath1 is given by @xmath25 , where @xmath26 is the maximum surface brightness the nebula can reach ( corresponding to an optical depth @xmath27 ) at @xmath1 .
@xmath28 depends solely on the angle of scattering @xmath1 .
column 5 of table [ tbl : neb ] gives the non - dimensional quantity @xmath29 for each of the observations .
@xmath29 was estimated by taking the maximum value ( over the @xmath30 $ ] iue wavelength range ) of @xmath31 , the ratio of the maximum surface brightness of the nebula to the star flux , normalized to a distance of @xmath32 to the star .
@xmath33 arcsec@xmath34 is the surface of the iue camera large aperture . if , as will be suggested in sect .
[ intopt ] , the mean optical depth of the nebulae is close to @xmath35 , @xmath29 should be close to @xmath36 .
the range of values the observed maximum of @xmath29 takes ( last column of table [ tbl : neb ] ) is quite narrow . within the restrictions due to the margin of error discussed in the following section
, @xmath37 can be taken to be @xmath38 , for @xmath39 between @xmath40 and a few arcminutes . from the present data ,
it is not possible to discern whether the observed maximum of @xmath29 in table [ tbl : neb ] is a constant or varies with the nebula .
table [ tbl : neb ] suggests a power law dependence of the nebulae surface brightness with angular distance to the illuminating star .
for the observations of nebulae in the same region ( e.g. , and nebulae ) , illuminated by the same star , @xmath29 is remarkably constant .
when different positions of the same nebula are available and proportional , the data show that there are strong variations of the brightness with angular distance to the star but @xmath29 stays relatively constant .
differences between the observed values of @xmath24 , reported in the table [ tbl : neb ] , and its theoretical definition can arise from a mistaken evaluation of either the nebula surface brightness or the unreddened flux of the star . a major difficulty when estimating the exact surface brightness is the width of the beam of the observations , which in many cases has dimensions comparable to the distance to the star .
it should be reliable when the dimensions of the beam are small compared to the distance to the star but errors may arise when the observation is made at close distance to the star ( a few @xmath41 ) . in the latter case ,
the variation of the nebula surface brightness between the two ends ( facing and opposite the star ) of the @xmath2-aperture , and its orientation , have to be taken into account .
the margin of error can be estimated by comparing the brightnesses obtained from observations at the same position with the @xmath2-aperture on the one hand , and from the @xmath4-aperture on the other hand .
the expected ratio is 7 .
few nebulae were observed with the @xmath2 and the @xmath4 aperture .
the ratio of the brightnesses in these observations ranges from 7 , as expected , to 100 .
although the pointing may explain part of this large factor , the exact surface brightness of a nebula , estimated by dividing the brightness of the nebula ( @xmath42 in eq .
( [ transfert ] ) ) by the solid angle intercepted by the @xmath2aperture , may be overestimated by a factor of 10 , for positions at less than a few @xmath41 to the star .
the second uncertainty in the evaluation of @xmath24 comes from the radiation field .
if the starlight is extinguished between the star and the scattering volume and if material in front of the star does not compensate for this extinction , the observed value of @xmath24 is superior to the exact one .
conversely , if there is more extinction between the star and the observer than between the nebula and the observer , the observed value of @xmath24 will be an inflated estimate of the exact one .
table [ tbl : neb ] shows a tendency for the observed maximums of @xmath29 to increase with the reddening of the star .
this increase may in part be explained by extinction of starlight in the vicinity of the star .
the stars and have no bump and are slightly reddened ( @xmath43 , table [ tbl : neb ] ) . the nebulae emissions , which have no peculiarities at 2200 , can safely be attributed to scattering .
the linear transformation which permits passage from the star to the nebula spectrum is to be interpreted as the slope of the surface brightness versus the optical depth curve ( fig .
[ bumpfig ] ) in the optical depth range of the spectrum .
the same should hold for nebulae with bump .
( [ transfert ] ) expresses that around the mean optical depth @xmath44 of the nebula and for the wavelength range under consideration , @xmath42 variations with @xmath45 can be approximated using the slope of @xmath46 ( fig .
[ bumpfig ] ) at @xmath44 .
if @xmath47 , @xmath18 grows with @xmath45 and @xmath44 must be less than @xmath48 . if @xmath49 , @xmath50 , the spectrums are close to being proportionate .
last , if @xmath51 , @xmath52 .
the @xmath53 coefficient has a remarkable property .
@xmath54 is the only source of @xmath42 variations with @xmath14 and must be proportional to the proportion of scattered starlight ( restricted to @xmath45-values close to @xmath44 ) , the @xmath9-axis of fig .
[ bumpfig ] .
hence , @xmath55 will always correspond to the point where the slope crosses the abscissa , with @xmath56 scaled to be @xmath0 . if @xmath47 ; the slope is positive , @xmath57 , and @xmath44 increases when @xmath58 decreases ( fig .
[ bumpfig ] ) .
if @xmath51 ; the slope is negative , @xmath52 , and @xmath44 increases with @xmath59 , if @xmath60 is not too large .
the @xmath55 coefficient provides a way to classify nebulae according to their optical depths .
the spectrums of fig .
[ nebfig ] are plotted by increasing optical depths , according to the considerations of the previous paragraph .
two nebulae only , nebula 2 and have their spectrums in exact proportion to the spectrum of the star .
these nebulae have an optical depth @xmath44 equal to @xmath48 .
their surface brightness , as a function of the optical depth , is at its maximum .
there are striking relations between the optical depth and the changes in the shape of the uv spectrum of fig .
[ nebfig ] . increasing the optical depth of a nebula
will soften the slope of the rise between 2200 and 1300 .
the effect can be observed by comparing the spectrums of the stars and the nebulae . increasing the optical depth also attenuates the fuv emission .
both phenomenae are the consequences of the increase of absorption toward shorter wavelengths , and contribute to the emphasis of the bump s visual impression .
it is particularly evident for .
most of the nebulae spectrums are close to being in proportion with the illuminating star s spectrum : for each nebula the zero order fit to the spectrum of the star is in reasonable agreement with the spectrum of the star .
it can be interpreted as @xmath44 values which are close to @xmath48 .
@xmath24 , calculated as indicated in section [ analyse2 ] , must be close to @xmath28 , defined in the same section . the interpretation which is given of the eq .
( [ transfert ] ) involves only the change in optical depth with wavelength .
there is no evidence or need for a variation of the albedo or of the phase function of interstellar grains in the wavelength range [ 1000 , 3000 ] of the iue observations .
there is also no evidence for variations of the properties of the grains with the direction of space .
the analysis of eq . ( [ transfert ] ) , has focused on the variation of the brightness of a nebula with wavelength , which depends on variations of the optical depth only .
the effect of the phase function of interstellar grains on the brightness of a nebula will be more evident through the comparison of the brightnesses of different observations because each of these will correspond to different angles of scattering , @xmath1 in fig .
[ schema ] .
iue observations are particularly useful in this respect since they provide observations of the same nebula at different angular distances to the star and observations of nebulae illuminated by different stars .
assume an isotropic phase function .
@xmath42 will depend on the radiation field at the position of the scattering volume and not on @xmath1 .
the large variations of the brightness with the distance to the star ( @xmath61 , section [ analyse2 ] ) will be justified only if @xmath39 is proportional to the distance @xmath62 of the scattering volume to the star .
it implies that the nebulae are all approximately at the same distance as the star from the observer : @xmath63 and @xmath64 .
the surface brightness of the nebula is : @xmath65 with @xmath66 the luminosity of the star and @xmath67 the proportion of scattered starlight by a medium of optical depth @xmath68 in direction @xmath1 . for @xmath69 ,
the simple model developed in zagury , boulanger and banchet @xcite , shows that @xmath70 should be at least @xmath71 .
@xmath37 will be found to be : @xmath72 @xmath73 even though there is some uncertainty in the observed values of @xmath37 ( @xmath80.003 , section [ analyse2 ] ) , isotropic scattering probably implies an albedo of at most @xmath74 .
this very low albedo , and the systematic configuration in which all the nebulae need to be to justify isotropic scattering , are improbable .
a more likely explanation for the variation of the surface brightness with the angular distance to the star can be attributed to variations of @xmath1 rather than variations of @xmath62 .
the quick decrease of the surface brightness with @xmath39 , i.e. with @xmath1 , indicates a strong forward scattering phase function . assuming a strong forward scattering phase function , the scattering volumes are likely to be in front of the stars , in which case we have @xmath75 ( fig .
[ schema ] ) .
three independent factors will modify @xmath26 : the star luminosity , @xmath76 , the distance to the star , @xmath62 , and the angle of scattering , @xmath1 .
the surface brightness of a nebula is proportionate to @xmath76 and to @xmath77 .
the remaining dependence on @xmath1 will be defined by the function @xmath78 . then : @xmath79 the dependence of @xmath80 on @xmath81 in equation [ eq : hphi ] was e.g. verified by the observations ( section [ analyse2 ] ) .
concerning the observations for which @xmath36 is nearly constant ( @xmath82 ) , we have : @xmath83 .
since @xmath84 does not depend on @xmath39 this relation must be true for the range of @xmath1values considered by the observations .
it suggests that the differences between the values of @xmath85 in table [ tbl : neb ] is due to errors on the estimate of either @xmath24 ( section [ subsec4.1 ] ) or @xmath39 .
hence , the maximum surface brightness a nebula can reach in the direction @xmath1 from the direction of the star varies as : @xmath86 eq .
( [ eq : sb ] ) applies to @xmath87 values between a few @xmath41 ( or less ) and a few @xmath88 .
an important consequence of the relation found between the spectrum of the nebulae and the spectrum of the illuminating stars concerns the 2200 bump . in none of the uv observations of the nebulae presented here is there evidence for the presence of the 2200 carriers . if starlight is not affected by the presence of the bump carriers when the nebula is on the side of its illuminating star , why should it be affected when the nebula is in front of the star ? from eq .
( [ eq : sb ] ) it is clear that considerable starlight can be scattered into the beam of the telescope at close angular distance to a star . if @xmath89 is the smallest angle down to which eq .
( [ eq : sb ] ) holds , and if we assume @xmath28 to be constant for @xmath90 , the proportion of starlight scattered into the beam of semi angle @xmath91 can reach : @xmath92 it can easily be a few percent of the direct starlight corrected for extinction
. the proportion of scattered starlight into the beam will be even more important when compared to the direct starlight @xmath19 received by the telescope , since the latter is extinguished .
the implication may be that there are no bump carriers .
it was noted by bless & savage @xcite ( see also cardelli & clayton @xcite pp .
519 and 520 ) that the bump will also be explained if enough starlight scattered in the beam of the observation was added to the direct starlight .
this possibility was ruled out ( bless & savage @xcite ) on the basis of arguments which do not hold if grains strongly scatter light in the forward direction .
contrary to witt @xcite , who finds variations of grain properties in one direction to another , i find homogenous properties of interstellar grains in all directions and constant albedo and phase function in the uv .
witt & bohlin @xcite claim the discovery of enhanced scattering emission in reflection nebulae associated with the 2175 extinction band but no such irregularities can be observed in any of the nebulae presented here .
the spectrums of and presented in witt & bohlin are close to the raw data which may mean they have not been treated according to iue most recent recommendations .
consequently , large differences exist between the nebulae spectrum in witt & bohlin and the ones presented at the iue website and reproduced here .
the long wavelength range observation of used by the authors , lwr 17725 , on which most of their discussion relies , is classified ` @xmath93 ' or ` abnormal reading ' and ` non standard acquisition ' at ines archive data mirror .
the spectrums of and presented in fig .
[ nebfig ] are an average of the best observations of the nebulae , most of which were not available at the time witt & bohlin did their work .
the spectrums have no unique features , except for the large optical depth met in .
it is most probable that witt & bohlin s analysis as well as their conclusions are questionable .
another study , the calzetti et al .
@xcite paper on , uses the same data as that found in the present paper .
the authors have followed the witt & bohlin analysis and attribute the variation of @xmath94 with wavelengths to changes in the albedo and/or phase function of the grains .
they did nt search for a relation between the star flux , the nebula surface brightness and @xmath0 .
[ figure-3 ] the power law dependence of the surface brightness of with distance to the star was investigated from the optical to the uv by witt & cottrell @xcite and witt et al .
@xcite .
the power law which is found in these two papers has an exponent @xmath81.5 in absolute value , smaller than the one found in this paper .
this exponent is not the one which can be deduced from the raw data since the authors try to include a correction for the width of the aperture , which introduces an error on the surface brightness estimate at close angular distance to the star .
the raw data are not given in witt et al .
the observational results given in witt & cottrell @xcite for the optical wavelengths give an exponent of @xmath7 ( fig .
[ witt80fig ] ) .
other observational data , for instance the witt & schild @xcite study of fifteen reflection nebulae , give a similar exponent ( zagury , submitted ) . precise comparison of the different papers can not be broached here .
an exponent of 2 is logical within the present analysis .
the exponent of 1.5 found in witt et al .
@xcite is also justifed by the model developed in their paper .
another idea which emerges from the comparison of the data in the different papers mentioned in this section , is that the uv and optical wavelengths for which @xmath29 can be calculated give similar values .
if true , it may imply similar properties ( albedo and phase function ) of interstellar grains from the near infrared to the far uv .
comparison of the uv spectrum of several nebulae observed by the iue satellite to the spectrum of the illuminating stars shows that the spectrum of a nebula is always the product of two terms : the flux of the star @xmath19 and a linear function of @xmath0 .
this function was interpreted as the slope of the surface brightness @xmath4 of the nebula versus optical depth at the optical depth @xmath44 of the nebula .
the variation of @xmath23 with wavelength , and of its maximum with the angular distance @xmath39 to the star , provides additional informations on the properties of interstellar dust and on the nebulae .
the linear @xmath0 function gives an estimate of @xmath44 , since the slope of @xmath95 decreases with @xmath68 .
it allows a classification of the nebulae according to their mean optical depth .
for all the nebulae of the sample , @xmath44 should be close to @xmath48 , the optical depth at which maximum surface brightness is reached .
this is in conformity with the high brightnesses of the nebulae .
the important variations of the ratio @xmath23 with the angular distance @xmath39 to the star is explained if the interstellar grains have a strong forward scattering phase function in the uv . in some observations the maximum value of @xmath24
, @xmath28 varies as @xmath96 .
this result is justified if the phase function is so that @xmath28 is proportional to @xmath97 , where @xmath1 is the angle of scattering .
no additional absorption or other particular behaviour at 2200 was noticed in the spectrum of the nebulae .
the absence of bump carriers in the nebulae calls to question the interpretation of the 2200 bump in the spectrum of the stars . when a star is observed behind a nebula a considerable amount of starlight scattered into the beam
can be added to the direct starlight and may explain the bump .
the understanding and interpretation we have reached of the uv spectrum of nebulae applies similarly to different regions of the interstellar medium .
it indicates that the uv properties ( albedo and phase function ) of interstellar grains are identical in all directions of space
. these properties may be the same from the near infrared to the far uv .
[ figure-4 ] elementary properties of scattering used in this paper are summarized by the plot of fig .
[ bumpfig ] .
the plot represents surface brightness variations with optical depth for a cloud illuminated by a far away source ( the source of light comes from one direction ) .
the @xmath11-axis of the plot is the optical depth , related to the quantity of interstellar grains photons have crossed on their way to the telescope .
the optical depth depends on wavelength and on the column density along the path the photons have followed . optical depth and column density
are proportional .
optical depth is a linear function of @xmath0 .
thus , depending on the kind of observation which is involved , spectroscopy or photometry , @xmath68 will vary from a point to another according to the wavelength or to the quantity of matter on the photons way .
the quantity of matter on the photons way depends on the angle made by the observer , the source radiation field and the nebula . only for directions close to the illuminating source
can @xmath68 be assimilated to the cloud column density .
the @xmath9-axis is the nebula surface brightness scaled to 1 at maximum surface brightness . at very small @xmath68 ,
variations of @xmath4 and @xmath68 are proportional .
absorption will quickly diminish the slope d@xmath4/d@xmath68 .
@xmath4 will have a maximum at a value @xmath98 and decreases for @xmath99 .
for any given column density , there is a wavelength at which @xmath98 is reached and reciprocally .
hence there is a correspondence between wavelengths and column densities which allow to predict at which wavelength a cloud will preferably be observed and the possible range of column densities of a cloud bright in a certain wavelengths range . for a constant albedo
, @xmath100 should depend solely on the phase function , i.e. only on the angle @xmath1 of the source radiation field direction with the direction of the scattering volume and the observer .
i thank professor y. fukui for the support and freedom i took benefit from during two years in japan and without which this work would not have been possible .
i received constant help and a warm welcome from all the radio astronomy group of nagoya university .
a special thank to dr onishi and to dr mizuno . | this paper presents an analysis of the uv spectrum of some nebulae with clearly identified illuminating stars , all observed by the iue satellite .
the data show remarkable properties of the uv spectrum of the nebulae .
each spectrum is the product of the star spectrum and a linear function of @xmath0 .
there is no peculiar behaviour in the spectrums at 2200 : no bump created in the spectrum of a nebula and no excess of scattering . when moving away from the star , the surface brightness of a nebula decreases as the inverse of the square of the angular distance to the star .
these results can logically be interpreted in terms of scattering of starlight .
they imply constant properties of the interstellar grains in the uv and in the directions of space sampled by the nebulae , and probably a strong forward scattering phase function .
there is no evidence for any particular type of grain which would specifically extinguish starlight at 2200 . concerning the uv spectrum of a star
, this may imply a revisal of the traditional interpretation of the 2200 bump . |
a very elegant mechanism for obtaining lattice chiral fermions was recently proposed @xcite .
the idea is to start from wilson fermions in @xmath0 dimensions , where @xmath1 is the dimension of space - time , and give these fermions a mass @xmath2 of the order of the cutoff .
however , a domain wall like defect is introduced by making the mass term dependent on the extra dimension , choosing @xmath3 for @xmath4 and @xmath5 for @xmath6 ( @xmath7 labels the coordinates of the extra dimension ) .
it was shown that a massless mode with positive chirality exists in this model @xcite , which is exponentially bound to the defect at @xmath8 , which is identified with @xmath1 dimensional space - time . in a finite volume
an antidomain wall exists due to the ( anti)periodic boundary conditions , and consequently , a negative chirality mode , bound to this defect , exists ( other choices of boundary conditions are possible , but do not alter the conclusions @xcite ) .
one can study this system coupled to external smooth gauge fields , and one finds that any anomalous charge created on the domain wall by this gauge field is carried off through a goldstone - wilczek current to the antidomain wall @xcite . in this work ,
we study the coupling to dynamical gauge fields .
a detailed account can be found in ref .
@xcite , to which we also refer the reader for notations .
we introduce a waveguide @xmath9 , with @xmath10 on one side , and @xmath11 on the other side of the domain wall . only within the waveguide ,
the fermions are coupled to a dynamical gauge field @xmath12 .
this breaks gauge invariance in the fermion hopping terms across the waveguide boundary , which is restored by promoting these terms to yukawa couplings with a scalar field @xmath13 , which takes values in the gauge group @xmath14 @xcite . for another approach with @xmath7-independent gauge fields ,
the model is defined by the action @xmath15 \nonumber \\ & - & y(\psb^{s_0 - 1}vp_l\psi^{s_0 } + \psb^{s_0}v^{\dg}p_r\psi^{s_0 - 1 } ) \nonumber \\ & - & y(\psb^{s_0'}v^{\dg}p_l\psi^{s_0'+1 } + \psb^{s_0'+1}vp_r\psi^{s_0'}),\end{aligned}\ ] ] where @xmath16 is the @xmath7-dependent mass . for @xmath17 the regions inside and outside the waveguide decouple , and new zeromodes appear on one of the waveguide boundaries , with negative chirality just inside , and with positive chirality just outside the waveguide @xcite .
( they appear only at one of the boundaries : which one depends on details . )
the inside mode couples to the gauge field , rendering the theory vectorlike .
the key question now is whether this situation pertains for all values of the yukawa coupling @xmath18 , or that a pms phase exists for large values of @xmath18 @xcite , where the fermion spectrum around the waveguide boundary does not contain light fermion modes .
this phenomenon takes place in a number of higgs - yukawa models , and if this would be the situation in the case at hand , the unwanted modes at the waveguide boundary would decouple .
we emphasize that there are no simple anomaly arguments to rule out this possibility @xcite .
we have studied this question for @xmath19 , as it is the fermion - scalar dynamics that would be responsible for the existence of a pms phase .
this is reasonable , as @xmath12 ( in the landau gauge ) should be close to one in an asymptotically free theory . at @xmath17
the action can be diagonalized , and one finds two massless dirac modes , @xmath20 and @xmath21 , where @xmath22 is the right(left)-handed mode at the ( anti)domain wall , and @xmath23 is the left(right)-handed mode just inside ( outside ) the waveguide boundary . these modes all have a chiral spectrum for momenta in the region @xmath24 around @xmath25 @xcite .
modes with momenta outside this region are delocalized in @xmath7 , and their wavefunctions are negligible at the domain walls and waveguide boundary .
this includes all species doubler momenta @xmath26 _ etc_. in writing an effective action for the light modes , we may therefore ignore these large @xmath27 modes .
we will also ignore the yukawa couplings of the domain wall modes ( @xmath28 ) , since their wavefunctions are exponentially suppressed at the waveguide boundaries .
this leads us to an effective model for @xmath29 and @xmath30 : @xmath31 this form of the effective action is reminiscent of a hypercubic higgs - yukawa model @xcite , in that in both models the yukawa couplings are suppressed for large momenta .
this leads to the suspicion that no pms phase will exist in our model , as no such phase exist in the hypercubic model .
the obvious thing for a numerical investigation , in order to find out whether a pms phase exists or not , is to compute fermion masses .
however , in particular for large @xmath18 and small scalar hopping parameter @xmath32 ( the region of interest @xcite ) , the propagators become very small and noisy due to the strongly fluctuating scalar field , and we have not been able to compute reliable fermion propagators in that region ( we mostly used the quenched approximation ) .
we studied the eigenvalue spectrum of the dirac operator for the effective action , eq .
( 2 ) , for @xmath33 and @xmath34 , and compared it with @xmath34 higgs - yukawa models which are known to have or lack a pms phase .
the distribution of eigenvalues in the complex plane is a very good measure of the existence of the pms phase @xcite .
results are shown in fig .
1 , indicating that indeed no pms phase exists .
= 5 cm we have also computed the @xmath30 mass in the broken phase ( _ i.e. _ for large @xmath32 ) of the full model ( eq . ( 1 ) ) , where the signal to noise ratio is better , again for @xmath35 and @xmath34 .
( for a discussion of the finite volume @xmath34 scalar phase structure , see ref .
@xcite . ) results for the @xmath32 dependence are shown in fig . 2 for strong yukawa coupling @xmath36 , which are consistent with the weak coupling behavior @xmath37 for @xmath38 . near a fm - pms phase transition one
would have expected a fermion mass increasing with @xmath39 @xcite .
again , this indicates that no pms phase exists in our model .
we conclude with some remarks , most of which are more extensively discussed in ref .
@xcite .
@xmath40 we have tried unquenched computations , which resulted in very poor statistics , and very low acceptance rate of the hybrid mc algorithm . unquenched results were not inconsistent with our quenched computations .
@xmath40 the scalar field is nothing else than the longitudinal gauge field , and it is due to the fluctuations of this field that the model produces mirror fermions , rendering the model vectorlike : both @xmath41 and @xmath42 couple to the gauge field .
@xmath40 a crucial role is played by the effective momentum cutoff " @xmath43 , which is critical in removing the doublers .
however , from eq .
( 2 ) it is also clear that this cutoff plays an important role in the ( non)existence of a pms phase .
@xmath40 we have also considered a staggered fermion formulation of the domain wall approach , based on the notion that in that case no @xmath43 appears . however , in this case , the flavor structure causes similar problems @xcite .
@xmath40 there exists a proposal to keep the volume in the extra dimension strictly infinite @xcite , based on the idea that in that way the zeromode on the antidomain wall is avoided from the start .
@xcite discusses a possible relationship between this approach and the one reported on here
. 9 d.b .
kaplan , phys .
b288 ( 1992)342 .
y. shamir , nucl .
b406 ( 1993)90 .
k. jansen , phys .
b288 ( 1992)348 .
golterman , k. jansen and d.b .
kaplan , phys .
b301 ( 1993)219 .
golterman , k. jansen , d.n .
petcher and j.c .
vink , ucsd / wash.u .
preprint hep - lat/9309015 .
kaplan , nucl .
b ( proc.suppl . ) 30 ( 1993)597 . r. narayanan and h. neuberger , phys .
b302 ( 1993)62
. w. bock _ et al .
b344 ( 1990)207 and refs . therein .
k. jansen and m. schmaltz , phys .
b296 ( 1993)374 .
lee , j. shigemitsu and r.e .
schrock , nucl .
b330 ( 1990)225 .
i. barbour _ et al .
b368 ( 1992)390 .
golterman and j.c .
vink , unpublished .
r. narayanan and h. neuberger , rutgers preprints hep - lat/9307006 and 9308011 ; r. narayanan , these proceedings .
y. shamir , weizmann inst .
preprint hep - lat/9310006 . | we have investigated a proposal to construct chiral gauge theories on the lattice using domain wall fermions .
the model contains two opposite chirality zeromodes , which live on two domain walls .
we couple only one of them to a gauge field , but find that mirror fermions which also couple to the gauge field always seem to exist . |
arguably , one of the most fascinating recent astrophysical discoveries is the rich population of young luminous blue `` globular - like '' clusters in strongly interacting and merging systems ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* hereafter ws95 ) .
these clusters all appear much younger ( @xmath11 100 myrs ) than both the old globular cluster system and underlying background population and have dynamical masses ( m@xmath12few @xmath9 10@xmath13 m@xmath8 in a few cases * ? ? ?
* ) and effective radii very similar to the globulars ( r@xmath14@xmath1210 pc
the formation of such clusters appears to be generic to the merger process and is likely the result of the high pressures that are induced during the merger process ( e.g. , * ? ? ?
* ) . the star formation activity in ngc 4038/4039 the `` antennae '' has been known from studies at various wavelengths , for example from h@xmath4 spectroscopy @xcite or the very blue colors observed in the hst images ( ws95 ) .
its relative proximity ( 19 mpc assuming h@xmath15=75 km s@xmath16mpc@xmath16 ) makes it the ideal candidate for the study of merger induced star formation on scales of single star clusters and has allowed for the detection and investigation of a very large number ( @xmath1800 ) of compact clusters ( * ? ? ?
the heavily extincted overlap region was revealed to be the most active star formation zone by radio observations @xcite and imaging and spectroscopy performed with iso @xcite .
the overlap region is , however , relatively inconspicuous at optical wavelengths .
the iso results suggest very young ages for these clusters and the presence of high mass ( 50 - 60 m@xmath8 ) stars .
interestingly , a significant fraction of the total bolometric flux from the antennae appears to originate in two clusters eastward of the ngc 4039 nucleus ( numbers 86/87/88/89/90 in ws95 ) , building up an optically bright knot , and number 80 in ws95 , which is optically extincted and declared a `` very red object '' .
these two regions , which are the highest surface brightness features in the mid - infrared , contribute @xmath125% of the total flux at 15@xmath0 m @xcite and @xmath120% of the total emission from discrete radio sources in ngc 4038/4039 @xcite .
these large fractions suggest that these two regions must constitute a significant portion of the very recent or on - going star - formation within the antennae .
understanding their properties within the perspective of the overall star formation history of the antennae is of prime importance to understanding the bolometric output of these merging galaxies .
we discuss the properties of these regions as determined through ir integral field spectroscopy in combination with data available from the literature .
models of the merger of two spiral galaxies ( like that currently being observed in ngc4038/4039 ) predict multiple bursts of star formation during the merger event @xcite . depending on the stage of the merger
, these bursts can be in the region of overlap of the two galaxy disks , in the tidally shocked regions of the outer disks of each galaxy , or as significant quantities of gas fall in towards the two nuclei , in the nuclei of each galaxy .
we observed the two most prominent clusters within the interaction zone , two other ir bright clusters in the outer arms of ngc 4038/4039 ( likely to be tidally shocked regions ) , and the two nuclei .
such a selection would allow us ( at a minimum ) to probe the various regions of star - formation during the merger process within the inevitable constraint of only having a limited amount of observing time . while only observing six regions will not allow us to investigate the properties of the whole ensemble of compact clusters in the antennae , our sample allows us to crudely investigate the possible range of ages across the antennae .
figure 1 shows the two clusters in the interaction zone ( [ ws95]86 and [ ws95]80 ) , the two nuclei ( ngc 4038 and ngc 4039 ) and two clusters in the outer loops ( [ ws95]417/[ws95]428 in the eastern and [ ws95]405 in the western ) , observed with the 3d integral field spectrometer @xcite in the k band . for the observations ,
3d was combined with the tip - tilt - guider rogue @xcite on the anglo - australian telescope in february , 1998 and april , 1999 .
we used a k - band grism with a resolution of @xmath17 and a 256x256 nicmos 3 array .
256 spectra are obtained simultaneously , arranged in a 16x16 pixel field on the sky .
the detector integration time per exposure was 100 seconds .
the on - source exposures were interleaved with off - source ( offset 60e of each cluster ) exposures of identical integration time for the sky subtraction . to remove telluric features ,
the a2v star hd106819 was observed roughly once every hour . in order to nyquist sample the spectra ,
we dithered in the spectral direction by half a pixel using a piezo - driven mirror .
the spatial pixel scale of 04 provided a field of view of 64x64 .
the conditions were generally not photometric and flux calibration achieved via aperture photometry available from the literature @xcite in combination with ks - band imaging ( see also next section ) .
the seeing during the observations of each individual region only showed moderate variation ( @xmath1102 ) but was significantly different for different objects .
the values for each observation are discussed in section [ results ] .
the total on - source integration time was 6200s on the first field ( hereafter abbreviated to [ ws95]86 ) , which is the optically bright star cluster , corresponding to the numbers 86/87/88/89/90 in ws95 ; 4200s on - source time for the second field ( hereafter [ ws95]80 ) , which corresponds to the `` very red object '' number 80 in ws95 ; 3120s on - source time for the field in the eastern loop ( numbers 417 and 428 in ws95 ) ; 3840s for the targets in the western loop ( ws95 - number 405 and several fainter objects , hereafter [ ws95]405 ) .
the nuclei of ngc 4038 and ngc 4039 were observed for 3360s and 2160s respectively . due to a piezo failure ,
the spectrum of the nucleus of ngc 4039 is not nyquist sampled .
data reduction was performed using the 3d data reduction routines developed at mpe within the gipsy @xcite data reduction environment .
single frames were dark subtracted , linearized and sky subtracted using the sky frame that was closest chronologically and had the right spectral dithering position .
the wavelength calibration was accomplished by exposures of a neon discharge lamp .
the re - binning of the spectrally interleaved data was performed onto a 600 pixel linear wavelength axis .
bad pixels and cosmic rays were removed from the 16x16x600 data cube , either by interpolation or by masking out .
after spatial re - binning onto a 64x64 pixel grid , the single cubes were stacked by centering each on its k - band continuum peak or the br@xmath2 peak for [ ws95]86 and [ ws95]80 , respectively .
telluric features were removed by division by the atmospheric transmission profile obtained from the observations of the a2v star .
the br@xmath2 absorption line in the calibrator spectrum was interpolated linearly between the adjacent continuum . in the resulting cluster spectra
, only the ngc 4038 nucleus spectrum shows a little remaining br@xmath2 emission from the calibrator , which is not a worry , because it is at zero redshift and does not contaminate the br@xmath2 emission from the object .
object spectra were extracted using a square aperture centered on the continuum peak and sized to maximize the snr of the resulting spectrum .
the individual apertures and their sizes are shown in figure 2 . for fields which showed significant offsets between continuum and br@xmath2 line emission peaks ( [ ws95]405 and the nucleus of ngc 4038 ) ,
the aperture sizes were chosen so as to minimize contamination between the continuum and line emitting regions .
the spectra were flux calibrated using the k@xmath18-magnitudes from our flux calibrated sofi image , obtained in may 99 at the ntt . the extracted spectra were normalized by dividing by the continuum value , estimated using a linear fit . from this we estimated the equivalent widths w@xmath19 of the br@xmath2- and the he i - lines and the co - band - heads .
for the creation of linemaps , the linear fit was performed for each spatial pixel and subtracted .
the spectra of all the regions ( divided into subregions for some of the fields ) are shown in fig .
2 , together with the k@xmath18-band images . overlaid are the contours of br@xmath2 or the h@xmath20(1 - 0)s(1)-line ( labelled in fig .
results obtained from our data , incorporating results from publicly available hst data ( h@xmath4 and v - band images * ? ? ?
* ) , are listed in table 1 .
[ ws95]86 was slightly resolved while [ ws95]80 was unresolved ( both combined data sets had a final fwhm@xmath1210 ; fig .
the fwhm of the seeing was @xmath1208 during [ ws95]417&428 observations , and the two clusters are marginally resolved into single components . during [ ws95]405 observations , the seeing was poor ( 16 ) , but to the ne of the bright cluster ,
a fainter second cluster is visible .
the tabulated values for [ ws95]405-a and b refer to the bright cluster ( a ) and to the location of the br@xmath2 peak ( b ) .
we used the measured equivalent width of br@xmath2 , w@xmath21 , and co 2.29@xmath0 m band - head , w@xmath22 ( table 1 ) , to determine the ages of the starbursts using stellar synthesis models @xcite . see plots in @xcite ( their figures 89 and 101 ) , showing the evolution of these parameters with burst age . assuming an upper mass cutoff m@xmath23=100m@xmath8 , salpeter imf slope and an instantaneous burst at solar metallicity
, we derive ages of 3.7@xmath24yr for [ ws95]86 and 5.5@xmath25yr for [ ws95]80 . @xcite
derive a value of @xmath124 myrs for [ ws95]80 , assuming the same starburst parameters , in reasonable agreement with our value .
the ages of the two clusters in [ ws95]417&428 are not very well constrained , due to the low s / n of the data and the ambiguous value of the w@xmath22
. however , the values of w@xmath22 are sufficiently large to exclude additional errors introduced by the non - inclusion of the thermally pulsing asymptotic giant branch phase in the geneva stellar evolution tracks that were incorporated in @xcite @xcite .
age estimates of [ ws95]417&428 lie between 8.5 and 12.8 myrs old .
for the [ ws95]405 spectra , it is not possible to get a consistent model age with the presence of both strong co - band - heads and br@xmath2 emission with an instantaneous burst . but given the size of the cluster ( r@xmath26=12.3pc , * ? ? ? * ) it is plausible to assume that the star - formation has a finite duration of a few myrs @xcite .
other possibilities that might explain why these clusters have both co absorption and br@xmath2 emission are shock heating of the gas due to supernova explosions , the seeing broadened br@xmath2 emission from the region to the s - e of [ ws95]405-a which contaminates the cluster spectrum , supersolar metallicity , and possible gaps in our knowledge of stellar evolution and model atmospheres .
it is interesting to compare the 3d image and spectra of a region outside both [ ws95]405 clusters with the multi - color hst image , since our data show a br@xmath2 peak to the se of the k - band continuum peak of cluster a. this coincides with a dust filament obvious in the hst image , and it could be the site of more recent star formation or of shock excitation in a more dusty environment .
a detailed discussion of the nuclear spectra is beyond the scope of the current paper .
here we only wish to emphasize how the spectra of the two nuclei differ from the young compact clusters .
the w@xmath22 of the two nuclei suggest the same possible age range for both , 65@xmath2715 myrs , considerably older than the ages estimated for the prominent young clusters .
the northern nucleus also shows evidence for a more recent starburst to the north of the k - band peak , but it is difficult to estimate its age due to the underlying continuum contribution by the older background nuclear stellar population .
but the presence of br@xmath2 emission ( no evidence for agn activity was seen in other observations so far ) requires the presence of o or b stars or recent shock excitation , thus implying that it is younger than @xmath1210 - 20 myrs . under the assumption that there is no intrinsic gradient in w@xmath22 across the ngc 4038 nucleus , meaning that the observed lower w@xmath22 at the location of the br@xmath2 peak
can be attributed to the additional continuum from the young stellar population , we estimate the young star light continuum to contribute about 1/5 of the total continuum , which increases the observed w@xmath21 to 58 .
this then implies an age of the starburst of around 6 myrs , consistent with the effective temperature of 33000k derived from the low ratio of he i / br@xmath2 of 0.17 .
the age uncertainties given in table 1 only take into account uncertainties in the signal - to - noise of the data and do not take into account the systematic effects of selecting different imf slopes , upper and lower mass cutoffs , abundances , or uncertainties in the models themselves .
the extinction was derived by comparing the theoretical ratio of h@xmath4/br@xmath2 ( case b recombination ratio assuming t@xmath14=10000 k from * ? ? ?
* ) with the observed ratio .
using the extinction curve of @xcite for a foreground screen model , we derived a@xmath6 of [email protected] and [email protected] for [ ws95]86 and [ ws95]80 , respectively .
they agree within the errors with the values we derived from the v - k color excess with the intrinsic color derived from the models for the respective cluster ages ( a@xmath6 = 1.6@xmath28 and 3.7@xmath28 ) .
for [ ws95]417&428 and [ ws95]405 , we used the v - k colors not to confirm the extinction , but rather to further constrain the age .
the ratio of he i @xmath32.058@xmath0m / br@xmath2 is not an unambiguous tracer of the effective temperature of the highest mass stars present in a stellar population @xcite , but it was determined to compare with the iso results at higher spatial resolution .
the effective temperatures derived from the ratios of he i / br@xmath2=0.47 and 0.5 are 36,000 and 38,000k ( [ ws95]86 and [ ws95]80 , respectively ) , which corresponds to masses of roughly 28 m@xmath8 and is below the 44,000k ( 50 - 60 m@xmath8 ) derived from mid - infrared line ratios by @xcite .
given the sensitivity of the he i @xmath32.058@xmath0 m line to nebular parameters @xcite and the large aperture ( 14@xmath927 ) used for the iso measurement , some disagreement between our estimate and that obtained using iso data for the temperature of the hottest stars is not surprising .
moreover , the relationship between temperature of the hottest stars and he i / br@xmath2 is not monotonic and thus our estimates for the mass should be considered lower limits .
these upper mass limits are consistent with our estimated ages of the [ ws95]80 cluster , while the younger age of [ ws95]86 would allow for the detection of higher mass stars .
a qualitative comparison of the derived effective temperatures / ages of all fields agrees with the gradient in the @xmath29ne iii@xmath30$]/@xmath29ne ii@xmath30 $ ] observed by @xcite . from the extinction
corrected br@xmath2 flux we can estimate the number of lyman continuum photons , n@xmath31 . for this calculation , we used the extinction corrected br@xmath2 flux , a distance of 19.2 mpc for the antennae , and recombination coefficients from @xcite and @xcite .
we find that n@xmath31=3.1@xmath3210@xmath33 s@xmath16 and 1.9@xmath3410@xmath33 s@xmath16 for [ ws95]86 and [ ws95]80 respectively .
this somewhat lower than the value derived from the thermal radio continuum flux density estimated by @xcite , which results in n@xmath35 s@xmath16 for the sum of [ ws95]86 and [ ws95]80 . given that our apertures are smaller , the difficulty in estimating the thermal contribution to the total radio emission , and that there is some diffuse radio emission , it is not surprising that there is a difference between the two estimates .
it is also significantly less than the value of 1.0@xmath36 photons s@xmath16 derived by @xcite .
both our observed flux and estimated extinction are lower by factors of @xmath37 and both of these differences account for the discrepancy between our respective estimates of the lyman continuum flux .
however , @xcite caution that the extinction value they derived ( a@xmath38 = 1.1mag ) should be considered an upper limit .
comparing our results ( m@xmath38 ( 0 ) and n@xmath31 ) with the models of ( * ? ? ?
* assuming a salpeter imf slope , m@xmath23=100 m@xmath8 , high mass stars have already evolved off the main sequence and are therefore not observed , and m@xmath39=1 m@xmath8 ) for the age determined for each cluster in the previous section , the total mass of each cluster would be 1.6@xmath40 and 3.0@xmath4110@xmath42 m@xmath8 for [ ws95]86 and [ ws95]80 respectively .
a lower mass limit equal to 0.1 m@xmath8 ( using miller - scalo slope for the lower mass end of the imf ) would imply masses a factor of 1.6 higher .
for [ ws95]417&428 and [ ws95]405 , we crudely estimated their masses using only their extinction - corrected k - magnitudes compared to the magnitudes expected from our derived ages .
their masses are lower than the masses of [ ws95]86 and [ ws95]80 , and lie between 0.3 and 1.1 @xmath43 m@xmath44 .
this places each of the very massive [ ws95]86 and [ ws95]80 clusters into the same mass regime as m82a , which produces roughly the same amount of n@xmath31 , and has the same burst age @xcite .
their masses are more than a factor of ten higher than the average mass of a globular cluster @xcite .
even if they are expected to lose 60% of their mass over a hubble time @xcite , they will still represent the top - end of the globular cluster mass function .
the [ ws95]417&428 and [ ws95]405 clusters have masses which are more moderate and comparable to the average globular cluster mass .
this mass range in the observed clusters is not surprising given that those two fields contribute a substantial amount of the bolometric luminosity of the antennae . but
a more accurate mass estimate of a larger fraction of the population of young star clusters is necessary to judge if the young clusters could evolve to form a part of the globular cluster population of an elliptical galaxy .
we performed integral field spectroscopy in the k - band ( 1.9 - 2.4@xmath0 m ) on four star clusters and the nuclei in ngc 4038/4039 in order to derive the starburst properties and their variations with the location in the merger .
our analysis of br@xmath2 , co band - heads , he i @xmath32.058@xmath0m / br@xmath2 , h@xmath4 , and v - k colors indicates that the clusters show a considerable age spread from 4 - 13 myrs .
the two star clusters in the interaction zone have ages near the lower limit of that range , while those in the outer loops lie near the top of that age range .
these clusters sample the overall age gradient visible in the antennae , the details of which we will discuss in a forthcoming paper .
the equivalent widths of the co band - heads indicate that the starbursts in the nuclei are considerably older ( 65 myrs ; with the nucleus of ngc 4038 having evidence for a significant contribution from younger population with an age of @xmath126 myrs northward of its k - band continuum peak ) .
the extinction is highly variable and especially high in some parts of the overlap region ( a@xmath6=0.2 to 4.3 ) . using these age estimates and assuming the parameters of the imf ( salpeter slope , upper and lower mass cutoffs , etc . )
, we find that the clusters have masses ranging between 0.3 to 3 @xmath9 10@xmath10 m@xmath8 , ( larger if the imf is extended below 1 m@xmath8 ) .
our observations that the k - band continuum peak ( with its strong co absorption ) and the highest surface brightness br@xmath2 emission are not coincident in some regions and possible variations in w@xmath22 for clusters in [ ws95]405 and the nucleus of ngc 4038 argues for sequential star formation on small scales .
van der hulst , j. m. , terlouw , j. p. , begeman , k. , zwister , w. , & roelfsema , in astrophysical data analysis software and systems i , ( eds .
d. m. worall , c. biemesderfer , and j. barnes ) , asp conf .
series num .
25 , p. 131
j2000 & 12@xmath46 & 12@xmath47 & 12@xmath48 & 12@xmath49 & 12@xmath50 & 12@xmath51 + dec . & j2000 & -18@xmath5253022 & -18@xmath5253060 & -18@xmath5252103 & -18@xmath5252116 & -18@xmath5253103 & -18@xmath5252043 + apert .
size & & 22@xmath922 & 22@xmath922 & 14@xmath914 & 22@xmath922 & 38@xmath938 & 38@xmath926 + & & & & 14@xmath910 & 20@xmath918 & & 30@xmath922 + w@xmath21 & & 210@xmath2710 & 116@xmath2710 & 2@xmath272 & [email protected] , & [email protected] & [email protected] + & & & & [email protected] & 54@xmath275 & & 11@xmath272 ( 58 ) + w@xmath22 & & 0@xmath53 & 0@xmath53 & 19@xmath274 & [email protected] & [email protected] & [email protected] + & & & & 15@xmath272 & 0@xmath272 & & [email protected] + age & 10@xmath10yr & 3.7@xmath54 & 5.5@xmath55 & 8.5 ... 12.8 & 8.1@xmath56 & 65@xmath2715 & 65@xmath2715 + & & & & each & & & ( 5.9 ) + he i / br@xmath2 & & [email protected] & [email protected] & & & & [email protected] + t@xmath26 & 10@xmath57k & 36@xmath271 & 38@xmath271 & & & & 33@xmath271 + m@xmath38 & mag & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + & & & & each & & & [email protected] + m@xmath6 & mag & [email protected] & [email protected] & [email protected] , & [email protected] & [email protected] & [email protected] , + & & & & [email protected] & & & [email protected] + a@xmath6 & mag & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0@xmath58 , + & & & & & & & [email protected] + f@xmath21 & 10@xmath59 erg & 83@xmath278 & 42@xmath278 & 1.5@xmath271 & 18@xmath273 & 20@xmath271 & 7@xmath272 + & s@xmath16@xmath60 & & & & ( a+b ) & & 21@xmath273 + f@xmath61 & 10@xmath62 erg & 32@xmath273 & [email protected] & [email protected] & 10@xmath271 & 17@xmath271 & 8@xmath271 + & s@xmath16@xmath60 & & & ( a+b ) & ( a+b ) & & 13@xmath271 + mass & 10@xmath10m@xmath44 & 1.6@xmath40 & 3.0@xmath63 & 0.4 - 0.6 & [email protected] + & & & & each & + | integral field spectroscopy in the k - band ( 1.9 - 2.4@xmath0 m ) was performed on four ir - bright star clusters and the two nuclei in ngc 4038/4039 ( `` the antennae '' ) .
two of the clusters are located in the overlap region of the two galaxies , and together comprise @xmath125% of the total
15@xmath0 m and @xmath110% of the total 4.8 ghz emission from this pair of merging galaxies .
the other two clusters , each of them spatially resolved into two components , are located in the northern galaxy , one in the western and one in the eastern loop of blue clusters . comparing our analysis of br@xmath2 , co band - heads , he i @xmath32.058@xmath0
m , h@xmath4 ( from archival hst data ) , and v@xmath5k colors with stellar population synthesis models indicates that the clusters are extincted ( a@xmath6@xmath7 mags ) and young , displaying a significant age spread ( 4 - 13 myrs ) .
the starbursts in the nuclei are much older ( 65 myrs ) , with the nucleus of ngc 4038 displaying a region of recent star formation northward of its k - band peak . using our derived age estimates and assuming the parameters of the imf ( salpeter slope , upper mass cut - off of 100 m@xmath8 , miller - scalo between 1 m@xmath8 and 0.1 m@xmath8 )
, we find that the clusters have masses between 0.5 and 5 @xmath9 10@xmath10 m@xmath8 . |
since its introduction as a model for the periodic appearance of the ice ages @xcite , stochastic resonance has been observed in a large number of physical and biological systems , including lasers , electronic circuits and the sensory system of crayfish ( for reviews of applications , see for instance @xcite ) .
the mechanism of stochastic resonance can be illustrated in a simple model .
consider the overdamped motion of a particle in a double - well potential .
the two potential wells describe two macroscopically different states of the unperturbed system , for instance cold and warm climate .
the particle is subject to two different kinds of perturbation : a deterministic periodic driving force ( such as the periodic variation of insulation caused by the changing eccentricity of the earth s orbit ) , and an additive noise ( modeling the random influence of the weather ) .
each of these two perturbations , taken by itself , does not produce any interesting dynamics ( from the point of view of resonance ) .
indeed , the periodic driving is assumed to have too small an amplitude to allow for any transitions between the potential wells in the absence of noise . on the other hand , without periodic forcing ,
additive noise will cause the particle to jump from one potential well to the other at random times .
the expected time between transitions is given asymptotically , in the small noise limit , by kramers time , which is proportional to the exponential of the barrier height @xmath5 over the noise intensity squared , namely @xmath6 .
when both perturbations are combined , however , and their amplitudes suitably tuned , the particle will flip back and forth between the wells in a close to periodic way .
thus the internal noise can significantly enhance the weak external periodic forcing , by producing large amplitude oscillations of the system , hence the name of resonance .
the choice of the term resonance has been questioned , as it would be more appropriate to refer to _ noise - induced signal - to - noise ratio enhancement _
@xcite . in the regime of a periodic
driving whose amplitude is not a small parameter , one also speaks of _ noise - induced synchronization _ @xcite .
appreciable , though still sub - threshold amplitudes of the periodic driving have the advantage to enable transitions for small noise intensities , without requiring astronomically long driving periods .
while the heuristic mechanism of stochastic resonance is rather well understood , a complete mathematical description is still lacking , though important progress has been made in several limiting cases . depending on the regime
one is interested in , several approaches have been used to describe the phenomenon quantitatively .
the simplest ones use a discretization of either time or space .
when the potential is considered as piecewise constant in time , the generator of the autonomous case can be used to give a complete solution @xcite , showing that resonance occurs when driving period and kramers time are equal .
alternatively , space can be discretized in order to obtain a two - state model , which is described by a markovian jump process @xcite .
the two - state model has also been realised experimentally by an electronic circuit , called the schmitt trigger @xcite . in physical experiments , one has often access to indirect characteristics of the dynamics , such as the power spectrum , which displays a peak at the driving frequency .
the strength of the resonance is quantified through the signal - to - noise ratio ( snr ) , which is proportional to the area under the peak ( this definition obviously leaves some liberty of choice ) .
the snr has been estimated , in the limit of small driving amplitude , by using spectral theory of the fokker
planck equation @xcite , or a rateequation for the probability density @xcite .
the signal - to - noise ratio is found to behave like @xmath7 , which reaches a maximum for @xmath8 .
the probability density of the process , however , only gives part of the picture , and a more detailed understanding of the behaviour of individual paths is desirable .
some interesting progress in this direction is found in @xcite .
the approach applies to a very general class of dynamical systems , in the limit of vanishing noise intensity . when the period of the forcing scales like kramers time , solutions of the stochastic differential equation are shown to converge to periodic functions in the following sense : the @xmath9-distance between the paths and the periodic limiting function converges to zero in probability as the noise intensity goes to zero .
due to its generality , however , this approach does not give any information on the rate of convergence of typical paths to the periodic function , nor does it estimate the probability of atypical paths . also , since the period of the forcing must scale like kramers time , the assumed small noise intensity goes hand in hand with exponentially long waiting times between interwell transitions . in the present work ,
we provide a more detailed description of the individual paths behaviour , for small but finite noise intensities and driving frequencies .
we consider two classes of one - dimensional double - well potentials , symmetric and asymmetric ones .
the height of the potential barrier is assumed to become small periodically , which allows us to consider situations where the period need not be exponentially large in @xmath10 for transitions between the wells to be likely . in the case of an asymmetric potential , we are interested , in particular , in determining the optimal noise intensity as a function of the driving frequency and the minimal barrier height , guaranteeing a close - to - periodic oscillation between both wells
. we will estimate both the deviation ( in space and time ) of typical paths from the limiting periodic function , and the asymptotics of the probability of exceptional paths .
the case of a symmetric potential shows an additional feature .
for the right choice of the noise intensity , transitions become likely once per period , at which time the new well is chosen at random .
we will again estimate the deviation from a suitable reference process and the asymptotics of the probability of exceptional paths .
the systems are described by stochastic differential equations ( sdes ) of the form @xmath11 where @xmath12 is a brownian motion .
the potential @xmath13 is @xmath0-periodic in @xmath14 , and admits two minima for every value of @xmath14 .
the frequency @xmath1 , the minimal barrier height between the wells and the noise intensity @xmath2 are considered as ( moderately ) small parameters , the relation between which will determine the transition probability . the first class of potentials we consider is symmetric in @xmath15 .
a typical representative of this class is the potential @xmath16 here @xmath17 is a parameter controlling the minimal barrier height . we introduce the slow time @xmath18 for convenience .
the potential has two wells , located at @xmath19 , separated by a barrier of height @xmath20 .
the distance between the wells and the barrier height become small simultaneously , at integer values of @xmath21 .
our results for symmetric potentials can be summarized as follows : in the deterministic case @xmath22 , we describe the dependence of solutions on @xmath21 , @xmath23 and @xmath1 ( theorem [ thm_sdet ] ) .
solutions starting at @xmath24 are attracted by the potential well at @xmath25 , which they track with a small lag . if @xmath26 , this lag is at most of the order @xmath27 ; if @xmath28 , it is at most of the order @xmath29 , but solutions never approach the saddle closer than a distance of order @xmath29 ( even if @xmath30 ) . when noise is present , but @xmath2 is small compared to the maximum of @xmath23 and @xmath31 , the paths are likely to track the solution of the corresponding deterministic differential equation at a distance of order @xmath32 ( theorem [ thm_snearxdet ] ) . the probability to reach
the saddle during one time period is exponentially small in @xmath33 .
if @xmath2 is larger than both @xmath23 and @xmath31 , transitions between potential wells become likely , but are concentrated on the time interval @xmath34 $ ] ( repeated periodically ) . during this time interval , the paths may jump back and forth frequently between both potential wells , and they have a typical spreading of the order @xmath35 . after time @xmath36 ,
the paths are likely to choose one of the wells and stay there till the next period ( theorem [ thm_sescape ] ) .
the probability to choose either potential well is exponentially close to @xmath3 , with an exponent of order @xmath37 , which is independent of @xmath23 ( theorem [ thm_strans ] ) .
this picture remains true when @xmath2 is larger than both @xmath38 and @xmath29 , but note that the spreading of paths during the transition may become very large . thus increasing noise levels
will gradually blur the periodic signal .
these results show a rather sharp transition to take place at @xmath39 , from a regime where the paths are unlikely to switch from one potential well to the other one , to a regime where they do switch with a probability exponentially close to @xmath3 ( ) . the second class of potentials we consider is asymmetric , a typical representative being @xmath40 this is a double - well potential if and only if @xmath41 .
we thus choose @xmath42 of the form @xmath43 near @xmath44 , the right - hand potential well approaches the saddle at a distance of order @xmath38 , and the barrier height is of order .
a similar encounter between the left - hand potential well and the saddle occurs at @xmath45 .
our results for asymmetric potentials can be summarized as follows : in the deterministic case , solutions track the potential wells at a distance at most of order @xmath46 . if @xmath47 , they never approach the saddle closer than a distance of order @xmath48 ( theorem [ thm_adet ] ) . when @xmath2 is small compared to the maximum of and @xmath49 , paths are likely to track the deterministic solutions at a distance of order @xmath50 ( theorem [ thm_anearxdet ] ) . the probability to overcome the barrier is exponentially small in @xmath51 . for larger
@xmath2 , transitions become probable during the time interval @xmath52 $ ] . due to the asymmetry , the probability to jump from the less deep potential well to the deeper one
is exponentially close to one , with an exponent of order @xmath53 , while paths are unlikely to come back ( theorem [ thm_atrans ] ) .
this picture remains true when @xmath2 is larger than both and , but the spreading of paths during the transition may become very large .
again , we find a rather sharp transition to take place , this time at @xmath54 .
in contrast to the symmetric case , for large @xmath2 the paths are likely to jump from one potential well to the other at every half - period ( ) . in both the symmetric and the asymmetric case , we thus obtain a high switching probability between the potential wells even for small noise intensities , provided minimal barrier height and driving frequency are sufficiently small .
they only need , however , to be smaller than a power of @xmath2 : @xmath55 and @xmath56 in the symmetric case , and @xmath57 , @xmath58 in the asymmetric case are sufficient conditions for switching dynamics .
our results require a precise understanding of dynamical effects , and the subtle interplay between the probability to reach the potential barrier , the time needed for such excursions , and the total number of excursions with a chance of success . in this respect ,
they provide a substantial progress compared to the quasistaticapproach , which considers potentials that are piecewise constant in time . note that some of our results may come as a surprise . in particular , neither the width ( in time ) of the transition zone nor the asymptotics of the transition probability depend on the minimal barrier height @xmath23 .
in fact , the picture is independent of @xmath23 as soon as @xmath23 is smaller than @xmath31 ( in the symmetric case ) or @xmath1 ( in the asymmetric case ) , even for @xmath30 .
this is due to the fact that when @xmath23 is small , the time during which the potential barrier is low is too short to contribute significantly to the transition probability . the remainder of this paper is organized as follows .
the results are formulated in detail in section [ sec_res ] , subsection [ ssec_rsym ] being devoted to symmetric potentials , and subsection [ ssec_rasym ] to asymmetric potentials . section [ sec_sym ] contains the proofs for the symmetric case , while section [ sec_asym ] contains the proofs for the asymmetric case .
the fascinating problems related to stochastic resonance were brought to our attention by anton bovier , whom we would like to thank for stimulating discussions .
thanks the wias for kind hospitality .
we consider non - autonomous sdes of the form . introducing the slow time @xmath18 allows to study the system on a time interval of order one .
when substituting @xmath21 for @xmath59 , brownian motion is rescaled and we obtain an sde @xmath60 where @xmath61 is the force , derived from the potential @xmath62 , and @xmath63 is a standard wiener process on some probability space @xmath64 .
initial conditions @xmath65 are always assumed to be square - integrable with respect to @xmath66 and independent of @xmath63 . without further mentioning
we always assume that @xmath61 satisfies the usual ( local ) lipschitz and bounded - growth conditions which guarantee existence and pathwise uniqueness of a strong solution @xmath67 of . under these conditions
, there exists a continuous version of @xmath67 .
therefore we may assume that the paths @xmath68 are continuous for @xmath66-almost all @xmath69 .
we introduce the notation @xmath70 for the law of the process @xmath71 , starting in @xmath65 at time @xmath72 , and use @xmath73 to denote expectations with respect to @xmath70 .
note that the stochastic process @xmath71 is an inhomogeneous markov process .
we are interested in first exit times of @xmath74 from space time sets .
let @xmath75 $ ] be borel - measurable .
assuming that @xmath76 contains @xmath77 , we define the first exit time of @xmath78 from @xmath76 by @xmath79}{(x_t , t)\not\in{{\mathcal a}}},\ ] ] and agree to set @xmath80 for those @xmath69 which satisfy @xmath81 for all @xmath82 $ ] . for convenience ,
we shall call @xmath83 the _ first exit time of @xmath74 from @xmath76_. typically , we will consider sets of the form @xmath84}{g_1(t)<x
< g_2(t)}$ ] with continuous functions @xmath85 .
note that in this case , @xmath83 is a stopping time , the first exit time @xmath83 is still a stopping time with respect to the canonical filtration , completed by the null sets . ] with respect to the canonical filtration of @xmath64 generated by @xmath71 . before turning to the precise statements of our results ,
let us introduce some notations .
we shall use @xmath86 for @xmath87 to denote the smallest integer which is greater than or equal to @xmath88 , and @xmath89 and @xmath90 to denote the maximum or minimum , respectively , of two real numbers @xmath88 and @xmath91 .
if @xmath92 and @xmath93 are defined for small @xmath1 and for @xmath21 in a given interval @xmath94 , we write @xmath95 if there exist strictly positive constants @xmath96 such that @xmath97 for all @xmath98 and all sufficiently small @xmath1 .
the constants @xmath96 are understood to be independent of @xmath21 and @xmath1 ( and hence also independent of quantities like @xmath2 and @xmath23 which we consider as functions of @xmath1 ) . by @xmath99
we indicate that there exist @xmath100 and @xmath101 such that @xmath102 for all @xmath103 $ ] , where @xmath104 and @xmath105 of course do not depend on @xmath1 or on the other small parameters @xmath23 and @xmath2 .
similarly , @xmath106 is to be understood as @xmath107
. finally , let us point out that most estimates hold for small enough @xmath1 only , and often only for @xmath66-almost all @xmath69 .
we will stress these facts only where confusion might arise .
we consider in this subsection the sde in the case of @xmath61 being periodic in @xmath21 , odd in @xmath15 , and admitting two stable equilibrium branches , with a barrier between the branches becoming small once during every time period .
a typical example of such a function is @xmath108 we will consider a more general class of functions @xmath109 , which we assume to satisfy the following hypotheses : _ smoothness : _
@xmath110 , where @xmath111\times{{\mathbb{r}}\,}$ ] and @xmath112 is a constant ; _ periodicity : _ @xmath113 for all @xmath114 ; _ symmetry : _ @xmath115 for all @xmath114 ; _ equilibrium branches : _ there exists a continuous function @xmath116 $ ] with the property that @xmath117 in @xmath118 if and only if @xmath119 or @xmath120 ; _ stability : _ the origin is unstable and the equilibrium branches @xmath121 are stable , that is , for all @xmath122 , @xmath123 _ behaviour near @xmath44 : _ we want the three equilibrium branches to come close at integer times . given the symmetry of @xmath61 , the natural assumption is that we have an avoided pitchfork bifurcation , that is , @xmath124 where @xmath125 and @xmath126 are fixed ( of order one ) , while @xmath127 is a positive small parameter .
is is easy to show that @xmath128 behaves like @xmath25 for small @xmath21 , and admits a quadratic minimum at a time @xmath129 .
moreover , @xmath130 near @xmath44 .
we can choose a constant @xmath131 such that the derivatives of @xmath132 and @xmath128 vanish only once in the interval @xmath133 $ ] .
we finally require that @xmath128 , @xmath132 and @xmath134 are bounded away from zero outside this interval .
we can summarize these properties as @xmath135 we start by considering the deterministic equation @xmath136 without loss of generality , we may assume that @xmath137 starts at time @xmath138 in some @xmath139 .
theorem @xcite , applied on the interval @xmath140 $ ] , implies that @xmath137 converges exponentially fast to a of order @xmath1 of @xmath128 .
we may thus assume that @xmath141 .
in fact , since @xmath142 is decreasing at time @xmath143 , we may even assume that @xmath144 .
the motion of @xmath137 in the interval @xmath133 $ ] is described in the following theorem .
[ thm_sdet ] the solution @xmath137 and the curve @xmath128 cross once and only once during the time interval @xmath133 $ ] .
this crossing occurs at a time @xmath145 satisfying @xmath146 . there exists a constant @xmath147 such that @xmath148 and thus @xmath149 in these time intervals . for @xmath150
, @xmath151 finally , the linearization of @xmath61 at @xmath137 satisfies @xmath152 we give the proof in subsection [ ssec_symdet ] .
the relation may be surprising , since it means that no matter how small we make @xmath23 , @xmath137 never approaches the saddle at @xmath119 closer than a distance of order @xmath29 .
this fact can be intuitively understood as follows .
even if @xmath30 and near @xmath44 , we have @xmath153 since for @xmath154 , @xmath74 can not approach the origin significantly during any time interval of order @xmath29 .
after such a time , however , the repulsion of the saddle will make itself felt again , preventing the solution from further approaching the origin .
in other words , the time interval during which @xmath132 is smaller than @xmath31 is too short to allow the deterministic solution to come close to the saddle .
we return now to the sde with @xmath155 .
assume that we start at some deterministic @xmath156 .
theorem 2.3 in @xcite shows that the paths are likely to track the deterministic solution @xmath137 with the same initial condition at a distance of order @xmath157 for any @xmath100 ( with probability @xmath158 ) , as long as the equilibrium branches are well separated , that is , at least for @xmath159 .
a transition between the potential wells is thus unlikely if @xmath160 , and interesting phenomena can only be expected between the times @xmath143 and @xmath161 . upon completion of one time period ,
i.e. , at time @xmath161 , the markov property allows to repeat the above argument .
hence there is no limitation in considering the sde on the time interval @xmath133 $ ] , with a fixed initial condition @xmath162 satisfying @xmath163 .
we will denote by @xmath137 and @xmath74 , respectively , the solutions of and with the same initial condition @xmath162 .
let us start by describing the dynamics in a of @xmath137 .
the main idea is that for @xmath2 sufficiently small , the typical spreading of paths around @xmath137 should be related to the variance @xmath164 of the solution of , linearized around @xmath137 .
this variance is given by @xmath165 the variance is equal to zero at time @xmath143 , but behaves asymptotically like @xmath166 .
in fact , if we define the function @xmath167 then @xmath164 differs from @xmath168 by a term that becomes negligible as soon as @xmath169 is larger than a constant times @xmath170 .
@xmath171 has the advantage to be bounded away from zero for all @xmath21 , which avoids certain technical problems in the proofs .
we shall show that @xmath172 we introduce the set @xmath173 and denote by @xmath174 the first exit time of @xmath74 from @xmath175 .
[ thm_snearxdet ] there exists a constant @xmath176 , depending only on @xmath61 , such that if @xmath177 and @xmath178 , then @xmath179 if @xmath180 and @xmath181 , then @xmath182 in both cases , @xmath183 we give the proof in subsection [ ssec_symnear ] .
this result has several consequences .
observe first that the exponential factors in and are very small as soon as @xmath184 is significantly larger than @xmath2 .
the prefactor @xmath185 ( which , unlike the exponent , we do not believe to be optimal ) leads to subexponential corrections , which are negligible as soon as @xmath186 .
it mainly accounts for the fact that the probability for a path to leave @xmath175 increases slowly with time .
the theorem shows that the typical spreading of paths around @xmath137 is of order @xmath187 if @xmath188 , we may choose @xmath189 for all times , and thus the probability of leaving a of @xmath137 , let alone approach the other stable branch , is exponentially small ( in @xmath190 ) . on the other hand ,
if @xmath2 is not so small , can still be applied to show that a transition is unlikely to occur before a time of order @xmath191 .
figure [ fig3 ] illustrates this phenomenon by showing typical paths for two different noise intensities .
let us now assume that @xmath2 is sufficiently large to allow for a transition , and examine the transition regime in more detail .
we will proceed in two steps .
first we will estimate the probability of _ not _ reaching the saddle at @xmath119 during a time interval @xmath192 $ ] .
the symmetry of @xmath61 implies that for any @xmath193 and @xmath194 , @xmath195 } \\ & \geqs \frac12 - \frac12\mskip1.5mu\bigprobin{t_0,x_0}{x_s>0\;\forall s\in[t_0,t_1]}.\end{aligned}\ ] ] in the second step , we will show , independently , that paths are likely to leave a of @xmath119 after time @xmath196 . thus if the probability of not reaching @xmath119 is small , the probability of making a transition from the positive well to the negative one will be close to @xmath3 ( it can never exceed @xmath3 because of the symmetry ) . this does not exclude , of course , that paths frequently switch back and forth between the two potential wells during the time interval @xmath197 $ ] .
but it shows that can indeed be interpreted as a lower bound on the transition probability .
let @xmath100 be a constant such that @xmath198 our hypotheses on @xmath61 imply that such a @xmath104 of order one always exists .
in some special cases , for instance if @xmath199 , @xmath104 may be chosen arbitrarily large .
[ thm_strans ] let @xmath200 be a constant and assume @xmath201
. choose times @xmath202 with @xmath203 $ ] , and let @xmath204 be such that @xmath205 for all @xmath206 $ ] . then ,
if @xmath207 is sufficiently small and @xmath208 $ ] , @xmath209 } \\
\leqs \frac52 \bigpar{\frac{\abs{\balpha(t_1,t_0)}}\eps + 1 } \operatorname{e}^{-\bar\kappa h^2/\sigma^2 } + \mskip 1.5mu 2 \exp\bigset{-\bar\kappa \frac1{\log(h/\sigma ) } \frac{\alpha(t_1,-c_1\sqrt\sigma)}\eps } , \qquad\end{gathered}\ ] ] where @xmath210 is a positive constant , and @xmath211 .
the proof is given in subsection [ ssec_symtrans ] .
the first term in is an upper bound on the probability that @xmath74 escapes upward . indeed
, our hypotheses on @xmath61 do not exclude that other stable equilibria exist for sufficiently large @xmath15 , which might trap escaping trajectories .
the second term bounds the probability of @xmath212 remaining between @xmath213 and @xmath214 for @xmath215 .
this estimate lies at the core of our argument , and can be understood as follows .
assume @xmath212 starts near @xmath216 .
it will perform a certain number of excursions to attempt reaching the saddle at @xmath119 .
each excursion requires a typical time of order @xmath217 , such that @xmath218 ( that is , @xmath219 , in the sense that the probability of reaching @xmath213 before time @xmath220 is small . after an unsuccessful excursion
, @xmath212 may exceed @xmath216 , but will return typically after another time of order @xmath217 .
thus the total number of trials during the time interval @xmath221 $ ] is of order @xmath222 . under the hypotheses of the theorem , the probability of not reaching the saddle during one excursion is of order one , and
thus the total number of trials determines the exponent in . before discussing the choice of the parameters giving an optimal bound in theorem [ thm_strans ] ,
let us first state the announced second step , namely the claim that the paths are likely to escape from the saddle after @xmath223 .
for @xmath224 , let us introduce the set @xmath225\times[c_1\sqrt\sigma , t ] } { \frac{f(x , t)}x >
\kappa a(t)}.\ ] ] the upper boundary of @xmath226 is a function @xmath227 .
let @xmath228 denote the first exit time of @xmath74 from @xmath226 .
[ thm_sescape ] let @xmath229 and assume @xmath230 .
then there exist constants @xmath231 and @xmath232 such that @xmath233 for all @xmath234 with @xmath235 .
the proof is adapted from the proof of the similar theorem 2.9 in @xcite .
compared to that result , we have sacrificed a factor @xmath236 in the exponent , in order to get a weaker condition on @xmath2 .
we discuss the changes in the proof in subsection [ ssec_symesc ] . for the moment , let us consider @xmath237 .
we want to choose a @xmath21 such that @xmath238 .
since @xmath239 is larger than a constant times @xmath240 , it suffices to choose a @xmath21 of order @xmath241 for to become small .
hence , after waiting for a time of that order , we find @xmath242 which shows that most trajectories will have left @xmath226 by time @xmath21 , see .
it remains to show that the paths are likely to approach either @xmath128 or @xmath243 after leaving @xmath226 .
let us first consider the solution @xmath244 of the deterministic differential equation with initial time @xmath245 and initial condition @xmath246 for some small constant @xmath247 .
here we need to choose @xmath248 small in order to arrange for @xmath249 which allows us to proceed as in our investigation of the motion for @xmath250 , cf .
subsections [ ssec_symdet ] and [ ssec_symnear ] . under these assumptions , @xmath244 approaches a neighbourhood of @xmath251 exponentially fast and then tracks the equilibrium branch at distance @xmath252 .
as before , one can show that the path @xmath74 of the solution of the sde with the same initial condition is likely to remain in a strip around @xmath244 of width scaling with @xmath253 .
so if a path @xmath74 leaves @xmath226 at time @xmath228 , then this path is likely to approach @xmath128 , if @xmath254 is positive , and @xmath243 , otherwise .
note that @xmath255 has to be smaller than @xmath256 , which restricts the possible values for @xmath257 .
therefore , we choose @xmath257 small enough to guarantee that @xmath258 for all @xmath259 .
finally note that paths which are not in @xmath226 at time @xmath260 but are not further away from @xmath121 than @xmath261 at some time @xmath21 will also approach the corresponding equilibrium branch .
let us now discuss the choice of the parameters in giving an optimal bound . for @xmath262 ,
theorem [ thm_snearxdet ] shows that up to time @xmath21 slightly less than @xmath263 , the paths are concentrated around @xmath137 .
therefore , here we should choose an initial time @xmath72 slightly before @xmath263 . in order for the second term in to be small ,
we want to choose @xmath264 as large as possible .
note , however , that @xmath265 has to be smaller than @xmath266 for all @xmath14 .
since the order of @xmath267 is increasing for @xmath268 , it turns out to be more advantageous to choose @xmath269 negative and of order @xmath191 , say @xmath270 . in this case , @xmath271 is larger than a constant times @xmath272 ( independently of @xmath23 ) , and this does not improve significantly for larger admissible @xmath269 .
at the same time , this choice allows us to take @xmath273 .
we find @xmath274 } & \leqs \bigprobin{t_0,x_0}{x_s>0\;\forall s\in[t_0,-\tfrac12 c_1\sqrt\sigma\ , ] } \\
\nonumber & \leqs \biggpar{\frac{\abs{t_0}}{\sqrt\sigma}}^3 \frac{\sigma^{3/2}}{\eps } \operatorname{e}^{-\order{\delta^2/\sigma } } + \exp\biggset{-\frac{\text{\it const}}{\log(\delta^2/\sigma ) } \frac{\sigma^{3/2}}{\eps}}.\end{aligned}\ ] ] consider first the generic case @xmath275 . the second term in becomes small as soon as @xmath276 holds in addition to the general condition @xmath277 .
the first term is small as long as @xmath278 .
we thus obtain the following regimes : for @xmath279 , the transition probability is exponentially small in @xmath190 ; for @xmath280 with @xmath281 , the probability of a transition between the wells is exponentially close to @xmath3 , with an exponent given essentially ( up to logarithmic corrections ) by @xmath282 for @xmath283 , the paths become so poorly localized that it is no longer meaningful to speak of a transition probability .
shows that the transition probability becomes optimal for @xmath284 . for larger values of the noise intensity
, the possibility of paths escaping upward becomes sufficiently important to decrease the transition probability .
however , if the function @xmath61 is such that @xmath104 can be chosen arbitrarily large , the second term in can be removed without changing the first one ( up to logarithmic corrections ) by taking @xmath285 , for instance . in that case , transitions between the wells become the more likely the larger the ratio @xmath286 is . one should note that a typical path will reach maximal values of the order @xmath287 .
thus , due to the flatness of the potential near @xmath44 , if @xmath2 is larger than @xmath288 , the spreading of the paths during the transition interval is larger than the maximal distance between the wells away from the transition .
in general we can not exclude that paths escape to other attractors , if the potential has more than two wells .
it may be surprising that the order of the transition probability is independent of @xmath23 as soon as @xmath289 .
intuitively , one would rather expect this probability to depend on the ratio @xmath290 , because of kramers law .
the fact that this is not the case illustrates the necessity of a good understanding of dynamical effects ( as opposed to a quasistatic picture ) .
although the potential barrier is smallest between the times @xmath291 and @xmath38 , the paths have more opportunities to reach the saddle during larger time intervals .
the optimal time interval turns out to have a length of the order @xmath196 , which corresponds to the regime where diffusive behaviour prevails over the influence of the drift .
we consider in this subsection the sde in the case of @xmath61 being periodic in @xmath21 and admitting two stable equilibrium branches , but without the symmetry assumption .
instead , we want each of the potential wells to approach the saddle once in every time period , but at different times for the left - hand and the right - hand potential well .
a typical example of such a function is @xmath292 here @xmath293 is defined by the fact that @xmath61 has two stable equilibria if and only if @xmath294 .
observe that @xmath295 vanishes at @xmath296 , and @xmath297 here the function @xmath298 plays the role that @xmath132 played in the symmetric case , and near @xmath44 the right - hand potential well and the saddle behave like @xmath299 , while the left - hand potential well is isolated .
near @xmath45 , a similar close encounter takes place between the saddle and the left - hand potential well .
we will consider a more general class of functions @xmath109 , which we assume to satisfy the following hypotheses : _ smoothness : _
@xmath300 , where @xmath111\times{{\mathbb{r}}\,}$ ] and @xmath112 is a constant ; _ periodicity : _
@xmath113 for all @xmath114 ; _ equilibrium branches : _ there exist continuous functions @xmath301 from @xmath302 to @xmath303 $ ] with the property that @xmath117 in @xmath118 if and only if @xmath304 or @xmath305 ; the zeroes of @xmath61 should be isolated in the following sense : for every @xmath100 , there should exist a constant @xmath306 such that , if @xmath307 and @xmath308 , then @xmath309 . depends on a small parameter @xmath23 , we want to avoid that @xmath310 approaches zero elsewhere but near the three equilibrium branches , even when @xmath23 becomes small . ] _ stability : _ the equilibrium branches @xmath311 are stable and the equilibrium branch @xmath312 is unstable , that is , for all @xmath122 , @xmath313 _ behaviour near @xmath44 : _ we want @xmath314 and @xmath312 to come close at integer times . here
the natural assumption is that we have an avoided saddle
node bifurcation , that is , there exists an @xmath315 such that @xmath316 where @xmath125 and @xmath317 are fixed ( of order one ) , while @xmath318 is a positive small parameter .
these assumptions imply that @xmath319 reaches a local minimum at a time @xmath320 , and @xmath321 reaches a local maximum at a possibly different time @xmath322 .
we can assume that for a sufficiently small constant @xmath323 , the three equilibrium branches and the linearization of @xmath61 around them satisfy @xmath324 _ behaviour near @xmath325 : _ we want @xmath326 and @xmath312 to come close at some time @xmath327 .
this is achieved by assuming that similar relations as , but with opposite signs , hold at a point @xmath328 .
_ behaviour between the close encounters : _ to exclude the possibility of other almost - bifurcations , we require that @xmath329 and @xmath330 , as well as the derivatives , are bounded away from zero for @xmath331 and @xmath332 . note that a sufficient assumption for the requirements on the behaviour near @xmath328 to hold is that @xmath333 for all @xmath334 .
we start by considering the deterministic equation @xmath335 as in the symmetric case , it is sufficient to consider the dynamics in the time interval @xmath133 $ ] , with an initial condition satisfying @xmath336 .
the situation in the time interval @xmath337 $ ] can be described in exactly the same way .
[ thm_adet ] the solution @xmath137 and the curve @xmath338 cross once and only once during the time interval @xmath133 $ ] .
this crossing occurs at a time @xmath145 satisfying @xmath339 .
there exists a constant @xmath147 such that @xmath340 and thus @xmath341 in these time intervals . for @xmath342 , @xmath343
the linearization of @xmath61 at @xmath137 satisfies @xmath344 moreover , admits a particular solution @xmath244 tracking the _ unstable _ equilibrium branch @xmath345 .
it satisfies analogous relations , namely , @xmath244 and @xmath345 cross once at a time @xmath346 satisfying @xmath347 , and , and hold for @xmath348 and @xmath321 , but with opposite signs . the proof is similar to the proof of theorem [ thm_sdet ] , and we comment on a few minor differences in subsection [ ssec_asymdet ] .
note that implies that @xmath137 never approaches the saddle at @xmath321 closer than a distance of order @xmath48 .
we return now to the sde with @xmath155 .
we will denote by @xmath137 and @xmath74 , respectively , the solutions of and with the same initial condition @xmath162 satisfying @xmath349 .
we introduce again the function @xmath350 which behaves , in this case , like @xmath351 we define once more the set @xmath352 and denote by @xmath174 the first exit time of @xmath74 from @xmath175 .
[ thm_anearxdet ] there exists a constant @xmath176 , depending only on @xmath61 , such that if @xmath353 and @xmath354 , then @xmath355 if @xmath356 and @xmath357 , then @xmath358 in both cases , @xmath359 this result is proved in exactly the same way as theorem [ thm_snearxdet ] .
it has similar consequences , only with different values of the exponents .
the typical spreading of paths around @xmath137 is of order @xmath360 if @xmath361 , the probability of leaving a of @xmath137 , or making a transition to the other stable equilibrium branch , is exponentially small ( in @xmath362 ) . on the other hand , if @xmath2 is not so small , can still be applied to show that a transition is unlikely to occur before a time of order @xmath363 .
let us now assume that @xmath2 is sufficiently large for a transition to take place , i.e. that @xmath364 .
we want to give an upper bound on the probability _ not _ to make a transition .
let us introduce levels @xmath365 such that @xmath366 here @xmath367 and @xmath368 are always of order @xmath4 ( in fact , we must have @xmath369 for all @xmath370 $ ] ) , and we think of @xmath367 as being in the basin of attraction of @xmath326 .
our hypotheses imply that a @xmath371 of order one satisfying always exists , but @xmath371 may be chosen arbitrarily large in particular cases such as @xmath372 .
the non - transition probability can be estimated by distinguishing three cases : either @xmath74 , starting in @xmath373 at some @xmath374 , never reaches @xmath368 .
the probability of this event can be shown to be small in a similar way as in theorem [ thm_strans ] , the main difference being that due to the asymmetry , we can do better than estimating the probability not to reach the saddle . if @xmath74 reaches @xmath368 , one can estimate in a very simple way the probability not to reach @xmath367 as well , using the fact that the drift term is bounded away from zero .
if @xmath74 reaches @xmath367 , theorem 2.3 in @xcite shows that @xmath74 is likely to reach a small of @xmath326 as well .
[ thm_atrans ] let @xmath207 and @xmath375 be positive constants and assume that @xmath376 .
choose times @xmath377 with @xmath378 $ ] and @xmath379 .
let @xmath204 be such that @xmath380 for all @xmath206 $ ] .
then , for sufficiently small @xmath207 , sufficiently large @xmath375 and all @xmath381 $ ] , @xmath382 } \leqs { } & \frac52 \bigpar{\frac{\abs{\balpha(t_1,t_0)}}\eps + 1 } \operatorname{e}^{-\kappa h^2/\sigma^2 } \\ \nonumber & { } + \frac32 \exp\bigset{-\kappa \frac1{\log(h/\sigma)\vee\abs{\log\sigma } } \frac{\widehat\alpha(t_1,-c_1\sigma^{2/3})}\eps } \\ & { } + \operatorname{e}^{-\kappa/\sigma^2 } , \label{ares19}\end{aligned}\ ] ] where @xmath257 is a positive constant , and @xmath383 . the proof is given in subsection [ ssec_asymtrans ] .
the three terms on the right - hand side of bound , respectively , the probability that @xmath74 escapes through the upper boundary @xmath214 , the probability that @xmath74 reaches neither the upper boundary nor @xmath368 , and the probability that @xmath74 does not reach @xmath367 when starting on @xmath368 ( ) . the crucial term is the second one . let us now discuss the optimal choice of parameters .
if we choose @xmath384 , we can take @xmath385 , where @xmath386 , and we get the estimate @xmath387 } \\ \leqs \frac{t_0 ^ 2}\eps \operatorname{e}^{-\order{\tilde\delta_2 ^ 2/\sigma^{4/3 } } } + \exp\biggset{-\frac{\text{\it const}}{\log(\tilde\delta_2 ^ 2/\sigma^{4/3 } ) \vee\abs{\log\sigma } } \frac{\sigma^{4/3}}{\eps } } + \operatorname{e}^{-\kappa/\sigma^2}. \qquad\end{gathered}\ ] ] as in the symmetric case , when @xmath388 , we obtain the following regimes : for @xmath389 , the transition probability is exponentially small in @xmath390 ; for @xmath391 , the transition probability is exponentially close to @xmath4 , with an exponent given essentially ( up to logarithmic corrections ) by @xmath392 for @xmath283 , the paths become so poorly localized that it is no longer meaningful to speak of a transition probability .
the transition probability becomes optimal for @xmath393 .
note , once again , that the exponent is independent of @xmath23 .
if the function @xmath61 is such that @xmath371 can be chosen arbitrarily large , the second term in can be removed without changing the first one ( up to logarithmic corrections ) by taking @xmath394 for instance .
if @xmath395 , the paths may become extremely delocalised in the transition zone , and could escape to other attractors .
we consider in this section the nonlinear sde @xmath396 where @xmath61 satisfies the hypotheses given at the beginning of subsection [ ssec_rsym ] . by rescaling @xmath15
, we can arrange for @xmath397 , so that taylor s formula allows us to write @xmath398 where @xmath399 satisfy @xmath400 with continuously differentiable functions @xmath401 satisfying @xmath402 .
the implicit function theorem shows the existence , for small @xmath21 , of an equilibrium curve @xmath403 for small @xmath21 , the curve @xmath128 behaves like @xmath404 , and it admits a quadratic minimum at some time @xmath129 .
thus we can choose a constant @xmath131 such that @xmath405 in this subsection we consider the deterministic equation @xmath406 as already mentioned , tihonov s theorem allows us to restrict the analysis to the time interval @xmath133 $ ] , and to assume that @xmath407 .
during the time interval @xmath133 $ ] , the process @xmath74 crosses the equilibrium branch @xmath128 once and only once , the time @xmath408 of the crossing satisfying @xmath409 .
this fact is due to the property that @xmath74 is strictly decreasing when lying above @xmath128 , and strictly increasing when lying below .
let @xmath410 be the times of the successive crossings of @xmath74 and @xmath128 in @xmath133 $ ] .
then @xmath74 is decreasing between @xmath143 and @xmath411 ( since @xmath412 ) , increasing for @xmath413 , and so on .
thus @xmath128 must be increasing for @xmath21 slightly larger than @xmath414 , and decreasing for @xmath21 slightly larger than @xmath415 .
since , by assumption , @xmath128 is decreasing on @xmath416 and increasing on @xmath417 $ ] , this implies that @xmath418 and @xmath419 .
therefore , there is at most one crossing .
we shall see below that @xmath74 and @xmath128 actually cross and we will also determine the order of that time @xmath145 .
we consider now the difference @xmath420 .
it satisfies the equation @xmath421 where taylor s formula , and yield the relations @xmath422 with @xmath423 .
we start by giving a technical result that we will need several times .
[ lem_sdet1 ] let @xmath424 be a continuous function satisfying @xmath425 for @xmath426 , where @xmath427 .
let @xmath428 , and define @xmath429 .
then @xmath430 to prove the lemma , we take advantage of the fact that the expression on the left - hand side of is the solution of an ordinary differential equation . by the semi - group property , we may consider separately the following regimes : for @xmath431 , we distinguish the cases @xmath432 $ ] , @xmath433 $ ] , @xmath434 $ ] , @xmath435 $ ] , @xmath436 $ ] , @xmath437 $ ] and for @xmath438 , we deal separately with @xmath432 $ ] , @xmath439 $ ] , @xmath440 $ ] , @xmath441 $ ] , @xmath442 $ ] , @xmath443 $ ] .
on each of these time intervals the claimed behaviour follows easily by elementary calculus , see also the similar result ( * ? ? ?
* lemma 4.2 ) .
[ prop_sdet1 ] there exists a constant @xmath147 such that the solution of with initial condition @xmath444 satisfies @xmath445 let @xmath446 and @xmath447 be constants to be chosen later , and denote by @xmath448 the first exit time of @xmath449 from the strip @xmath450 .
set @xmath451 .
then , for @xmath452 , we get from and that @xmath453 for some constants @xmath454 . the relations and yield the existence of constants @xmath455 such that @xmath456 and @xmath457 for @xmath458 $ ] .
from and we obtain @xmath459 for any given @xmath207 , we can choose @xmath460 large enough for the term in brackets to be larger than @xmath3 .
then , by lemma [ lem_sdet1 ] , there exists a constant @xmath461 such that @xmath462 for all @xmath463 $ ] .
therefore , if @xmath464 , then @xmath465 follows .
the lower bound can be obtained in exactly the same way . for the remainder of this subsection , let @xmath466 with @xmath460 chosen according to the preceding proposition .
note that this proposition implies that @xmath467 for @xmath468 and , in particular , that @xmath469 .
we now consider the dynamics for @xmath470 , starting with the case of @xmath23 not too small , i.e. , the case of @xmath471 .
[ prop_sdet2 ] there exists a constant @xmath472 , depending only on @xmath61 and @xmath473 , such that , when @xmath474 , then @xmath475 for all @xmath470 .
again , we will only show how to obtain an upper bound , since the corresponding lower bound can be established in exactly the same way .
first we fix a constant @xmath476 .
we denote by @xmath448 the first exit time of @xmath449 from the strip @xmath477 . for @xmath478
, we have @xmath479 with constants @xmath454 . choosing @xmath480 large enough , we get @xmath481 which implies @xmath482 by partial integration . here
@xmath483 satisfies @xmath484 .
we want to estimate the contribution of the middle term on the right - hand side of .
assume first that @xmath485 and consider @xmath486 . by convexity , @xmath487 now , @xmath488 since @xmath489 as @xmath490 , we also have @xmath491 provided @xmath480 is large enough .
this shows the existence of constants @xmath492 and @xmath493 such that @xmath494 for @xmath495 , @xmath496 is immediate , and shows that also holds for @xmath497 .
note that in the case @xmath498 , holds trivially . since @xmath499 is a direct consequence of and
our choice of @xmath207 , @xmath500 follows , and , therefore , the upper bound holds for all @xmath501 .
note that the result implies that @xmath449 changes sign at a time @xmath502 , which shows that @xmath74 actually crosses @xmath128 at a time @xmath145 satisfying @xmath503 . for large enough @xmath480
, the proposition also shows that @xmath504 for @xmath505 and that @xmath506 .
we consider now the case @xmath507 with @xmath480 from proposition [ prop_sdet2 ] . without loss of generality , we may assume that @xmath508 . [ prop_sdet3 ] assume that @xmath507 .
then , for any fixed @xmath509 , @xmath510 and @xmath74 crosses @xmath128 at a time @xmath145 satisfying @xmath511 . in order to show , we rescale space and time in the following way : @xmath512 let @xmath513
. then @xmath514 , and @xmath91 satisfies the differential equation @xmath515 where @xmath516 is a perturbation of order @xmath29 of the bernoulli equation @xmath517 using gronwall s inequality , one easily shows that on an @xmath14-time scale of order @xmath4 , the solution of differs by @xmath518 from the solution of , which is @xmath519 this function is bounded away from zero , and remains of order one for @xmath14 of order one , which shows that @xmath520 on @xmath192 $ ] . since @xmath521 and @xmath522 , @xmath74 and @xmath128 necessarily cross at some time @xmath511 . note
that the above proposition also implies bounds on @xmath449 , namely , @xmath523 for @xmath524 , and there exist constants @xmath525 such that @xmath526 gathering the results for @xmath527 and @xmath528 , we see that there exists a time @xmath529 such that @xmath530 . by enlarging @xmath460 if necessary
, we may assume that @xmath531 .
[ prop_sdet4 ] on the interval @xmath532 $ ] , @xmath533 the proof is similar to the one of proposition [ prop_sdet1 ] .
note that the previous result implies @xmath534 , provided @xmath460 is large enough .
so far , we have proved that for @xmath370 $ ] , @xmath74 tracks @xmath128 at a distance of order @xmath535 and that the two curves cross at a time @xmath145 satisfying @xmath536 .
let us now examine the behaviour of the linearization @xmath537 which will determine the behaviour of orbits starting close to the particular solution @xmath74 .
[ prop_sdet5 ] for all @xmath370 $ ] and all @xmath538 , @xmath539 by taylor s formula we get @xmath540 consider first the case @xmath541 for @xmath480 large enough .
equation implies that @xmath542 for a constant @xmath543 . on the other hand , shows that the second term on the right - hand side of is bounded in absolute value by @xmath544 for a constant @xmath545 . thus if @xmath546 we obtain that @xmath547 .
we consider next the case @xmath507 . for @xmath548
, the above argument can be repeated .
the non - trivial case occurs for @xmath549 . by rescaling variables as in proposition [ prop_sdet3 ]
, we obtain that @xmath550 we have to show that @xmath551 which is equivalent to @xmath552 for @xmath14 of order one .
the lower bound is trivial as @xmath553 and @xmath554 . in order to show the upper bound , first note that for @xmath555 , we have @xmath556 which implies @xmath557 .
therefore , it is sufficient to consider @xmath558 . taking into account the expression
, we find that showing the upper bound amounts to showing that @xmath559 since @xmath560 is proportional to @xmath460 , choosing _ a priori _ a large enough @xmath460 also makes @xmath560 large .
thus it is in fact sufficient to verify that @xmath561 for all @xmath562 .
optimizing the left - hand side with respect to @xmath563 and @xmath14 shows that we may assume @xmath564 and that holds .
we now consider the sde @xmath565 on the time interval @xmath133 $ ] , where we assume @xmath566 . in order to compare the solution @xmath74 with the solution @xmath137 of the corresponding deterministic equation
, we introduce the difference @xmath567 , which satisfies the sde @xmath568 where @xmath569 is the linearization of @xmath61 along @xmath137 , and taylor s formula yields the relations @xmath570 whenever @xmath426 and @xmath571 , where @xmath572 is a positive constant .
let us first consider the linearization of , namely @xmath573 the random variable @xmath574 is gaussian with expectation zero and variance @xmath575 lemma [ lem_sdet1 ] and proposition [ prop_sdet5 ] imply that @xmath576 thus @xmath164 is of order @xmath577 , except for @xmath21 very close to @xmath143 .
we now show that @xmath574 is likely to remain in a strip of width proportional to @xmath578 .
[ prop_snear1 ] for @xmath579 , @xmath580 where @xmath581 let @xmath582 , with some @xmath101 , be a partition of @xmath583 $ ] . in (
* lemma 3.2 ) , we show that the probability is bounded above by @xmath584 now we choose the partition by requiring that @xmath585 since @xmath586 , we have @xmath587/\eps \leqs1/\eps$ ] , and thus @xmath588 if @xmath589 is such that @xmath590 , then by and , there is a constant @xmath591 such that @xmath592 and hence by ( choosing the same @xmath591 for brevity of notation ) @xmath593 for all other @xmath589 , we have @xmath594 in both cases , we find @xmath595 which leads to the result , using the definition of @xmath105 .
let us now compare solutions of the linear equation and the nonlinear equation .
we introduce the events @xmath596 } \\ \label{snear9b } \omega^0_t(h ) & = \bigsetsuch{\w}{\abs{y^0_s}\leqs h\sqrt{\z(s)}\;\;\forall s\in[-t , t]}. \end{aligned}\ ] ] [ n_snear ] for two events @xmath597 and @xmath598 , we write @xmath599 if @xmath66-almost all @xmath600 belong to @xmath598 .
[ prop_snear2 ] there exists a constant @xmath601 , depending only on @xmath61 , such that if @xmath177 and @xmath602 , then @xmath603 if @xmath180 and @xmath604 , then @xmath605 the proof is based on the fact that the variable @xmath606 satisfies the relation @xmath607 consider first the case @xmath177 .
let @xmath608 be a constant to be chosen later , and set @xmath609 .
we define the first exit time @xmath610}{\abs{z_s}\geqs\delta h\sqrt{\z(s ) } } \in [ -t , t]\cup \{\infty\}.\ ] ] pick any @xmath611 and @xmath612 $ ] .
then we have @xmath613 for all @xmath614 $ ] . from and
, we obtain the existence of a constant @xmath545 such that @xmath615 for these @xmath616 .
hence , by we get the estimate @xmath617 and thus , by and lemma [ lem_sdet1 ] , @xmath618 where we use again the same @xmath619 for brevity of notation .
using once again , we arrive at the bound @xmath620 now we choose @xmath621 which implies @xmath622 for all @xmath623 $ ] , by the definition of @xmath104 .
hence @xmath624 for almost all @xmath625 .
since we have @xmath626 whenever @xmath627 , we conclude that @xmath628 , and thus @xmath629 for almost all @xmath630 , which implies that @xmath631 for @xmath632 and these @xmath633 .
this completes the proof of .
the proof of is almost the same . in the case
@xmath634 , we take @xmath635 .
the estimate has to be replaced by @xmath636 and thus we get , instead of , the bound @xmath637 the remainder of the proof is similar .
now , the preceding two propositions immediately imply theorem [ thm_snearxdet ] , as proposition [ prop_snear1 ] shows the desired behaviour for the approximation by a gaussian process and proposition [ prop_snear2 ] allows to extend this result to the original process .
we consider now the regime of @xmath2 sufficiently large to allow for transitions from one stable equilibrium branch to the other . here
@xmath137 is the solution of the deterministic equation with the same initial condition @xmath638 as in the previous sections , which tracks @xmath128 at distance at most @xmath518 .
@xmath74 denotes a general solution of the sde .
our aim is to establish an upper bound for the probability of _ not _ reaching the axis @xmath119 , which , by using symmetry , will allow us to estimate the transition probability .
let @xmath100 be the constant defined in , i.e. by @xmath639 the basic ingredient of our estimate is the following comparison lemma which allows us to linearize the stochastic differential equations under consideration and , therefore , to investigate gaussian approximations to our processes .
the lemma gives conditions under which relations between initial conditions carry over to the sample paths .
[ lem_strans ] fix some initial time @xmath640 $ ] .
we consider the following processes on @xmath641 $ ] : the solution @xmath137 of the deterministic differential equation with initial condition @xmath642 $ ] ; the solution @xmath74 of the sde with initial condition @xmath643 $ ] ; the difference @xmath644 , which satisfies @xmath645 ; the solution @xmath574of the linear sde @xmath646 with initial condition @xmath647 $ ] .
if @xmath648 for all @xmath649 $ ] , then @xmath650 for those @xmath14 .
similarly , if @xmath651 for all @xmath649 $ ] , then @xmath652 for those @xmath14 .
the result remains true when @xmath21 is replaced by a stopping time .
the hypothesis implies that for all @xmath653 $ ] , @xmath654 let @xmath655}{y_s\not\in[0,\delta-\xdet_s ] } \in [ t_0,t ] \cup\{\infty\}$ ] .
for @xmath656 , the variable @xmath657 satisfies @xmath658 applying gronwall s inequality , we obtain @xmath659,\ ] ] where @xmath660 .
this proves the result for @xmath661 .
now if @xmath662 is negative , the result is trivially satisfied , and if @xmath662 becomes positive again , the above argument can be repeated . note that @xmath663 is immediate .
this proves the first assertion , and the second assertion can be proved directly , without use of @xmath448 .
we will now proceed as follows .
let @xmath171 be the function defined in , and let @xmath184 be such that @xmath205 for all @xmath649 $ ] .
given @xmath664 , we can write @xmath665 } \leqs { } & \bigprobin{t_0,x_0}{\sup_{t_0\leqs s\leqs t } \frac{x_s-\xdet_s}{\sqrt{\z(s ) } } > h } \\ & { } + \bigprobin{t_0,x_0}{0<x_s\leqs\xdet_s+h\sqrt{\z(s ) } \;\ ; \forall s\in[t_0,t]}.\end{aligned}\ ] ] we will estimate these two terms separately .
the first event is similar to the event we have examined in the previous subsection , but we need here an estimate valid for all times , even when @xmath2 is not very small , whereas the previous result is only useful for @xmath666 .
we will show the following .
[ prop_strans1 ] assume @xmath667 .
then @xmath668 where @xmath210 is a positive constant and @xmath669 .
we define a partition @xmath670 of the interval @xmath671 $ ] by requiring @xmath672 note that similar arguments as in the proof of proposition [ prop_snear1 ] yield @xmath673 now let @xmath674 and @xmath675 as usual .
define @xmath676 for @xmath677 , and @xmath678 then @xmath679 in order to estimate @xmath680 , we introduce the stochastic process @xmath681}$ ] defined by @xmath682 where @xmath683}$ ] is the brownian motion @xmath684 .
note that @xmath685 is the solution of the sde with initial condition at time @xmath686 .
we define the stopping times @xmath687}{y^{(k)}_s=0 } \\
\tau^+ & = \inf\bigsetsuch{s\in[u_k , u_{k+1}]}{y^{(k)}_s = h\sqrt{\z(s)}\ , } \end{split}\ ] ] describing the time when @xmath688 either reaches the @xmath21-axis or the upper boundary @xmath689 .
now , lemma [ lem_strans ] implies that if @xmath690 , then @xmath691 for @xmath692 .
this shows that @xmath693 for @xmath677 , and @xmath694 each of these terms depends only on @xmath685 , and can be easily estimated .
let @xmath695 denote the variance of @xmath696 .
then by symmetry ( as in ) , we have @xmath697 the second term on the right - hand side of or , respectively , can be estimated using the symmetry ( in distribution ) of under the map @xmath698 : @xmath699
\colon y^{(k)}_s \geqs h\sqrt{\z(s ) } } \\ \nonumber & \leqs \bigprobin{u_k,\rho_k}{\exists s\in[u_k , u_{k+1 } ] \colon y^{(k)}_s \leqs h \sqrt{\z(u_k ) } \operatorname{e}^{\balpha(s , u_k)/\eps } - h \sqrt{\z(s ) } } \\ & \leqs \bigprobin{u_k,\rho_k}{\tau^0<u_{k+1}}. \end{aligned}\ ] ] in order to estimate the third term on the right - hand side of , we will use the fact that for @xmath700 @xmath701 proposition [ prop_sdet5 ] and thus yield the existence of a constant @xmath543 such that @xmath702 this allows us to estimate ( for @xmath700 ) @xmath703 the estimates , and , inserted in and , imply that @xmath704 with @xmath705 by , for each @xmath589 , there exists a @xmath706 $ ] such that @xmath707 together with , this implies that @xmath708 for all @xmath589 , and thus the result follows from with @xmath709 .
we now give an estimate of the second term in .
the markov property implies that we will obtain an upper bound by starting at time @xmath710 .
[ prop_strans2 ] there exist constants @xmath200 and @xmath711 such that , if @xmath712 and @xmath204 , then @xmath713 } \\
\leqs 2 \exp\biggset{-\bar\kappa \mskip3mu \frac1{\log(h/\sigma ) } \frac{\alpha(t_1,-c_1\sqrt\sigma)}\eps } \qquad\qquad\end{gathered}\ ] ] holds with @xmath714 , for @xmath715 and all initial conditions @xmath65 satisfying @xmath716 .
let @xmath717 and define a partition @xmath718 of @xmath221 $ ] by @xmath719 we would like to control the probability of not reaching the @xmath21-axis during the time interval @xmath720 $ ] .
let @xmath721}.\ ] ] then the probability on the left - hand side of is @xmath722}\\ \nonumber&\qquad = \bigexpecin{-c_1\sqrt\sigma , x_0}{\indexfct{0<x_s\leqs\xdet_s+h\sqrt{\z(s ) } \;\ ; \forall s\in[-c_1\sqrt\sigma , u_{k-1 } ] } \\ \nonumber & \qquad\hphantom{{}={{\mathbb{e}}}^{\mskip1.5mu -c_1\sqrt\sigma , x_0}\bigl\{\null\bigr . }
\bigprobin{u_{k-1},x_{u_{k-1}}}{0<x_s\leqs\xdet_s+h\sqrt{\z(s ) } \;\ ; \forall s\in[u_{k-1},u_k ] } } \\ \nonumber & \qquad { } \leqs q_{k-1 } \ , \bigprobin{-c_1\sqrt\sigma , x_0}{0<x_s\leqs\xdet_s+h\sqrt{\z(s ) } \;\ ; \forall s\in[-c_1\sqrt\sigma , u_{k-1 } ] } \\ & \qquad { } \leqs \dots \leqs \prod_{k=0}^{k-1 } q_k.\end{aligned}\ ] ] if we manage to estimate each @xmath680 by a constant less than @xmath4 ( say , @xmath3 ) , then the probability will be exponentially small in @xmath105 . in the sequel
, we shall estimate @xmath680 uniformly in @xmath723 , and bound @xmath724 by @xmath4 , since the last interval of the partition may be too small to get a good bound .
so let @xmath700 from now on .
we consider first the case @xmath725 .
we define the process @xmath726 as the solution of the linearized sde @xmath727 where @xmath728}$ ] is the brownian motion @xmath729 .
let @xmath730 denote the variance of @xmath731 .
then @xmath732 we can now apply lemma [ lem_strans ] in the particular case @xmath733 to show that if @xmath734 for @xmath735 $ ] , then @xmath736 in the same interval .
we thus obtain @xmath737 } \leqs \bigprobin{u_k , x_k}{x^{(k)}_s > 0 \;\ ; \forall s\in[u_k , u_{k+1}]}. \lower1\jot\hbox{\strut}\ ] ] the probability on the right - hand side satisfies @xmath738 } & = 1- 2\mskip1.5mu \bigprobin{u_k , x_k}{x^{(k)}_{u_{k+1 } } < 0}\\ \label{strans2:6b } & = 2\mskip1.5mu \bigprobin{u_k , x_k}{x^{(k)}_{u_{k+1 } } \geqs 0 } -1 , \lower1\jot\hbox{\strut}\end{aligned}\ ] ] yielding @xmath739 } \\ \nonumber&\qquad \leqs \frac2{\sqrt{2\pi } } \int_0^{{x_k\operatorname{e}^{\alpha(u_{k+1},u_k)/\eps } } \bigpar{v^{(k)}_{u_{k+1}}}^{-1/2 } } \operatorname{e}^{-z^2/2 } \6z \\ \nonumber&\qquad \leqs \frac2{\sqrt{2\pi } } \frac{x_k}{\sqrt{\operatorname{e}^{-2\alpha(u_{k+1},u_k)/\eps } v^{(k)}_{u_{k+1 } } } } \\ & \qquad \leqs \frac2{\sqrt{\pi } } \sqrt{\frac1{1-\operatorname{e}^{-2\varrho } } } \frac{\sqrt{a(u_k)\vee a(u_{k+1})}}\sigma \,\xdet_{u_k}.\end{aligned}\ ] ] by making @xmath207 small enough , we can guarantee that this bound is smaller than some imposed constant of order @xmath4 , say @xmath3 .
this shows that the length of @xmath720 $ ] has been chosen large enough that the probability of reaching the @xmath21-axis during this time interval is appreciable .
we examine now the case @xmath740 .
we introduce a time @xmath741 , defined by @xmath742 our strategy will be to show that @xmath74 is likely to cross @xmath137 before time @xmath743 , which will allow us to use the previous result . proposition [ prop_sdet5 ] implies the existence of a constant @xmath744 such that @xmath745 let @xmath746 be the solution of the linear sde @xmath747 where @xmath728}$ ] is again the brownian motion @xmath729 .
the variance of @xmath748 is @xmath749 lemma [ lem_strans ] shows that if @xmath750 on the interval @xmath751 $ ] , then @xmath752 on that interval . if we introduce the stopping time @xmath753}{x_s=\xdet_s } \in[u_k , u_{k+1}]\cup\{\infty\},\ ] ] then we have @xmath754 } \\
\nonumber&\quad \leqs \bigprobin{u_k , x_k}{\xdet_s <
x_s\leqs\xdet_s+h\sqrt{\z(s ) } \;\ ; \forall s\in[u_k,\tilde u_k ] } \\ & \quad \phantom{\leqs { } } + \bigexpecin{u_k , x_k}{\indexfct{\tau_k<\tilde u_k } \mskip1.5mu \bigprobin{\tau_k,\xdet_{\tau_k}}{0<x_s\leqs\xdet_s+h\sqrt{\z(s ) } \;\ ; \forall s\in[\tau_k , u_{k+1}]}}. \label{strans2:11}\end{aligned}\ ] ] the second term on the right - hand side can be bounded , as in , by @xmath755 using , the first term on the right - hand side of can be estimated in the following way : @xmath756 } \\
\nonumber&\qquad \vrule height 12pt depth 7pt width 0pt \leqs \bigprobin{u_k , y_k}{y^{(k)}_s > 0 \;\ ; \forall s\in[u_k,\tilde u_k ] } \\
\nonumber&\qquad \leqs \frac2{\sqrt{2\pi } } \frac{y_k \operatorname{e}^{\balpha(\tilde u_k , u_k)/\eps } } { \bigpar{\tilde v^{(k)}_{\tilde u_k}}^{1/2 } } \\
& \qquad \leqs \frac2{\sqrt{\pi } } \sqrt{\frac1{1-\operatorname{e}^{-\varrho / l } } } \sqrt{\abs{\ba(u_k)}\vee\abs{\ba(\tilde u_k ) } } \,\frac h\sigma \sqrt{\z(u_k ) } \operatorname{e}^{-\varrho/2l}. \label{strans2:12}\end{aligned}\ ] ] using proposition [ prop_sdet5 ] and , it is easy to show that the expression @xmath757 is uniformly bounded by a constant independent of @xmath589 and @xmath1 .
the sum of and of the last term in provides an upper bound for @xmath680 .
using the fact that for @xmath758 and @xmath759 , one has @xmath760 and @xmath761 , we arrive at the bound @xmath762 where the constant @xmath763 can be chosen independent of @xmath601 because @xmath764 by assumption .
thus if we choose @xmath765 and @xmath766 , we obtain that @xmath767 for @xmath768 .
this yields @xmath769 and the result follows from our choice of @xmath601 .
now the proof of theorem [ thm_strans ] follows from and the two preceding propositions , where we use the markov property to restart at time @xmath710 before applying proposition [ prop_strans2 ] . in this subsection
, we investigate the behaviour of the random motion @xmath74 given by the sde for @xmath770 , i.e. , after the transition regime .
we want to show that @xmath74 is likely to leave a suitably defined neighbourhood of the saddle within time @xmath771 .
the proof of theorem [ thm_sescape ] is very similar to the proof of ( * ? ? ?
* theorem 2.9 ) , and for the sake of brevity , we will refrain from giving all the details . instead
, we will discuss how to proceed and then focus on those parts which need to be modified .
from now on , we will assume that @xmath772 and that @xmath2 is large enough in order to allow for transitions , i.e. , @xmath773 .
we want to estimate the first exit time @xmath228 of @xmath74 from the set @xmath774\times[c_1\sqrt\sigma , t ] } { \frac{f(x , t)}{x } > \kappa a(t)},\ ] ] where @xmath224 is a constant .
note that the upper boundary @xmath775 of @xmath226 satisfies @xmath776 .
our first step towards estimating @xmath228 is to estimate the first exit time @xmath777 from a smaller strip @xmath778 , defined by @xmath779\times[c_1\sqrt\sigma , t ] } { \abs{x } < \frac h{\sqrt{a(s)}}},\ ] ] where we will choose @xmath184 later .
note that @xmath780 for some ( small ) constant would assure @xmath781 .
we will not impose such a restrictive condition on @xmath184 but replace @xmath778 by @xmath782 in case @xmath778 is not a subset of @xmath226 .
the following proposition gives our estimate on the first exit time from @xmath778 .
[ p_sescape ] let @xmath783 and @xmath784 .
then there exists a constant @xmath744 such that for any @xmath785 , we have @xmath786 under the condition @xmath787 the proof follows along the lines of the one of ( * ? ? ?
* proposition 4.7 ) , the main difference being the quadratic behaviour @xmath788 of @xmath789 in our case as opposed to the linear one in @xcite .
we start by defining a partition @xmath790 of @xmath791 $ ] , given by @xmath792 on each interval @xmath720 $ ] , we consider a gaussian approximation @xmath793}$ ] of @xmath74 , defined by @xmath794 where @xmath795 .
if @xmath796 for all @xmath735 $ ] , then by and , there is a constant @xmath797 such that @xmath798 for all @xmath735 $ ] , provided the condition @xmath799 holds for all @xmath589 .
now , @xmath800 where @xmath248 is a constant satisfying @xmath801 for all @xmath802 $ ] .
this shows that there exists a constant @xmath744 such that the condition is satisfied whenever @xmath803 this condition is equivalent to .
assume @xmath804 for the moment .
then , @xmath805 where @xmath730 denotes the variance of @xmath731 . by partial integration , we find @xmath806 now , the markov property yields @xmath807 and the bound follows by a straightforward calculation . the preceding proposition shows that a path starting in @xmath778 is likely to leave @xmath778 after a short time .
we want to show that such a path ( or any path starting in @xmath808 ) is also likely to leave @xmath226 . for this purpose
, we will again compare @xmath74 to a gaussian approximation , given by @xmath809 where @xmath810 , so that @xmath811 in @xmath226 .
assume that @xmath812 .
then @xmath813 holds as long as @xmath212 neither leaves @xmath226 nor crosses the @xmath21-axis , cf .
* lemma 4.8 ) .
therefore we can proceed as follows .
once a path is in @xmath808 , there are two possibilities . either , @xmath814 does not return to zero , or it does .
if @xmath815 does not return to zero , then it is likely to leave @xmath226 via the upper boundary and so is @xmath74 .
so we are left with the case of @xmath815 returning to zero .
this event has a small but not negligible probability .
note that if @xmath815 returns to zero , then @xmath74 is still non - negative . if @xmath74 has nevertheless left @xmath226 , we are done . if not , @xmath74 is either in @xmath778 or in @xmath808 . since we may assume that , after a short time , @xmath74 is in @xmath808 again , we can repeat the above argument . making the above - said precise
, we obtain an integral equation for an upper bound on the probability that @xmath816 does not leave @xmath226 up to time @xmath21 , which will be solved by iterations .
we will cite the integral equation from @xcite , as the general arguments leading to it do not require adaptation .
let us first introduce the necessary notations .
we choose @xmath817 for some ( possibly large ) constant @xmath101 . for @xmath224
, we choose @xmath785 in such a way that @xmath818 for all large enough @xmath105 .
note that choosing @xmath257 too close to @xmath4 requires large @xmath819 and is thus not desirable .
since we want to apply proposition [ p_sescape ] on the first exit time from @xmath778 with @xmath245 for a suitably chosen @xmath375 , condition must be satisfied .
therefore , we choose @xmath820 large enough for @xmath821 to hold .
now , set @xmath822 and @xmath823 for @xmath824 , let @xmath825 , and define @xmath826 for @xmath827 by @xmath828 then , ( * ? ? ?
* ( 4.95 ) and ( 4.107 ) ) show that for any @xmath829 and @xmath830 , @xmath831 and @xmath832 next we estimate @xmath833 by showing that @xmath834 holds with @xmath835 , @xmath836 in the case @xmath837 , and with @xmath838 for @xmath839 , provided @xmath840 .
note that the latter imposes a condition on @xmath841 . to obtain the bound
, we proceed as in @xcite , the only difference lying in the term @xmath842 , where we sacrifice a factor of @xmath236 in the exponent in order to gain a smaller coefficient @xmath843 .
our choice of @xmath843 yields a less restrictive condition on @xmath844 , namely we only need @xmath845 which is satisfied whenever @xmath841 is large enough .
now , and imply that for @xmath105 and @xmath846 large enough , @xmath847 with some constant @xmath848 .
this completes our outline of the proof of theorem [ thm_sescape ] .
we consider in this section the nonlinear sde @xmath849 where @xmath61 satisfies the hypotheses given at the beginning of subsection [ ssec_rasym ] . by rescaling @xmath15
, we can arrange for @xmath850 , so that taylor s formula allows us to write @xmath851 where @xmath852 and @xmath853 . since @xmath854 by assumption , @xmath855 vanishes on a curve @xmath856 .
we further obtain that @xmath857 where @xmath858 and @xmath859 .
thus @xmath310 vanishes on two curves @xmath319 and @xmath321 , which behave near @xmath44 like @xmath860 $ ] , as indicated in .
the behaviour of the linearization follows from .
the proof of theorem [ thm_adet ] follows closely the proof of theorem [ thm_sdet ] , with some minor differences we comment on here .
the dynamics of @xmath861 is still governed by an equation of the form @xmath862 but now taylor s formula yields the relations @xmath863 while relation holds for the derivative of @xmath864 , with @xmath865 replaced by @xmath866 .
lemma [ lem_sdet1 ] becomes proposition [ prop_sdet1 ] carries over with some obvious adjustments , and shows the existence of a constant @xmath460 such that @xmath869 in particular , @xmath870 .
an adaptation of proposition [ prop_sdet2 ] yields the existence of a constant @xmath472 such that , for @xmath871 , @xmath872 for all @xmath470 .
this shows in particular that @xmath449 vanishes at a time @xmath145 satisfying @xmath873 .
proposition [ prop_sdet3 ] is replaced by let @xmath878 .
we first observe that by taylor s formula , @xmath879 this shows that @xmath880 thus , with the rescaling @xmath881 we obtain that @xmath882 obeys a perturbation of order @xmath48 of the riccati equation @xmath883 one easily shows that the solution satisfies @xmath884 for @xmath14 of order 1 , and this property carries over to the perturbed equation with the help of gronwall s inequality . finally , since @xmath885 and @xmath886
, these curves necessarily cross at a time @xmath887 .
the assertion on the existence of a particular solution @xmath348 tracking the unstable equilibrium branch @xmath321 follows from the observation that @xmath888 satisfies the equation @xmath889 this system admits @xmath890 as a stable equilibrium branch .
thus the same arguments as above can be used to show the existence of a solution @xmath891 tracking @xmath892 , with similar properties . proposition [ prop_sdet5 ] admits the following counterpart : we consider now the regime of @xmath2 sufficiently large to allow for transitions from the potential well at @xmath314 to the potential well at @xmath326 , by passing over the saddle at @xmath312 . here
@xmath137 and @xmath244 denote solutions of the deterministic equation @xmath895 tracking , respectively , the stable equilibrium branch @xmath319 and the unstable equilibrium branch @xmath321 , while @xmath74 denotes a general solution of the sde .
our aim is to establish an upper bound for the probability _ not _ to reach a level @xmath367 between @xmath321 and @xmath896 , situated at a distance of order @xmath4 from both equilibria .
* theorem 2.3 ) shows that if @xmath74 reaches @xmath367 , and @xmath367 is close enough to @xmath896 ( but it may still be at a distance of order @xmath4 ) , then it is likely to reach a of @xmath896 as well .
if @xmath904 for all @xmath649 $ ] , then @xmath650 for those @xmath14 .
similarly , if @xmath905 for all @xmath649 $ ] , then @xmath652 for those @xmath14 .
the result remains true when @xmath21 is replaced by a stopping time .
we will proceed as follows .
let @xmath171 be the function defined in , and let @xmath184 be such that @xmath906 for all @xmath649 $ ] .
given @xmath907 and times @xmath908 in @xmath133 $ ] , we consider the solution @xmath74 of the sde with initial condition @xmath909 .
we introduce the stopping time @xmath910}{x_s\leqs\delta_1 } \in[t_0,t_1 ] \cup\set{\infty}.\ ] ] we can thus write @xmath911 } \leqs { } & \bigprobin{t_0,x_0}{x_s>\delta_1 \;\ ; \forall s\in[t_0,t_1 ] } \\ \label{atrans5 } & { } + \bigexpecin{t_0,x_0}{\indexfct{\tau\leqs t_1 } \bigprobin{\tau,\delta_1}{x_s>\delta_0 \;\ ; \forall s\in[\tau , t]}}. \end{aligned}\ ] ] the first term can be further estimated by @xmath912 } \leqs { } & \bigprobin{t_0,x_0}{\sup_{t_0\leqs s\leqs t_1 } \frac{x_s-\xdet_s}{\sqrt{\z(s ) } } > h } \\ \label{atrans6 } & { } + \bigprobin{t_0,x_0}{\delta_1<x_s\leqs\xdet_s+h\sqrt{\z(s ) } \;\ ; \forall s\in[t_0,t_1]}.\end{aligned}\ ] ] the two summands in can be estimated in a similar way as in the symmetric case .
the first one is dealt with in the following result .
the proof is almost the same as the proof of proposition [ prop_strans1 ] . instead of
, we may define @xmath916 and @xmath917 as the first times when @xmath918 either reaches @xmath919 or the upper boundary @xmath265 .
then lemma [ lem_atrans ] implies @xmath920 for @xmath921 $ ] .
however , when estimating the probability that @xmath922 as in , it is sufficient to use the fact that @xmath916 is larger than the first time @xmath918 reaches @xmath213 . finally , still holds with the present definitions of @xmath923 and @xmath924 , because of and proposition [ prop_adet5 ] .
[ prop_atrans2 ] there exist constants @xmath200 and @xmath711 such that , if @xmath925 and @xmath204 , then @xmath926 } \\
\leqs \frac32 \exp\biggset{-\bar\kappa \mskip3mu \frac1{\log(h/\sigma ) \vee\abs{\log\sigma } } \frac{\halpha(t_1,-c_1\sigma^{2/3})}\eps } \qquad\qquad\qquad\end{gathered}\ ] ] holds with @xmath927 , for @xmath928 and all initial conditions @xmath65 satisfying @xmath929 .
let @xmath244 be the deterministic solution tracking the saddle at @xmath321 and set @xmath930 .
then @xmath931 where , and imply @xmath932 and @xmath933$.}\ ] ] let @xmath717 and define a partition @xmath934 of @xmath935 $ ] by @xmath936 let @xmath937}.\ ] ] then we have , as in , @xmath938 } \leqs \prod_{k=0}^{k-1 } q_k.\ ] ] the result will thus be proved if we manage to choose @xmath601 in such a way that @xmath680 is bounded away from @xmath4 for @xmath768 .
we will estimate the @xmath680 in a similar way as in proposition [ prop_strans2 ] , but we shall distinguish three cases instead of two .
these cases correspond to @xmath212 reaching the levels @xmath216 , @xmath939 and @xmath368 .
we introduce a subdivision @xmath940 defined by @xmath941 and stopping times @xmath942 } { z_s\leqs\xdet_s-\xhatdet_s } \\
\tau_{k,2 } & = \inf\bigsetsuch{s\in[u_k,\tilde u_{k,2 } ] } { z_s\leqs 0}. \end{split}\ ] ] then we can write , similarly as in , @xmath943}\\ \nonumber & \vrule height 13pt depth 7pt width 0pt \leqs \bigprobin{u_k , z_k}{\xdet_s<\xhatdet_s+z_s\leqs\xdet_s+h\sqrt{\z(s ) } \;\ ; \forall s\in[u_k,\tilde u_{k,1 } ] } \\ & \phantom{\leqs { } } { } + \bigexpecin{u_k , z_k}{\indexfct{\tau_{k,1}<\tilde u_{k,1 } } \bigprobin{\tau_{k,1},z_{\tau_{k,1 } } } { \delta_1<\xhatdet_s+z_s\leqs\xdet_s+h\sqrt{\z(s ) } \;\ ; \forall s\in[\tau_{k,1},u_{k+1 } ] } } \nonumber\end{aligned}\ ] ] the first term can be bounded by comparing with the solution of the sde linearized around @xmath137 , with the help of lemma [ lem_atrans ] . as in
, we obtain the upper bound @xmath944 where @xmath744 is a constant such that @xmath945 for all @xmath616 . now
if @xmath946 , we also have @xmath947}\\ \label{atrans2:10 } \vrule height 13pt depth 7pt width 0pt & \leqs \bigprobin{\tau_{k,1},z_{\tau_{k,1 } } } { 0<z_s\leqs\xdet_s-\xhatdet_s+h\sqrt{\z(s ) } \;\ ; \forall s\in[\tau_{k,1},\tilde u_{k,2 } ] } \\
\nonumber & \phantom{\leqs { } } { } + \bigexpecin{\tau_{k,1},z_{\tau_{k,1 } } } { \indexfct{\tau_{k,2}<\tilde u_{k,2 } } \\ & \qquad\qquad\qquad\quad \bigprobin{\tau_{k,2},z_{\tau_{k,2 } } } { \delta_1<\xhatdet_s+z_s\leqs\xdet_s+h\sqrt{\z(s ) } \;\ ; \forall s\in[\tau_{k,2},u_{k+1}]}}. \nonumber\end{aligned}\ ] ] comparing with the solution of the sde linearized around @xmath244 , the first term can be bounded , as in , by @xmath948 this estimate shows that a path starting on @xmath949 at time @xmath950 has an appreciable probability to reach the saddle before time @xmath951 .
note , however , that we can not obtain directly a similar estimate for the probability to reach @xmath368 as well , which is why we restart the process in @xmath348 .
in order to estimate the second summand in , let @xmath952 be the process starting in @xmath213 at time @xmath953 which satisfies the linear equation @xmath954 with @xmath955 .
the variance @xmath730 of @xmath956 satisfies , as in , @xmath957 thus we obtain , using lemma [ lem_atrans ] , @xmath958 } \\
\nonumber&\qquad \vrule height 13pt depth 7pt width 0pt \leqs \bigprobin{\tau_{k,2},0}{z^{(k)}_{u_{k+1 } } \geqs-(\xhatdet_{u_{k+1}}-\delta_1 ) } \\ \nonumber&\qquad { } = \frac1{\sqrt{2\pi } } \int_{-(\xhatdet_{u_{k+1}}-\delta_1)(v^{(k)}_{u_{k+1}})^{-1/2}}^\infty \operatorname{e}^{-z^2/2}\6z \\ \label{atrans2:14}&\qquad
{ } \leqs \frac12 + \frac1{\sqrt\pi}\sqrt{\frac1{1-\operatorname{e}^{-2\varrho/3 } } } \sup_{u_k\leqs u\leqs u_{k+1 } } \frac{\sqrt{\ha(u)}}\sigma ( \xhatdet_{u_{k+1}}-\delta_1 ) \operatorname{e}^{-\varrho/3}.\end{aligned}\ ] ] here the introduction of the stopping time @xmath953 turns out to play a crucial role .
the above probability is indeed close to @xmath3 when @xmath601 is larger than a constant times @xmath959 , which shows that once a path has reached the saddle , it also has about fifty percent chance to reach the level @xmath368 in a time of order @xmath960 . from , and and
the fact that @xmath764 , we obtain the existence of a constant @xmath961 such that @xmath962 since @xmath963 , the properties of @xmath949 , @xmath348 , @xmath923 and @xmath964 imply the existence of another constant @xmath965 such that @xmath966 now choosing @xmath967 and @xmath968 , we get @xmath969 for @xmath768 and thus @xmath970 and the result follows from our choice of @xmath601 .
it remains to estimate the second term in , describing the probability not to reach @xmath367 when starting in @xmath368 .
this is done by using the fact that , by assumption , the drift term is bounded away from zero on the interval @xmath971 $ ] .
we will need to assume that it can be bounded away from zero on a slightly larger interval , which is possible by continuity of @xmath61 .
[ prop_atrans3 ] let @xmath972 be a constant such that @xmath973 for @xmath974 and @xmath426
. then @xmath975 } \leqs \operatorname{e}^{-\tilde\kappa/\sigma^2}\ ] ] holds for all @xmath976 $ ] , where @xmath977 and @xmath978 .
let @xmath815 be defined by @xmath979 by gronwall s inequality , it is easy to see , as in lemma [ lem_atrans ] , that if @xmath980 for all @xmath981 $ ] , then @xmath982 for those @xmath21 .
we thus have @xmath983 } \leqs { } \bigprobin{t_0,\delta_1}{\sup_{t_0\leqs s\leqs t_0+c\eps } x^0_s + \frac{f_0}{\eps}(s - t_0 ) > \delta_1+\rho } \\ \label{atrans3:2 } { } + \bigprobin{t_0,\delta_1}{\delta_0<x^0_s<\delta_1+\rho-\frac{f_0}{\eps}(s - t_0 ) \;\ ; \forall s\in[t_0,t_0+c\eps]}. \end{gathered}\ ] ] note , however , that for @xmath984 , @xmath985 so that the second term in is equal to zero .
the first term equals @xmath986 by doob s submartingale inequality . | additive white noise may significantly increase the response of bistable systems to a periodic driving signal .
we consider two classes of double - well potentials , symmetric and asymmetric , modulated periodically in time with period @xmath0 , where @xmath1 is a moderately ( not exponentially ) small parameter .
we show that the response of the system changes drastically when the noise intensity @xmath2 crosses a threshold value . below the threshold ,
paths are concentrated near one potential well , and have an exponentially small probability to jump to the other well . above the threshold ,
transitions between the wells occur with probability exponentially close to @xmath3 in the symmetric case , and exponentially close to @xmath4 in the asymmetric case .
the transition zones are localised in time near the points of minimal barrier height .
we give a mathematically rigorous description of the behaviour of individual paths , which allows us , in particular , to determine the power - law dependence of the critical noise intensity on @xmath1 and on the minimal barrier height , as well as the asymptotics of the transition and non - transition probabilities .
stochastic resonance , noise - induced synchronization , double - well potential , additive noise , random dynamical systems , non - autonomous stochastic differential equations , singular perturbations , pathwise description , concentration of measure . |
experimental demonstrations of raman - induced spin - orbit coupling in cold atoms have motivated theoretical proposals for using these techniques to observe topological edge modes @xcite .
a practical concern with these proposals is that the raman process introduces an intrinsic heating mechanism . as discussed by wei and mueller @xcite
, this heating is particularly problematic for @xmath0 , one of the work - horse fermions used in cold atom experiments .
here we show how to exploit the rich hyperfine structure of alkali atoms to ameliorate this heating .
excitement about raman - induced spin - orbit coupling in cold atoms is related to broader interest in topological " states of matter , i.e. , those which are characterized by topological invariants which lead to non - trivial edge modes .
spin - orbit coupling is an essential ingredient in many electronic topological states @xcite , and these raman experiments promise to enable cold - atom studies of analogous physics .
a particularly natural setup would involve a gas of fermionic atoms in a quasi-1d `` wire - trap '' .
a magnetic - field induced feshbach resonance could drive the atoms into a paired superfluid state . adding spin - orbit coupling would lead to majorana edge modes @xcite .
the same photons which provide spin - orbit coupling also heat the cloud . here
we propose minimizing this heating by dividing our atoms into two populations , distinguished by their hyperfine spins .
both the feshbach resonant interactions and the raman couplings are state selective , and one of these populations can be chosen to be strongly interacting , yet feel no spin - orbit coupling .
the other will be essentially non - interacting , but experience a substantial spin - orbit coupling .
the raman lasers will produce heat , but the heating rate will be proportional to the number of atoms in the raman - coupled states . by making the superfluid population large compared to the raman - coupled population , one can make studying this system practical
. figure [ fig : fig1 ] shows the hyperfine level diagram of @xmath0 , and the states we propose using for this experiment .
the strongly interacting population ( @xmath1 and @xmath2 ) will be formed from the lowest and second lowest hyperfine states .
these states have a feshbach resonance at @xmath3 in 3d @xcite .
the superfluid formed by these states has been observed @xcite .
the raman - coupled population ( @xmath4 and @xmath5 ) will be formed from the third and fourth lowest hyperfine states .
figure [ fig : fig2 ] shows the ratio of raman coupling strength to inelastic scattering rate as a function of magnetic field for @xmath5 and @xmath4 states . at fields near the feshbach resonance , these matrix elements are comparable to those used in prior raman - induced spin - orbit coupling experiments .
interconversion between @xmath2 and @xmath4 can be driven by radio waves .
microwaves can independently drive the @xmath1 to @xmath5 transition . by appropriately detuning these transitions , one can set the relative equilibrium populations of each of the states . in the presence of such interconversion
, we find that there are parameters for which ground state will possess majorana edge modes . .
( a ) schematic of proposed experiment in real space .
there is a large superfluid reservoir composed of @xmath1 and @xmath2 fermions overlapping a smaller cloud of raman - induced spin - orbit coupled particles with spin states @xmath5 and @xmath4 .
( b ) hyperfine structure of @xmath0 , illustrating the states used in the proposed experiment .
the tunneling @xmath6 is turned on by micro ( radio ) waves between @xmath1 ( @xmath2 ) and @xmath5 ( @xmath4).,scaledwidth=47.0% ] in addition to reducing heating , our two - population approach avoids the difficult problem of finding spin states which can both have strong interaction and be coupled by raman lasers .
the r - atoms have no readily accessible feshbach resonances , and the raman coupling strengths for the s - atoms are 40 times smaller than those of the r - atoms ( when the intensity is adjusted to give the same inelastic scattering rates ) .
furthermore , our approach provides a nice analogy to the solid state experiments , where the superconductivity and the spin - orbit coupling comes from separate materials .
it is natural to ask if similar physics could be seen in higher dimension .
unfortunately , current cold atom techniques give a unidirectional spin - orbit coupling , and do not generalize to 2d or 3d . in the remainder of this paper , we model our proposed experiment . in section
ii we write a hamiltonian for the system , and explain how it can be analyzed . in section iii
we show a bogoliubov - de gennes calculation of the properties of this system .
we calculate the local density of states , which can be studied with spatially resolved radio frequency spectroscopy .
we find two regimes , one of which possesses topological edge modes .
we further explain this physics through a local density approximation . at the end
, we estimate the heating and collisional losses in our system .
we model our system with a hamiltonian @xmath7 where @xmath8 describes the energy of the strongly interacting reservoir within a mean field approximation .
the field operator @xmath9 corresponds to atoms with spin @xmath10 . here
@xmath11 with @xmath12 the chemical potential of these atoms , and @xmath13 is the harmonic potential with frequency @xmath14 .
the superfluid energy gap @xmath15 is taken to be real and positive .
the hamiltonian of the raman - coupled atoms is @xmath16 with field operator @xmath17 , and @xmath18 where @xmath19 is the rf detuning .
the raman coupling is described by @xmath20 where @xmath21 is the two - photon rabi rate and @xmath22 is the recoil momentum .
the radio and microwave driven tunneling between the superfluid and raman states are characterized by @xmath23 with tunneling strength @xmath6 , proportional to the intensity of these fields . to inelastic scattering rate @xmath24 as a function of magnetic field for transition between the third and fourth lowest hyperfine states of @xmath0 @xcite.,scaledwidth=40.0% ] the full hamiltonian can be written in a matrix form @xmath25 \left[\begin{array}{cc } \mathcal{h}_{\rm s } & \mathcal{h}_{\rm t } \\
\mathcal{h}_{\rm t } & \mathcal{h}_{\rm r } \end{array}\right ] \left[\begin{array}{c } \psi_{\rm s}\\ \psi_{\rm r } \end{array}\right],\\ = & & \int dx \psi^\dagger\mathcal{h}\psi\end{aligned}\ ] ] where @xmath26^t$ ] and @xmath27^t$ ] are length 4 vectors of field operators , and the equality defines @xmath28 and @xmath29 .
the matrix hamiltonian density is made of @xmath30,\nonumber\ ] ] @xmath31,\nonumber\ ] ] @xmath32.\nonumber\ ] ] here @xmath33 .
we expand the field operator in a complete basis @xmath34 for clarity , in eq .
( [ eq : psi ] ) we have added hats to all operators and explicitly show all @xmath35 dependence .
( for notational simplicity , these typographic clues are left out at our other equations . )
the index @xmath36 runs from 1 to 8 , and @xmath37 is summed over all eigenstates .
here @xmath38 is the excitation operator with energy @xmath39 : @xmath40 the energies are found from the eigenvalue problem , @xmath41 where @xmath42 is a 8@xmath431 matrix with the @xmath36-th row element @xmath44 . in this particle - hole symmetric representation of the system , we have @xmath45 , and each mode effectively appears twice in eq .
( [ eq : eng ] ) .
the majorana states corresponds to the zero energy states , i.e. , @xmath46 .
we numerically solve the bogoliubov - de gennes equation , eq .
( [ eq : bdg ] ) .
the local density of states of superfluid and raman coupled particles then can be written as @xmath47 where the summation is over all eigenstates . .
energies @xmath48 are measured in terms of the recoil energy , @xmath49 where @xmath22 is the wave - vector of the raman lasers . in this calculation
we choose @xmath50 , @xmath51 .
the chemical potential is set to be @xmath52 and @xmath53 .
the harmonic potential @xmath13 has @xmath54 .
the horizontal dashed lines show @xmath55 .
( a ) topologically trivial state without tunneling .
( b ) topologically non - trivial state when zero energy modes appear on the edges .
( c ) transition from non - trivial to trivial state when edge modes begin to disappear .
( d ) topologically trivial state at large @xmath6.,scaledwidth=48.0% ]
our results are summarized by the local density of states shown in fig .
[ fig : fig3 ] . in fig .
[ fig : fig3 ] ( b ) we clearly see zero energy modes near the edge of the cloud .
these are majorana modes . at stronger coupling @xmath6
, this feature disappears . below we will present an argument for when majorana modes will be present .
first we rewrite the hamiltonian density in momentum space @xmath56 , \label{eq : hk}\ ] ] where @xmath57 , @xmath58 , and @xmath59 .
the eigenstates of eq .
( [ eq : hk ] ) define a mapping from @xmath60 [ the values of @xmath61 to @xmath62 [ 8-dimensional complex space ] .
as explicitly explored in @xcite , these mappings can be divided into topological distinct equivalence classes .
as one varies the parameters @xmath15 , @xmath63 , @xmath6 , etc . , the topological classification of the bands do not change , unless two bands touch .
the locations of the majorana modes in the trap are roughly the points where the spatially changing @xmath64 leads to exactly such a band crossing . by symmetry
, we expect the band crossings to occur at @xmath65 , and they can be found by setting @xmath66 .
this yields the simple condition @xmath67 where @xmath68 and @xmath69 .
figure [ fig : fig4 ] shows the phase diagram of the homogeneous system , resulting from this argument . further treating @xmath70 as the chemical potential at the center of the trap
, there will be a majorana mode somewhere in the cloud if the system is locally topologically non - trivial at the center .
the plot of @xmath71 in fig .
[ fig : fig4 ] separates the phase diagram into two phases : i. topologically non - trivial phase and ii .
topologically trivial phase .
the boundary is a non - monotonic function of @xmath6 , however at large enough @xmath6 , the state is always topologically trivial .
the line @xmath72 also corresponds to a band crossing , but if one reflects fig .
[ fig : fig4 ] across the vertical axis , the picture remain the same . and @xmath73 .
the diagram can be separated into two phases : i. topologically non - trivial state which has zero energy modes at the edge and ii .
topologically trivial state where no majorana modes should be found .
the curve shows where @xmath74.,scaledwidth=40.0% ] treating the @xmath75 and @xmath76 in eq .
( [ eq : phase ] ) as spatially dependent : @xmath11 , @xmath18 , the relationship in eq .
( [ eq : phase ] ) gives the rough location of the majorana mode . in fig .
[ fig : fig5 ] , we plot this location as a function of @xmath6 , for fixed @xmath50 , @xmath51 , @xmath52 , @xmath53 , and the harmonic trap @xmath13 with @xmath54 .
we can see that the system is always topologically trivial at the edge .
the center is trivial at large @xmath6 and non - trivial at small @xmath6 .
typically one goes from non - trivial to trivial as one moves from the center to the edge . for these parameters ,
the entire cloud is topologically trivial when @xmath77 .
we also plot the locations and widths of the majorana modes from our numerical simulations in fig .
[ fig : fig3 ] .
we can see that this simple argument gives a good approximation to the location of the majorana modes , even for the small system we are considering .
is the recoil energy with @xmath22 the recoil momentum from the raman laser , and @xmath6 is the strength of coupling between the superfluid and raman coupled states ( see fig .
[ fig : fig1 ] ) .
parameters are chosen as in fig .
[ fig : fig3 ] . the solution of eq .
( [ eq : phase ] ) ( the curve ) separates the diagram into two regimes : i. locally topologically non - trivial phase and ii . locally topologically trivial phase .
the edge of the trapped cloud is always trivial . within a local density approximation ( lda )
majorana edge modes appear at the spatial boundary between these phases , and will be found if @xmath78 . the points with error bars " show the location and width of the majorana mode in numerical simulations using the parameters in fig .
[ fig : fig3 ] .
the discrepancy between the numerical calculations and lda becomes smaller as the number of particles is increased.,scaledwidth=38.0% ]
one reason for introducing this model was to control heating . only the raman - coupled atoms experience heating from off - resonant light scattering .
the energy absorbed from one scattering event is roughly the recoil energy @xmath79 .
the number of scattering events per unit time is @xmath80 , where the single particle inelastic scattering rate is @xmath24 .
as shown by wei and mueller , the inelastic scattering rate is proportional to the raman coupling strength @xcite , and as seen in fig .
[ fig : fig2 ] , for our states @xmath81 .
a typical raman strength is @xmath82 , yielding an absorbed power of @xmath83 . due to atomic collisions
this energy gets redistributed among all the atoms .
superfluidity will be destroyed when the energy per atom @xmath84 is of order @xmath85 .
taking @xmath86 , @xmath52 , and the harmonic potential @xmath54 , we have @xmath87 , @xmath88 , yields @xmath89 , and the system should remain superfluid for a time of order @xmath90 .
by contrast , if all atoms experienced the raman coupling , the lifetime would be @xmath91 .
we must also consider inelastic processes coming from atomic collisions .
for example , an atom in state 4 would collide with one in state 1 , flipping into state 2 .
this process is allowed because it conserves @xmath92 .
the energy from such a spin - flip is sufficiently large that the atom would be lost from the trap .
considering our states , there are three such allowed exothermic processes : @xmath93 , @xmath94 , and @xmath95 .
the rate of atom loss is parameterized by the two - body coefficients @xmath96 , appearing in the rate equation @xmath97 where @xmath98 is the average density , and @xmath99 is the particle number . in fig .
[ fig : fig6 ] we plot the exchange loss rate coefficient @xmath96 as a function of magnetic field using asymptotic boundary condition approximation @xcite . near the feshbach region @xmath100 , the two - body loss rate coefficient goes to @xmath101 .
thus for a system with density @xmath102 , the @xmath103 lifetime is about @xmath104 . as a function of magnetic field @xmath105 using the formalism from @xcite .
the four states are labeled 1 to 4 from the lowest to the forth lowest energy state .
the solid line is collision from state @xmath93 , the dashed one is @xmath94 , and the dotted one is @xmath106.,scaledwidth=45.0% ] we must also consider three - body losses , which are enhanced by the large mutual interaction between the lowest three hyperfine states near feahbach resonance at @xmath100 .
experiment observations show that @xmath107 with the three - body loss rate coefficient is @xmath108 @xcite .
thus for a low density system with density @xmath109 , the @xmath103 lifetime is about @xmath110 .
in summary , we have devised an improved protocol to produce topological edge modes in raman - induced spin - orbit coupled fermions .
we divide the atoms into two populations , distinguished by their hyperfine states , yet spatially overlapping .
the majority of the atoms experience strong interactions , while a minority experiences a raman - induced spin - orbit coupling .
radio and microwaves mix the various states . by calculating the spatially resolved single particle density of states , we showed that this system has majorana edge modes .
we estimated heating rates from off - resonant light scattering , finding superfluid lifetimes of order of @xmath111 for @xmath0 .
we also estimate the two and three - body loss rate in the system , finding that the timescale is limited by the three - body losses : for a density of @xmath109 , the @xmath103 lifetime is @xmath110 .
we acknowledge useful comments from m. w. zwierlein .
j.x . would like to thank r. wei for helpful discussions and providing the data shown in fig .
[ fig : fig2 ] .
this research is supported by the national science foundation ( phy-1068165 ) , the national key basic research program of china ( grant no .
2013cb922000 ) , the national natural science foundation of china ( grant no .
11074021 ) , and the army research office with funds from the darpa ole program .
is also supported by china scholarship council . | we show how to exploit the rich hyperfine structure of fermionic alkali - metal atoms to produce a quasi-1d topological superfluid while avoiding excessive heating from off - resonant scattering .
we model interacting fermions where four hyperfine states are coupled by a variety of optical and microwave fields .
we calculate the local density of states in a trap , finding regimes with zero energy topological edge modes .
heating rates in this system are significantly suppressed compared to simple raman - induced spin - orbit coupling approaches .
we also estimate the two- and three - body decay rates and find a reasonable lifetime at small , but experimentally relevant densities . |
an important class of instabilities of one - dimensional patterns are side - band instabilities .
the most important one among them is the eckhaus instability @xcite which is a monotonic long - wave instability that renders the pattern unstable with respect to infinitesimal compressions and dilatations .
its onset can be characterized by the vanishing phase diffusion coefficient @xcite . near the threshold , where the pattern can be described by a real ( dissipative ) ginzburg landau ( gl ) equation ,
the eckhaus instability arises generically if the deviation @xmath0 of the wave - number of the pattern from that corresponding to the bifurcation point exceeds a certain value , i.e. if @xmath1 where @xmath2 is the deviation of the control parameter from the threshold . a similar instability of wave patterns ( e.g. described by the complex gl equation ) is usually referred to as the benjamin - feir instability @xcite .
typically , the eckhaus instability restricts the range of allowed wave numbers of the pattern , i.e. its nonlinear evolution eventually brings the system back to the stable range of wave numbers .
the eckhaus instability of periodic patterns has been studied extensively , both theoretically and experimentally , in various pattern - forming systems , such as rayleigh - bnard convection @xcite , convection in binary mixtures @xcite , taylor vortex flow @xcite , electrohydrodynamic convection in nematics @xcite , directional solidification @xcite . even quite close to the threshold
, the boundaries of the eckhaus instability of stationary periodic patterns can differ from ( [ ech ] ) due to various factors , such as the system confinement @xcite , subcritical character of the instability @xcite , the shape of the neutral stability curve @xcite , the effect of nonlinearity @xcite and resonances @xcite , the presence of random fluctuations @xcite , chiral symmetry breaking ( e.g.due to rotation ) @xcite , etc .
although the eckhaus instability is the most common one - dimensional side - band instability , others are possible as well . in the present paper we show that spatially periodic patterns arising due to marangoni convection in a liquid layer with a deformable free surface heated from below can possess additional side - band instabilities : a short - scale steady side - band instability and a long - scale oscillatory side - band instability .
an important feature of this system is the presence of _ two _ primary modes of instability ( generating the marangoni convection itself ) : a long - scale instability ( with zero wave - number ) connected with the deformation of the free surface , and short - scale convection ( with a finite wave - number ) governed by surface tension gradients .
a typical neutral stability curve for this system has two minima , corresponding to _ two _ critical marangoni numbers @xcite .
the one minimum , ma@xmath3 , is at a zero wave - number and corresponds to a long - scale instability driven by deformation of the free surface coupled with surface tension gradients .
the other minimum , ma@xmath4 , is at a non - zero wave - number , and corresponds to a short - scale instability caused by surface tension gradients only ( surface deformations are not important in this regime ) .
when the two thresholds are close to each other , the nonlinear interaction between the two modes of marangoni instability with different scales can lead to various secondary instabilities of the short - scale convection generating various stationary and temporally modulated patterns @xcite . the nonlinear evolution and interaction of the two modes
is described by a system of coupled equations which can be written in the following scaled form @xcite : @xmath5 here @xmath6 is the amplitude of the short - scale convection , @xmath7 is the deformation of the free surface , @xmath8 and @xmath9 are long - scale space coordinate and slow time .
the form of the equation for @xmath7 is determined by the fact that @xmath7 is conserved due to mass conservation . in the non - conserved case
certain aspects of the interaction between short- and long - scale structures has been discussed in @xcite .
the parameter @xmath10 is proportional to the difference between the two thresholds , @xmath11 characterizes the effect of the laplace pressure , and @xmath12 is the computable coupling parameter which was shown to be positive @xcite .
note that for the terms in the second equation in ( [ ah ] ) to be of the same order , the capillary number , which parameter @xmath13 is proportional to , has to be large ( which is indeed the case in most of experimental conditions ) .
the coupling between @xmath6 and @xmath7 reflects the following mechanism of the modes interaction : as expressed by the coupling term in the first equation , the short - scale convection is more intensive beneath surface elevations ; since @xmath12 , the convection , in turn , suppresses the long - scale deformational mode by changing the average surface temperature or concentration @xcite .
this stabilization effect of the short - scale convection mode allows the difference between the critical marangoni numbers ma@xmath3 and ma@xmath4 to be @xmath14 and still the deformational instability to be restricted to long scales and have small growth rate
. however , this stabilization fails if ma@xmath15ma@xmath3 is too large ( see @xcite ) . in ref.@xcite
, the effect of the mode interaction on the stability of the short - scale convection was studied only in the vicinity of the band center ( @xmath16 ) . in the present paper
we extend the analysis to the full wave - number band of the short - scale convection and show that the interaction of the two modes strongly affects the width of the band and can even lead to a change of the type of the instability .
we consider the effect of surface deformations on the stability of short - scale convection rolls with wave - numbers lying within an @xmath17 range at small supercriticality ( ma@xmath18ma@xmath4)/ma@xmath19 .
this convection pattern is described by the following stationary solution of eq.([ah ] ) : @xmath20 where @xmath21 is the dimensionless ( in units of the characteristic width @xmath22 of the band of excited wave - numbers ) deviation from the critical wave - number corresponding to the threshold ma = ma@xmath4 of the short - scale instability .
consider a perturbed solution in the form @xmath23 where asterisks denote complex conjugates . substituting ( [ ptrb ] ) in system ( [ ah ] ) yields the following implicit dispersion relation for the side - band instability : @xmath24 where @xmath25 we see that the interaction of the two modes leads to a _
equation for the eigenvalues defining the limits of the side - band instability ( similar , say , to the case of the eckhaus instability of a hexagonal pattern when there is a coupling between three waves @xcite ) . in our case
, we have the coupling between two side - band modes , which correspond to longitudinal compression - dilatation waves propagating in opposite directions if @xmath26 is complex , and the surface deformation .
according to the routh - hurwitz criterion , it follows from ( [ disp ] ) that the stationary solution ( [ aq ] ) is stable if @xmath27 the analysis of conditions ( [ cond ] ) shows that , depending on the values of the parameters @xmath28 , the side - band instability in this system can be either long - wave or short - wave , monotonic or oscillatory , and the range of @xmath0 where the solution ( [ aq ] ) is stable can be either wider or narrower than the standard eckhaus interval defined by ( [ ech ] ) .
a detailed analysis of various possibilities is very complicated since the stability of the system is governed by four parameters , @xmath29 .
we shall dwell therefore mostly on the case when the solution ( [ aq ] ) becomes unstable with respect to perturbations with infinitesimal wave - number .
the corresponding stability limits can be determined from ( [ cond ] ) in the limit @xmath30 .
it is , however , more instructive to obtain this result using the so - called _ phase approximation _ which is based on the observation that in the long - wave limit the evolution of the amplitude of the convection pattern is slaved to the evolution of its phase @xcite .
thus , we set @xmath31 , and obtain from eq.([ah ] ) for the amplitude correction @xmath32 @xmath33 in our case , the amplitude near the instability threshold is slaved to both the phase @xmath34 and the surface deformation @xmath7 .
the evolution of these two variables is described by the following system of coupled equations : @xmath35 introducing the wave - number deviation , @xmath36 , one can rewrite system ( [ fih ] ) in the form of the coupled diffusion equations : @xmath37 where @xmath38 the system is stable if all eigenvalues of matrix @xmath39 have negative real parts .
the characteristic equation leads to the following stability conditions : @xmath40 it should be noted that this system obtained in the long - wave limit does not give the complete stability information .
it can be shown that there always exists an interval of @xmath10 where the relevant instability is a _ short - wave _ instability that can be either monotonic or oscillatory .
we shall dwell on the case when the instability is long - wave in the band center , and no short - wave _ oscillatory _ instability arises .
a typical stability diagram is shown in fig.1 for @xmath41 .
fig.1 is true provided @xmath42 ( see ref.@xcite ) and @xmath13 is larger than a certain value @xmath43 , respectively ( for @xmath44 .
fig.1 presents the region in the plane @xmath45 in which the solution ( [ aq ] ) is stable .
this region is bounded by the lines ab , bc , cd and de . the lines ab and de are the loci of @xmath46 and correspond to a _
monotonic _ long - wave instability ( cf .
( [ stab2 ] ) ) , while the line cd is the locus of @xmath47 and corresponds to an _ oscillatory _ long - wave instability ( cf .
( [ stab1 ] ) ) .
the line bc corresponds to a monotonic _ short - wave _ instability and can be found analytically from the condition @xmath48 ( see eq.([cond ] ) ) ( it is , however , too cumbersome to be reproduced here ) .
the horizontal dashed line corresponds to the threshold of the standard eckhaus instability , @xmath49 .
the coordinates of the points a , b , c , d , and e can be found analytically : a@xmath50 , b@xmath51 , d@xmath52 , e@xmath53 ; the expressions for the coordinates of the point c are very lengthy ( for @xmath41 in fig.1 , c@xmath54 ) .
thus , due to the coupling between the phase evolution and the surface deformation , both the interval of wave - numbers corresponding to stable stationary convection described by eq.([aq ] ) and the type of its instability depend on the parameter @xmath10 , i.e. on the difference between the two thresholds of the marangoni instability , ma@xmath15ma@xmath3 .
there are four particular values of @xmath55 , corresponding to the points b , c , e , d in fig.1 , respectively , which define four regions where qualitatively different instabilities are observed .
these regions in the @xmath56-parameter plane are shown in fig.2 for @xmath57 .
if @xmath58 , solution ( [ aq ] ) is stable when @xmath59 .
since @xmath12 in our case , the stability interval for @xmath0 is _ larger _ than in the case of the ordinary eckhaus instability , @xmath60 ; part of the energy of phase perturbations is transferred to the damped long - scale deformational mode .
this suppresses the eckhaus instability and makes the stability interval for @xmath0 larger .
outside this interval of @xmath0 , solution ( [ aq ] ) is unstable and the instability is _ monotonic _ and _ long - wave _ , like in the standard case with no surface deformation . in the limit @xmath61 , surface deformations are damped very strongly , and the eckhaus interval tends to its ordinary value .
when @xmath62 , the stability threshold for @xmath0 is determined by a cumbersome function of @xmath13 and @xmath63 which we do not represent here .
this threshold is still larger than @xmath64 .
the instability in this case is also monotonic , but _ short - wave_. at the point b the modulation wave number @xmath65 goes to zero and the short - wave instability continuously changes over to a long - wave instability .
this shows that the short - wave instability is not due to a subharmonic resonance as it has been found , e.g. , in taylor vortex flow @xcite .
when @xmath66 , the stability interval for @xmath0 is given by @xmath67 .
the instability in this case is _ oscillatory _ and _ long - wave _ and can lead to slow oscillations of the wave - number of the short - scale roll pattern coupled with oscillations of the surface deformation ( see sec.3.2 below ) . at @xmath68 ,
the stability region coincides with the ordinary eckhaus interval , @xmath60 .
if @xmath69 , the roll pattern corresponding to the band center , @xmath70 , becomes unstable .
this coincides with the results in @xcite . nevertheless , as long as @xmath71 , a _
stable _ pattern _ still exists _ under these conditions , but with a _ modified _ wave - number lying in the interval @xmath72 .
the lower boundary of this interval corresponds to the long - wave _ monotonic _ instability ( i.e. the usual eckhaus instability ) while the upper one to the long - wave _ oscillatory _ instability .
this splitting of the stable band into two sub - bands is reminiscent of a situation encountered in parametrically excited standing waves : under certain conditions the wave numbers in the center of the band become eckhaus - unstable while off - center wave numbers remain stable .
such a splitting was found to lead to the formation of stable kinks in the wave number separating domains with different wave numbers @xcite .
the nonlinear evolution of the present system in this regime will be discussed briefly below .
for @xmath73 , solution ( [ aq ] ) is unstable at any wave number due to the deformational instability of the surface .
thus , depending on @xmath10 , i.e. on the difference between the threshold ma@xmath4 of the short - scale convection and that of the long - scale deformational instability ( ma@xmath3 ) , the band of stable wave numbers can take on two qualitatively different forms .
for @xmath74 the band is contiguous ( hashed area in fig.3a ) while for @xmath75 convection is stable within two disconnected wave - number bands ( hashed area in fig.3b ) .
in conclusion of this section let us note that if the parameter @xmath13 is _ not _ large enough , the _ oscillatory short - wave _ side - band instability comes into play . in this case ,
the stability region shown in fig.1 slightly changes : there appears a new boundary corresponding to this short - wave oscillatory side - band instability which cuts off the corner at the point c and connects two boundaries bc and cd corresponding to the short - wave monotonic and to the long - scale oscillatory side - band instabilities , respectively .
in this section we shall study the nonlinear evolution of the long - wave instability near the thresholds given by the curves ab , cd and de in fig.1 .
the monotonic long - wave instability threshold corresponds to the curves ab and de in fig.1 and is given by @xmath76 we use the following scaling : @xmath77 , and apply the following ansatz : @xmath78 substituting ( [ arofi ] ) in system ( [ ah ] ) , one obtains in the first order that the change in the local wave - number , @xmath79 , is proportional to the surface deformation @xmath80 , @xmath81 proceeding to the sixth order , we obtain the following nonlinear evolution equation for the surface deformation @xmath80 : @xmath82 where @xmath83 the evolution equation for the long - wave modulation of the wave - number @xmath84 can be obtained from eq.([evh ] ) by a simple renormalization of the coefficient @xmath85 ( cf .
( [ k1h1 m ] ) ) . when @xmath61 , the surface is undeformable , and the coefficients in the equation for @xmath84 tend to those of the nonlinear evolution equation describing an ordinary eckhaus instability in the real gl equation .
it can be shown that the coefficients @xmath86 and @xmath87 are positive ( in the parameter region where the instability is long - wave ) , and the instability is subcritical .
therefore , all periodic solutions of ( [ evh ] ) blow up beyond the stability limit @xmath88 indicating a break - down of ( [ evh ] ) .
the same happens in the case of the _ monotonic short - scale _ instability , corresponding to the curve bc in fig.1 ; our computations show that it is also subcritical .
however , numerical simulations of the full amplitude equations ( [ ah ] ) show that different physical processes occur when crossing the lines ac and de in fig.1 , respectively .
this is related to the fact that along ac the deformational mode is linearly damped ( @xmath74 ) , whereas along de it is excited ( @xmath89 ) .
figs.4,5 show typical results of numerical simulations of ( [ ah ] ) .
fig.4 shows the evolution of the pattern occurring when the line ac is crossed .
while the surface deformation and the amplitude of the short - scale convection remain regular for all times , the wave - number goes through a true singularity .
it signifies a phase slip , which reduces the total phase in the system by @xmath90 . following this phase slip
, the pattern relaxes to a periodic pattern with a wave - number in the stable band .
the evolution below the line de is shown in fig.5 . here
the deformational mode is excited ; it has been shown previously that the instability is subcritical at @xmath70 @xcite .
the simulation of the coupled amplitude equations ( [ ah ] ) indicates that the instability does indeed not saturate which results in a singularity in the surface deformation and in the amplitude of the short - scale convection .
thus , even the amplitude equations ( [ ah ] ) become invalid .
this result suggests a dry - out of the fluid film as has been observed experimentally @xcite . despite the splitting of the wave - number band
, no stable wave - number kinks could be found .
this indicates that the higher - order contributions to the phase equation ( [ evh ] ) , which were not calculated , differ from those arising solely from a non - monotonicity of the phase diffusion coefficient @xcite .
it should be noted that at the point d the coefficients @xmath91 and @xmath87 tend to infinity , while at the point e the coefficient @xmath92 vanishes regardless of the value of @xmath93 ; the system dynamics in the vicinity of these points is more complex ; it can not be described by eq.([evh ] ) and requires separate analysis .
along line cd in fig.1 an oscillatory long - wave instability arises .
its threshold is given by @xmath94 to study its nonlinear behavior we set @xmath95 , introduce the slow coordinates @xmath96 , @xmath97 , and expand @xmath98 with @xmath99 substituting ( [ expo ] ) in system ( [ ah ] ) , we obtain in the first order a linear system for @xmath80 and @xmath100 which , after appropriate diagonalization , can be reduced to a linear free schrdinger equation @xmath101 where @xmath102 eq.([schr ] ) describes waves propagating along the convection pattern and modulating its wave - number and the surface deformation .
the phase velocity of the waves @xmath103 depends on the wave number @xmath65 of the modulation , as @xmath104 .
the wave - number @xmath105 and the surface deformation @xmath80 are proportional to each other @xmath106 and there is a phase shift between oscillations of the wave - number and the surface deformation equal to arg@xmath107 . at the onset of the oscillatory long - wave instability
, a whole band of wave - numbers @xmath108 is excited .
however , a strong ( quadratic ) dispersion of the waves described by ( [ schr ] ) allows only quartic interactions between waves with different wave - numbers within the excited interval .
the evolution of the wave amplitudes can thus be described in the third order of the perturbation theory by a system of landau equations .
a similar situation occured in a study of long - scale oscillatory double - diffusive convection @xcite . in order to study the wave interaction , we consider two pairs of oppositely propagating waves having wave - numbers @xmath109 and @xmath110 , and present @xmath105 and @xmath80 in the following form : @xmath111+c.c.,\label{ab}\\ \theta^{\pm}_{1,2}=\omega_{0}\kappa_{1,2}^{2}t_{0}\pm\kappa_{1,2}x , \nonumber \end{aligned}\ ] ] where the amplitudes @xmath112 and @xmath113 undergo slow evolution on the time scale @xmath114 . using ( [ ab ] ) and
( [ expo ] ) , we obtain from ( [ ah ] ) , as solvability conditions in the fourth and in the fifth orders , the following system of landau equations for the amplitudes @xmath112 and @xmath113 : @xmath115 the linear growth rate coefficients @xmath116 and the landau coefficients @xmath117 have the following form : @xmath118 where @xmath119 if @xmath120 .
the landau coefficients @xmath121 are complicated functions of @xmath122 satisfying the relationships @xmath123 the real parts @xmath124 of the functions @xmath125 in eq.([gamlam ] ) are shown in fig.6 as functions of @xmath10 .
they become unbounded in the vicinity of the point d ( @xmath126 ) where system ( [ land ] ) is not valid . system ( [ land ] ) allows to study the competition between travelling and standing waves .
let us consider two waves with a fixed wave - number propagating in opposite directions .
setting @xmath127 in ( [ land ] ) , we study the stability of the two solutions of system ( [ land ] ) @xmath128 with respect to infinitesimal perturbations of the amplitudes @xmath129 and @xmath130 .
the symbols of the landau coefficients in ( [ tw ] ) , ( [ sw ] ) and below denote now only the _ real parts _ of the respective coefficients .
the stability analysis shows that a)_travelling waves _ are stable if @xmath131 b)_standing waves _ are stable if @xmath132 otherwise , the system undergoes a subcritical instability which may lead to a blow - up and may return the system back to the stable region .
we have performed some numerical simulations of ( [ ah ] ) which support this expectation .
we have found ( see fig.6 ) that in our case ( @xmath133 ) , when @xmath134 standing waves are stable , and when @xmath135 traveling waves are stable . in the rest intervals of @xmath10 within the interval @xmath120 ,
close to the points c and d , both types of waves are subcritically unstable .
let us now study the stability of standing and travelling waves with respect to perturbations with different wave - numbers , i.e. the stability of solutions ( [ tw ] ) , ( [ sw ] ) with respect to perturbations of the amplitudes @xmath136 and @xmath137 .
the stability analysis shows that \a ) _ travelling waves _ are stable if , besides ( [ t ] ) , the following conditions are fulfilled : @xmath138
\b ) _ standing waves _ are stable if , besides ( [ s ] ) , the following condition holds : @xmath139 the analysis of conditions ( [ t1 ] ) and ( [ s1 ] ) shows that within the intervals defined by ( [ t ] ) and ( [ s ] ) there exist intervals of wave - numbers where the waves are also stable to perturbations with an arbitrary wave - number .
fig.7 shows how these intervals change with the variation of the parameter @xmath10 along curve cd in fig.1 .
it can be seen that standing waves are stable at intermediate wave - numbers , while travelling waves can be stable at infinitesimal wave - number as well .
thus , the oscillatory long - wave eckhaus instability in marangoni convection with deformable interface can generate either standing or travelling compression - dilatation waves propagating along the periodic pattern and modulating its wavelength together with the deformation of the free surface .
typical results of numerical simulations of ( [ ah ] ) are given in fig.8a , b .
it shows space - time diagrams for typical standing waves and traveling waves , respectively .
this research was supported by the israel science foundation .
a.a.g . acknowledges the support of the ministry for immigrant absorption .
gratefully acknowledges a travel grant by the minerva center and support by doe through de - fg02 - 92er14303 .
a.a.golovin , a.a.nepomnyashchy and l.m.pismen , `` interaction between short - scale marangoni convection and long - scale deformational instability , '' phys .
fluids a * 6 * , 34 ( 1994 ) .
w.eckhaus , _ studies in nonlinear stability theory _
( springer , new york , 1965 ) . l.kramer and w.zimmermann ,
on the eckhaus instability for spatially periodic patterns , " physica d * 16 * , 221 ( 1985 ) .
b.janiaud , a.pumir , d.bensimon , v.croquette , h.richter and l.kramer , `` the eckhaus instability for travelling waves , '' physica d * 55 * , 269 ( 1992 ) . m.m.chen and j.a.whitehead , `` evolution of two - dimensional periodic rayleigh convection cells of arbitrary wave - numbers , '' j. fluid
. mech . * 31 * , 1 ( 1968 ) .
r.m.clever and f.h.busse , `` transition to time dependent convection , '' j. fluid
. mech . * 65 * , 625 ( 1974 ) .
p.kolodner , `` observations of the eckhaus instability in one - dimensional travelling - wave convection , '' phys
a * 46 * , 1739 ( 1992 ) .
g.ahlers , d.s.cannell , m.a.dominguez-lerma and r.heinrichs , `` wave - number selection and eckhaus instability in couette - taylor flow , '' physica d * 23 * , 202 ( 1986 ) . h. riecke and h .-
paap , `` stability and wave - vector restriction of axisymmetric taylor vortex flow , '' phys .
rev . a * 33 * , 547 ( 1986 ) .
m.lowe and j.p.gollub , `` pattern selection near the onset of convection : the eckhaus instability , '' phys .
lett . * 55 * , 2575 ( 1986 ) . a.ghazali and c.misbah , `` phase instability and local dynamics in directional solidification , '' phys .
a * 46 * , 5026 ( 1992 ) .
l.s.tuckerman and d.barkley , `` bifurcation analysis of the eckhaus instability , '' physica d * 46 * , 57 ( 1990 ) .
h.r.brand and r.j.deissler , `` eckhaus and benjamin - feir instabilities near a weakly inverted bifurcation , '' phys .
a * 45 * , 3732 ( 1992 ) .
d.s.riley and s.h.davis , `` eckhaus instabilities in generalized landau - ginzburg equations , '' phys . fluids a * 1 * , 1745 ( 1989 ) .
d. raitt and h. riecke , `` domain structures in fourth - order phase and ginzburg - landau equations , '' physica d * 82 * , 79 ( 1995 ) .
b.a.malomed , i.e.staroselsky and a.b.konstantinov , `` corrections to the eckhaus stability criterion for one - dimensional stationary structures , '' physica d * 34 * , 270 ( 1989 ) .
e.hernandez-garcia , j.vinals , r.toral and m.san miguel , `` fluctuations and pattern selection near an eckhaus instability , '' phys .
70 * , 3576 ( 1993 ) .
a.c.or , `` new phenomena in the eckhaus instability of thermal rossby waves , '' j. fluid
* 216 * 613 ( 1990 ) .
m.takashima , `` surface tension driven instability in a horizontal liquid layer with a deformable free surface , '' j. phys .
soc . jpn . * 50 * , 2745 ( 1981 ) .
d.kazhdan , l.shtilman , a.a.golovin and l.m.pismen , `` nonlinear waves and turbulence in marangoni convection '' , phys .
fluids , ( 1995 ) to appear .
p. metzener and m.r.e .
proctor , `` interaction of patterns with disparate scales , '' eur .
b * 11 * , 759 ( 1992 ) .
g.z.gershuni , e.m.zhukhovitsky and a.a.nepomnyashchy , _ stability of convective flows _ , nauka , 1989 ( in russian ) .
m.m.sushchik and l.s.tsimring , `` the eckhaus instability in hexagonal patterns , '' physica d * 74 * , 90 ( 1994 ) . m.c.cross and a.newell , `` convection patterns in large aspect ratio systems , '' physica d * 10 * , 299 ( 1984 ) . m.c.cross and p.c.hohenberg ,
`` pattern formation outside of equilibrium , '' rev .
phys . * 65 * , 851 ( 1993 ) .
paap and h. riecke , `` wave - number restriction and mode interaction in taylor vortex flow : appearance of a short - wavelength instability , '' phys .
a * 41 * , 1943 ( 1990 ) .
h.riecke , `` stable wave - number kinks in parametrically excited standing waves , '' europhys . lett . * 11 * , 213 ( 1990 ) .
vanhook , m.f .
schatz , w.d .
mccormick , j.b .
swift and h.l .
swinney , `` long - wavelength instability in surface - tension - driven bnard convection '' , phys .
( submitted ) .
l.m.pismen , `` selection of long - scale oscillatory convective patterns , '' phys .
a * 38 * , 2564 ( 1988 ) .
* figure 1 .
* the dependence of the wavenumber @xmath0 corresponding to the onset of the various side - band instabilities on the parameter @xmath10 ( @xmath140 ) . different sections of the solid line correspond to different types of the instability : ab long - wave monotonic ( eckhaus ) ; bc short - wave monotonic ; cd long - wave oscillatory ; de long - wave monotonic ( eckhaus ) . the dashed line presents the standard eckhaus instability threshold @xmath49 . + + * figure 2 .
* regions in the @xmath141-parameter plane corresponding to different types of the side - band instability ( at @xmath57 ) : ml monotonic long - wave ( eckhaus ) , ms monotonic short - wave , ol
oscillatory long - wave .
+ + * figure 3 . *
two qualitatively different types of stable wave - number bands for the short - scale marangoni convection near the instability threshold : + a ) @xmath142 : the stable region ( hashed ) is larger than in the standard case ( dashed line ) , and its boundary ( solid line ) can correspond to either a monotonic , long- or short - wave instability , or to an oscillatory long - wave instability ; + b ) @xmath143 : the band center is unstable , and the stable region ( hashed ) is split up into two parts ; its inner boundary corresponds to a monotonic long - wave instability and the outer one to a long - wave oscillatory instability .
the dotted line is the neutral stability curve . + + * figure 4 . * typical space - time diagrams for the evolution of the long - wave monotonic instability along line
ac ( see fig.1 ) as obtained by numerical simulation of ( [ ah ] ) : the surface deformation a ) and the amplitude of the short - scale convection b ) remain regular while the wavenumber passes through a singularity indicating the phase slip .
+ + * figure 5 . * typical space - time diagrams for the evolution of the long - wave monotonic instability along line de ( see fig.1 ) as obtained by numerical simulation of ( [ ah ] ) . here
the surface deformation a ) and the amplitude of the short - scale convection b ) become singular indicating dry - out of the fluid film .
+ + * figure 6 . * the real parts of the interaction coefficients @xmath144 ( curve a ) and @xmath145 ( curve b ) from eq.([gamlam ] ) at @xmath41 , as functions of @xmath10 .
the vertical asymptote corresponds to @xmath146 ( point d in fig.1 ) .
+ + * figure 7 .
* regions of stability of standing and travelling waves with a wavenumber @xmath109 ( lying within the range of linearly excited modes ) with respect to perturbations with a wavenumber @xmath147 ( lying within the same region ) , at different values of the parameter @xmath10 : s stable , u unstable . + + * figure 8 . * a ) typical space - time diagram of a standing wave arising from the long - wave oscillatory side - band instability of the periodic short - scale convection .
+ b ) typical space - time diagram of a traveling wave arising from the long - wave oscillatory side - band instability of the periodic short - scale convection . + both diagrams were obtained by numerical simulation of ( [ ah ] ) . | the stability of marangoni roll convection in a liquid - gas system with deformable interface is studied in the case when there is a nonlinear interaction between two modes of marangoni instability : long - scale surface deformations and short - scale convection . within the framework of a model derived in @xcite
, it is shown that the nonlinear interaction between the two modes substantially changes the width of the band of stable wave numbers of the short - scale convection pattern as well as the type of the instability limiting the band .
depending on the parameters of the system , the instability can be either long- or short - wave , either monotonic or oscillatory .
the stability boundaries strongly differ from the standard ones and sometimes exclude the band center .
the long - wave limit of the side - band instability is studied in detail within the framework of the phase approximation .
it is shown that the monotonic instability is always subcritical , while the long - wave oscillatory instability can be supercritical , leading to the formation of either travelling or standing waves modulating the pattern wavelength and the surface deformation .
= 10000 = 6.25 in = 9.5 in |
the phenotypic diversity of a species impacts its ability to evolve . in particular ,
the importance of the variance of the population along a phenotypic trait is illustrated by the _
fundamental theorem of natural selection _
@xcite , and the _ breeder s equation _
@xcite : the evolution speed of a population along a one dimensional fitness gradient ( or under artificial selection ) is proportional to the variance of the initial population .
recently , the phenotypic variance of populations has also come to light as an important element to describe the evolutionary dynamics of ecosystems ( where many interacting species are considered ) @xcite . over the last decade ,
the thematic of _ evolutionary rescue _ has emerged as an important question @xcite ( see also the seminal work of luria and delbrck @xcite ) , and led to a new interest in the phenotypic distribution of populations , beyond phenotypic variance .
evolutionary rescue is concerned with a population living in an environment that changes suddenly
. the population will survive either if some individuals in the population carry an unusual trait that turns out to be successful in the new environment , or if new mutants able to survive in the new environment appear before the population goes extinct ( see @xcite for a discussion on the relative effect of _ de novo mutations _ and _ standing variance _ in evolutionary rescue ) . in any case , the fate of the population will not be decided by the properties of the bulk of its density , but rather by the properties of the tail of the initial distribution of the populations , close to the favourable traits for the new environment .
a first example of such problem comes from emerging disease @xcite : animal infections sometimes are able to infect humans .
this phenomena , called zoonose , is the source of many human epidemics : hiv , sars , ebola , mers - cov , etc .
a zoonose may happen if a pathogen that reaches a human has the unusual property of being adapted to this new human host .
a second example comes from the emergence of microbes resistant to an antimicrobial drug that is suddenly spread in the environment of the microbe .
this second phenomenon can easily be tested experimentally @xcite , and has major public health implications @xcite .
most papers devoted to the genetic diversity of populations structured by a continuous phenotypic trait describe the properties of mutation - selection equilibria .
it is however also interesting to describe the genetic diversity of population that are not at equilibrium ( _ transient dynamics _ ) : pathogen populations for instance are often in transient situations , either invading a new host , or being eliminated by the immune system .
we refer to @xcite for a review on transient dynamics in ecology . for asexual populations structured by a continuous phenotypic trait ,
several models exist , corresponding to different biological assumptions @xcite .
if the mutations are modeled by a diffusion , the steady populations ( for a model close to , but where mutations are modelled by a laplacian ) are gaussian distributions @xcite .
furthermore , @xcite have considered some transient dynamics for this model . in the model that we will consider ( see ) , the mutations are modelled by a non - local term .
it was shown in @xcite ( see also @xcite ) that mutation - selection equilibria are then cauchy profiles ( under some assumptions ) , and this result has been extended to more general mutation kernels in @xcite , provided that the mutation rate is small enough .
finally , let us notice that the case of sexual population is rather different , since recombinations by themselves can imply that a _ mutation - recombination equilibrium _ exists , even without selection .
we refer to the infinitesimal model @xcite , and to @xcite for some studies on the phenotypic distribution of sexual species in a context close to the one presented here for asexual populations . in this article , we consider a population consisting of individuals structured by a quantitative phenotypic trait @xmath6 ( @xmath7 open interval of @xmath8 containing @xmath9 ) , and denote by @xmath10 its density . here , the trait @xmath11 is fully inherited by the offspring ( if no mutation occurs ) , so that @xmath11 is indeed rather a breeding value than a phenotypic trait ( see @xcite ) .
we assume that the individuals reproduce with a rate @xmath12 , and die at a rate @xmath13 this means that the individuals with trait @xmath14 are those who are best adapted to their environment , and that the fitness decreases like a parabola around this optimal trait ( this is expected in the surroundings of a trait of maximal fitness ) .
it also means that the strength of the competition modeled by the logistic term is identical for all traits . when an individual of trait @xmath15 gives birth , we assume that the offspring will have the trait @xmath11 with probability @xmath16 , and a different trait @xmath17 with probability @xmath18 .
@xmath1 is then the probability that a mutation affects the phenotypic trait of the offspring .
we can now define the growth rate of the population of trait @xmath11 ( that is the difference between the rate of _ births without mutation _ , minus the death rate ) as @xmath19 when a mutation affects the trait of the offspring , we assume that the trait @xmath17 of the mutated offspring is drawn from a law over the set of phenotypes @xmath20 with a density @xmath21 .
the function @xmath22 then satisfies @xmath23 and we assume moreover that @xmath22 is bounded , @xmath24 , with bounded derivative and strictly positive on @xmath7 . the main assumption here
is that the law of the trait of a mutated offspring does not depend of the trait of its parent .
this classical assumption , known as _ house of cards _ is not the most realistic , but it can be justified when the mutation rate is small @xcite ( see also @xcite ) .
all in all , we end up with the following equation : @xmath25 this paper is devoted to the study of the asymptotic behaviour of the solutions of equation when @xmath26 is small and @xmath0 large and it is organized as follows . in the rest of section 1 the main results are quoted , first in an informal way , and then as rigourous statements .
section 2 contains the proof of theorem [ theorem1 ] and its corollary and finally , in section 3 , theorem [ thm : smallt ] is proved .
when we consider the solutions of , two particular profiles naturally appear : * _ a cauchy profile : _ for a given mutation rate @xmath27 small enough , one expects that @xmath28 will converge , as @xmath0 goes to infinity , to the unique steady - state of , wich is the following cauchy profile @xmath29 where @xmath30 is such that @xmath31 .
this steady - state of is the so - called _ mutation - selection equilibrium _ of the _ house of cards
_ model , which has been introduced in @xcite ( we also refer to @xcite for a broader presentation of existing results ) . *
_ a gaussian profile : _ if @xmath32 , the solution of ( [ eqq0 ] ) can be written @xmath33 where @xmath34 , so that a gaussian - like behavior ( with respect to @xmath11 ) naturally appears in this case
. surprisingly , we are not aware of any reference to this property in the population genetics literature .
we will show that , as suggested by the above arguments , we can describe the phenotypic distribution of the population , that is @xmath35 , when either @xmath36 ( large time for a given mutation rate @xmath37 ) , or @xmath38 ( small mutation rate , for a given time interval @xmath39 $ ] ) . before providing the precise statements of our results ( see subsection [ subsec : rigorous ] ) , we will briefly describe them here , and illustrate them with numerical simulations .
the numerical simulations presented in fig . [ fig1 ] and fig . [ fig2 ] are obtained thanks to a finite difference scheme ( explicit runge - kutta in time ) , and we illustrate our result with a single simulation of with @xmath40 , @xmath41 $ ] , @xmath42 and @xmath43 ( see the definition of @xmath44 in eq .
( [ eqq4 ] ) below ) .
the initial condition corresponds to a population at the mutation - selection equilibrium which environment suddenly changes ( the optimal trait originally in @xmath45 moves to @xmath14 at @xmath46 ) .
this example is guided by the evolutionary rescue experiments described in subsection [ subsec : sel - mut - comp ] , where the sudden change is obtained by the addition of e.g. salt or antibiotic to a bacterial culture .
we describe two phases of the dynamics of the population : * _ large time : cauchy profile .
_ we show that @xmath28 is asymptotically ( when the mutation rate @xmath27 is small ) close to @xmath47 provided @xmath48 .
the population is then a time - independent cauchy distribution for large times .
this theoretical result is coherent with numerical results : we see in fig .
[ fig1 ] that @xmath49 is well described by @xmath50 , as soon as @xmath51 , which is confirmed by the value of @xmath52 for @xmath51 given by fig .
[ fig2 ] . * _ short time : gaussian profile .
_ we also show that @xmath28 is asymptotically ( when the mutation rate @xmath27 is small ) close to @xmath53 provided @xmath54 .
the population has then a gaussian - type distribution for short ( but not too short ) times .
this theoretical result is coherent with numerical results : we see in fig .
[ fig1 ] that @xmath49 is well described by @xmath55 for @xmath56 $ ] , which is confirmed by the value of @xmath52 for @xmath56 $ ] given by fig .
[ fig2 ] . to @xmath57 , of the same simulation of for @xmath58
( see in the text for a complete description ) . in each of these plots , the blue ( resp .
red , black ) line represents @xmath35 ( resp .
@xmath59 , @xmath60 ) .
note that in this figure , the scales of both axis change from one graph to the other , to accommodate with the dynamics of the solution @xmath61.,title="fig : " ] to @xmath57 , of the same simulation of for @xmath58 ( see in the text for a complete description ) . in each of these plots , the blue ( resp .
red , black ) line represents @xmath35 ( resp .
@xmath59 , @xmath60 ) .
note that in this figure , the scales of both axis change from one graph to the other , to accommodate with the dynamics of the solution @xmath61.,title="fig : " ] to @xmath57 , of the same simulation of for @xmath58 ( see in the text for a complete description ) . in each of these plots , the blue ( resp .
red , black ) line represents @xmath35 ( resp .
@xmath59 , @xmath60 ) .
note that in this figure , the scales of both axis change from one graph to the other , to accommodate with the dynamics of the solution @xmath61.,title="fig : " ] to @xmath57 , of the same simulation of for @xmath58 ( see in the text for a complete description ) . in each of these plots , the blue ( resp .
red , black ) line represents @xmath35 ( resp .
@xmath59 , @xmath60 ) .
note that in this figure , the scales of both axis change from one graph to the other , to accommodate with the dynamics of the solution @xmath61.,title="fig : " ] to @xmath57 , of the same simulation of for @xmath58 ( see in the text for a complete description ) . in each of these plots , the blue ( resp .
red , black ) line represents @xmath35 ( resp .
@xmath59 , @xmath60 ) .
note that in this figure , the scales of both axis change from one graph to the other , to accommodate with the dynamics of the solution @xmath61.,title="fig : " ] to @xmath57 , of the same simulation of for @xmath58 ( see in the text for a complete description ) . in each of these plots , the blue ( resp .
red , black ) line represents @xmath35 ( resp .
@xmath59 , @xmath60 ) .
note that in this figure , the scales of both axis change from one graph to the other , to accommodate with the dynamics of the solution @xmath61.,title="fig : " ] to @xmath57 , of the same simulation of for @xmath58 ( see in the text for a complete description ) .
in each of these plots , the blue ( resp .
red , black ) line represents @xmath35 ( resp .
@xmath59 , @xmath60 ) .
note that in this figure , the scales of both axis change from one graph to the other , to accommodate with the dynamics of the solution @xmath61.,title="fig : " ] to @xmath57 , of the same simulation of for @xmath58 ( see in the text for a complete description ) . in each of these plots , the blue ( resp .
red , black ) line represents @xmath35 ( resp .
@xmath59 , @xmath60 ) .
note that in this figure , the scales of both axis change from one graph to the other , to accommodate with the dynamics of the solution @xmath61.,title="fig : " ] ( see in the text for a complete description ) .
the red line represents @xmath62 , while the black line represents @xmath52 . ]
another way to look at these results is to consider @xmath63 and @xmath27 as two parameters , and to see the approximations presented above as approximations of @xmath49 for some set of parameters : @xmath64 for @xmath65 , while @xmath66 for @xmath67 .
we have represented these sets in fig [ fig3 ] .
( in blue ) , where the approximation @xmath64 holds provided that @xmath27 is small enough ; and of the set @xmath68 ( in red ) , where the approximation @xmath66 holds provided that @xmath27 is small enough . ]
as described in the subsection [ subsec : sel - mut - comp ] , the phenotypic distribution of species is involved in many ecological and epidemiological problematics .
our study is a general analysis of this problem and we do not have a particular application in mind .
an interesting and ( to our knowledge ) new feature described by our study is that the tails of the traits distribution in a population can change drastically between `` short times '' , that is @xmath54 and `` large times '' , that is @xmath48 : the distribution is initially close to a gaussian distribution , with small tails , and then converges to a thick tailed cauchy distribution .
this result could have significant consequences in for _ evolutionary rescue _ :
the tails of the distribution then play an important role . quantifying
the effect of this property of the tails of the distributions would however require further work , in particular on the impact of stochasticity ( the number of pathogen is typically large , but finite ) .
the plasticity of the pathogen ( see @xcite ) may also play an important role . here
we state the two main theorems of the paper , each of them followed by a corollary . to do so we start by defining the linear operator @xmath69 and denoting by @xmath70 the dominant eigenvalue of @xmath71 and by @xmath72 the corresponding eigenvector ( see proposition [ proposition1 ] ) .
[ theorem1 ] let us assume that the initial datum @xmath73 is integrable on @xmath7 ( @xmath74a , b[$ ] , @xmath75 ) , and @xmath76 is not identically ( i .- e . a.e . )
@xmath9 .
then the initial value problem for ( [ eqq0 ] ) with @xmath77 has a unique ( global for positive times ) mild solution .
moreover , for @xmath37 small enough , and any @xmath78 , there exists a constant @xmath79 ( depending on @xmath76 and @xmath1 ) such that @xmath80 furthermore , taking @xmath81 , the following more explicit ( in terms of dependence w.r.t @xmath1 ) estimate holds @xmath82 where @xmath83 depend on @xmath76 but not on @xmath1 .
[ cor1 ] under the same hypotheses , there exists positive constants @xmath84 and @xmath85 ( independent of @xmath26 ) such that @xmath86 [ thm : smallt ] let @xmath87 such that @xmath88 .
let @xmath89 satisfying @xmath90 and @xmath91 .
there exists a constant @xmath92 such that the solution @xmath93 of satisfies @xmath94 [ rem : c ] as can be seen from the proof , the constant @xmath95 appearing in indeed only depends on some upper bounds on @xmath96 , @xmath97 and a lower bound on @xmath98 , and on @xmath99 .
in particular , theorem [ thm : smallt ] implies the following description of the population s phenotypic diversity during transitory times , that is times @xmath0 satisfying @xmath100 : [ cor : smallt ] let @xmath87 such that @xmath88 .
let @xmath89 satisfying @xmath90 and @xmath91 .
there exists @xmath92 such that for @xmath101 small enough , as soon as @xmath102 , the solution @xmath93 of satisfies @xmath103,\quad\bigg\|f(t , x)-\frac{f(0,x)\sqrt t e^{-x^2 t}}{f(0,0)\,\int_i e^{-y^2}\,dy}\bigg\|_{l^1}\leq c\kappa.\ ] ] these results hold for models which are slightly more general than equation ( [ eqq0 ] ) .
in fact , in both theorems one can assume that the competition term is a weighted population instead of the total population number . in theorem
[ thm : smallt ] , one could also assume that the mutation kernel depends on the parents trait .
we start here the proof of theorem [ theorem1 ] .
we recall that @xmath104a , b[$ ] , @xmath105 , and @xmath106 is a bounded , @xmath107 function with bounded derivative , such that @xmath108 and @xmath109 .
we begin with the study of the linear operator associated to eq .
( [ eqq0 ] ) .
let us recall that @xmath110 is the operator corresponding to the linear part in eq .
( [ eqq0 ] ) .
it acts on functions of the variable @xmath6 .
we begin with a basic lemma which enables to define the semigroup associated with this operator .
[ lemanou ] the linear operator @xmath71 , defined on @xmath111 and with domain @xmath112 , generates an irreducible positive @xmath113-semigroup ( denoted from now on by @xmath114 ) .
the multiplication linear operator @xmath115 is the generator of a positive @xmath113-semigroup . since @xmath22 is strictly positive
, @xmath116 is a positive bounded perturbation whose only invariant closed ideals are @xmath117 and the whole space @xmath111 .
so @xmath114 is irreducible ( see @xcite , corollary 9.22 ) .
next , we present a proposition which gives information about the spectrum of @xmath71 .
[ proposition1 ] the linear operator @xmath71 has only one eigenvalue .
it is a strictly dominant algebraically simple eigenvalue @xmath118 and a pole of the resolvent , with corresponding normalized positive eigenvector @xmath119 moreover , for @xmath26 small enough , @xmath120 .
the rest of the spectrum of the linear operator @xmath71 is equal to the interval @xmath121.$ ] in the sequel , the norm @xmath122 is the @xmath123 norm on @xmath7 .
let us first show that any @xmath124 belonging to the set @xmath125 belongs to the spectrum of @xmath71 .
in order to do this , for @xmath126 , let us define @xmath127}(x)-\chi_{[x_0-\frac{1}{n},x_0]}(x)\right)$ ] for @xmath128 such that @xmath129\subset i$ ] .
we then have @xmath130 and @xmath131 .
so @xmath132 $ ] is contained in the spectrum of @xmath71 .
the claim follows from the fact that the spectrum is a closed set .
on the other hand , notice that ( for @xmath133 ) , @xmath134 is not an eigenvalue , since the potential corresponding eigenfunction @xmath135 is not an integrable function on @xmath7 ( remember that @xmath22 does not vanish ) .
let us now compute the resolvent operator of @xmath71 , that is , let us try to solve the equation @xmath136 for @xmath137 , defining @xmath138 , ( [ eq3 ] ) gives @xmath139 integrating , we get @xmath140 and @xmath124 belongs to the resolvent set unless the factor of @xmath141 on the left hand side vanishes . therefore @xmath142 .
since for any real number @xmath143 , the function @xmath144 is continuous , strictly decreasing , and satisfies @xmath145 ( recall that @xmath146 ) and @xmath147 , we see that there is a unique real solution of @xmath148 in @xmath149 .
we denote it by @xmath70 .
taking @xmath150 in ( [ eq3 ] ) , we see that @xmath70 is an eigenvalue with corresponding normalized strictly positive eigenvector @xmath151 taking @xmath152 and @xmath153 we see that the left hand side in ( [ eq5 ] ) vanishes , whereas the right hand side is strictly negative , so that @xmath154 has no solution and hence @xmath70 is algebraically simple .
+ indeed , it also follows from ( [ eq5 ] ) that the range of @xmath155 coincides with the kernel of the linear form defined on @xmath111 by the @xmath156 function @xmath157 ( which is the eigenvector corresponding to the eigenvalue @xmath70 of the adjoint operator @xmath158 ) and hence it is a closed subspace of @xmath111 .
therefore , @xmath70 is a pole of the resolvent ( see theorem a.3.3 of @xcite ) .
furthermore , since @xmath159 we see that @xmath160 for @xmath26 small enough , and hence @xmath120 .
substituting @xmath124 by @xmath161 in the characteristic equation @xmath162 we have that the imaginary part is @xmath163 . since @xmath164 , there are no non real solutions of note that @xmath165 we now write an expansion of the eigenvalue @xmath70 .
[ proposition2 ]
let @xmath70 be the dominant eigenvalue of the operator @xmath71 .
then @xmath166 let us consider the change of variable @xmath167 where @xmath168 .
we have @xmath169 then @xmath170 where we have used @xmath171 and have denoted @xmath172 and @xmath173 .
+ since @xmath174 and @xmath175 we obtain @xmath176 which implies @xmath177 since @xmath178 we have @xmath179 for @xmath26 small enough .
+ therefore , using in we get @xmath180 finally , by and , let us start this subsection with a lemma in which properties of the spectrum of @xmath182 are used to study the asymptotic behavior of the semigroup @xmath183 generated by @xmath184 .
[ lemma2.0 ] * the essential growth bound of the semigroup generated by @xmath184 is @xmath185 * the growth bound of the semigroup generated by @xmath184 is @xmath186 * @xmath184 is a compact ( one rank ) perturbation of @xmath187 then @xmath188 where @xmath189 is the semigroup generated by @xmath190 ( see @xcite ) .
+ since @xmath190 is a multiplication operator , @xmath191 and the result follows .
* by proposition [ proposition1 ] , the spectral bound of @xmath184 is @xmath9 and the spectral mapping theorem holds for any positive @xmath192semigroup on @xmath193 ( see @xcite ) .
let us now write , for a positive non identically zero @xmath76 , @xmath194 where @xmath195 is the eigenvector corresponding to the eigenvalue 0 of @xmath184 and @xmath196 is the spectral projection of @xmath76 on the kernel of @xmath184 ( note that @xmath197 since @xmath198 is positive and @xmath184 is the generator of an irreducible positive semigroup ) .
we also define @xmath199 .
the following lemma gives the asymptotic behavior of @xmath200 : [ lemma4 ] let us assume that @xmath76 is a positive integrable function on @xmath7 .
then there exist positive constants @xmath201 , @xmath202 ( independent of @xmath26 but depending on @xmath76 ) such that @xmath203 moreover , @xmath204 recall that @xmath205 where @xmath206 is the eigenvector of the adjoint operator @xmath158 corresponding to the eigenvalue @xmath70 , normalized such that @xmath207 .
since @xmath208 we see that @xmath209 let us start by bounding the denominator from above . using that , by proposition [ proposition2 ] , for @xmath26 small enough ,
@xmath210 we obtain the bound @xmath211 similarly , since for @xmath26 small enough , @xmath212 , so that @xmath213}\gamma(x ) \int_{-a}^{a}\frac{\,dx}{\left(2(\gamma(0)\pi\varepsilon)^2+x^2\right)^2 } \geq \frac{k_3}{\varepsilon^2 } .\ ] ] for the numerator we have , on the one hand , @xmath214 where the right hand side tends to @xmath9 when @xmath26 goes to @xmath9 by an easy application of the lebesgue dominated convergence theorem ( note that the integrand is bounded above by @xmath215 ) .
on the other hand , notice that there exists an interval @xmath216 which does not contain @xmath9 such that @xmath217 .
then , since @xmath218 and @xmath219 there exists a constant @xmath220 such that @xmath221 by and , @xmath222 and by and , and @xmath26 small enough , @xmath223 this completes the proof .
if @xmath224 is bounded below by a positive number @xmath225 in a neighbourhood @xmath226 of @xmath9 , then the lower estimate can be improved using that @xmath227 indeed , for @xmath26 small enough @xmath228 so in this case , for @xmath26 small enough , @xmath229 for some constant @xmath230 independent of @xmath26 .
the next two lemmas enable to estimate @xmath231 ( defined above lemma [ lemma4 ] ) .
in the first one , the dependence w.r.t .
@xmath1 is not explicit .
[ lemma1 ] for @xmath26 small enough and any @xmath78 there exists @xmath232 such that @xmath233 since by lemma [ lemma2.0 ] @xmath234 , we can apply theorem @xmath235 in @xcite , and get the estimate @xmath236 proposition [ proposition2 ] gives then the statement .
we now give an estimate of the dependence of @xmath237 on @xmath26 , provided that @xmath238 is chosen far enough from its limit value .
more precisely , we choose @xmath239 [ lem6 ] for @xmath26 small enough there exists a constant @xmath230 independent of @xmath26 and of @xmath76 such that @xmath240 since the proof of this result is quite technical we delay it to the end of this section ( subsection [ app ] ) .
we now rewrite equation ( [ eqq0 ] ) as @xmath241 we look for solutions of ( [ eq6 ] ) ( with positive initial condition @xmath242 ) which can be written as @xmath243 , with @xmath244 a function of time such that @xmath245 . substituting in ( [ eq6 ] ) , it follows that @xmath246 is indeed a solution of eq .
( [ eqq0 ] ) if @xmath247 satisfies the following initial value problem for an ordinary differential equation : @xmath248 or equivalently @xmath249 the two next lemmas explain the asymptotic behavior of @xmath247 .
in the first one , the dependence w.r.t .
@xmath1 of the constants is not explicit .
[ lemma3 ] for @xmath37 small enough and any @xmath78 , there exists a positive constant @xmath250 such that @xmath251 the solution of is explicitly given by @xmath252 then @xmath253 where for the last inequality we have used that the denominator is a positive continuous function bounded below ( it takes the value @xmath12 for @xmath46 and its limit is @xmath254 when @xmath0 goes to infinity ) .
we also used the following estimate for the numerator : since , by lemma [ lemma1 ] , @xmath255 , then @xmath256 in order to give an estimate of the dependence of @xmath257 w.r.t .
@xmath258 we need to bound the denominator more precisely and to take a value of @xmath238 separated of its limit value . as in lemma
[ lem6 ] , we choose @xmath239 [ lem5 ] for @xmath37 small enough , there exist constants @xmath259 and @xmath260 ( independent of @xmath26 ) such that @xmath261 using lemma [ lemma1 ] and the fact that the second term is positive we see that @xmath262 for any @xmath263 such that @xmath264 ( notice that the left hand side in is an increasing function of @xmath263 ) .
this indeed happens if @xmath265 and @xmath266 since the second condition is weaker than the first one for @xmath26 small enough , holds whenever @xmath263 is such that @xmath267 , i.e. , @xmath268 and @xmath37 is sufficiently small .
so , @xmath269 is also a lower bound in , and we finally have @xmath270 using the bound on the numerator given in the proof of lemma [ lemma3 ] , the previous estimate and using also lemma [ lem6 ] , lemma [ lemma4 ] and proposition [ proposition2 ] , we obtain @xmath271 we are now in position to conclude the proof of theorem [ theorem1 ] .
we recall that @xmath272 satisfies the integral equation @xmath273 from which the following identity follows @xmath274 i.e. , @xmath275 is a solution of the variations of constants equation .
+ on the other hand , the nonlinear part of the right hand side of ( [ eq6 ] ) is a locally lipschitz function of @xmath276 . from this uniqueness follows , whereas global existence is clear from the previous lemmas .
finally , a standard application of the triangular inequality and lemmas [ lemma4 ] , [ lemma1 ] and [ lemma3 ] gives @xmath277 using lemmas [ lem6 ] and [ lem5 ] in the second inequality of , the last statement of theorem [ theorem1 ] follows . by the triangular inequality ,
@xmath278 hence by proposition [ proposition2 ] and theorem [ theorem1 ] , we only need to estimate the last term , for which we have @xmath279 let us bound the three terms . for the first one we have , by proposition [ proposition2 ] , @xmath280 for the second one , by proposition [ proposition2 ] and , @xmath281 for the third one , similarly to the proof of proposition [ proposition2 ] , denoting by @xmath172 , @xmath282 let us consider the linear initial value problem @xmath283 where @xmath284 .
let us recall that @xmath285 and @xmath286 ( see proposition [ proposition1 ] ) . applying the laplace transform with respect to @xmath0 to the previous equation
, we obtain the identity @xmath287(\mu , x)-u_{0}(x)= ( a_{\varepsilon}(x)-\lambda_{\varepsilon})\,\mathcal{l}[u](\mu , x)+\varepsilon\ , \gamma(x)\,\int_{i}\mathcal{l}[u](\mu , y)\ , \,dy,\ ] ] that is @xmath288(\mu , x)=\frac{u_{0}(x)}{\mu+\lambda_{\varepsilon}-a_{\varepsilon}(x)}+\frac{\varepsilon \,\gamma(x)}{\mu+\lambda_{\varepsilon}-a_{\varepsilon}(x)}\int_{i}\mathcal{l}[u](\mu , y)\,dy.\ ] ] integrating ( with respect to @xmath11 ) , we obtain @xmath289(\mu , x)\,dx=\frac{\int_{i}\frac{u_{0}(x)}{\mu+\lambda_{\varepsilon}-a_{\varepsilon}(x)}\,dx } { 1- \int_{i}\frac{\varepsilon \gamma(x)}{\mu+\lambda_{\varepsilon}-a_{\varepsilon}(x)}\,dx}=\frac{\int_{i}\frac{u_{0}(x)}{\mu+\lambda_{\varepsilon}-a_{\varepsilon}(x)}\,dx } { \varepsilon \mu \int_{i}\frac{\gamma(x)}{(\lambda_{\varepsilon}-a_{\varepsilon}(x))(\mu+\lambda_{\varepsilon}-a_{\varepsilon}(x))}\,dx},\ ] ] where we have used , for the second equality , @xmath290 . substituting in ,
we get @xmath291(\mu , x)=\frac{u_{0}(x)}{\mu+\lambda_{\varepsilon}-a_{\varepsilon}(x)}+ \frac{\int_{i}\frac{u_{0}(x)}{\mu+\lambda_{\varepsilon}-a_{\varepsilon}(x)}\,dx } { \mu \int_{i}\frac{\gamma(x)}{(\lambda_{\varepsilon}-a_{\varepsilon}(x))(\mu+\lambda_{\varepsilon}-a_{\varepsilon}(x))}\,dx}\frac{\gamma(x)}{(\mu+\lambda_{\varepsilon}-a_{\varepsilon}(x))}.\ ] ] this laplace transform is analytic for re @xmath292 ( note that @xmath293 is positive and tends to zero when @xmath26 tends to zero ) .
then , for @xmath294 , we know , by the inversion theorem , that @xmath295(\mu , x)\,e^{\mu t}\ , \,d\mu.\ ] ] using the theorem of residues , we can shift the integration path to the left in order to obtain , for any @xmath296 @xmath297(\mu , x)e^{\mu t}\big)+\frac{1}{2\pi i}\int_{s'-i\infty}^{s'+i\infty}\mathcal{l}[u](\mu , x)e^{\mu\ , t}\ , \,d\mu,\ ] ] where @xmath298(\mu , x)e^{\mu t}\big)&= & \lim_{\mu \to 0 } \mu \mathcal{l}[u](\mu , x)\\ \\ & = & \lim_{\mu \to 0 } \left(\frac{\mu u_{0}(x)}{\mu+\lambda_{\varepsilon}-a_{\varepsilon}(x)}+\frac { \int_{i}\frac{u_{0}(x)}{\mu+\lambda_{\varepsilon}-a_{\varepsilon}(x)}\,{d}x } { \int_{i}\frac{\gamma(x)}{(\lambda_{\varepsilon}-a_{\varepsilon}(x))(\mu+\lambda_{\varepsilon } -a_{\varepsilon}(x))}\,{d}x}\,\frac{\gamma(x)}{\mu+\lambda_{\varepsilon}-a_{\varepsilon}(x)}\right)\\ \\ & = & \frac{\langle u_0 , \psi_{\varepsilon}^{*}\rangle}{\langle \psi_{\varepsilon } , \psi_{\varepsilon}^{*}\rangle}\,\psi_{\varepsilon}(x)=c_{u_0}\,\psi_{\varepsilon}(x ) \end{array}\ ] ] ( let us recall that @xmath299 and @xmath300 ) .
thus , we obtain that , for @xmath301 , @xmath302(s'+i\tau , x)\ , e^{(s'+i\tau)\ , t}\ , \,d\tau .\ ] ] we now define @xmath303 so that we can write @xmath304(s'+i\tau , x)e^{(s'+i\tau ) t } \,d\tau&= & \frac{1}{2\pi } u_0(x)e^{s't}\int_{-\infty}^{\infty}\frac{e^{i\tau t}}{s'+\lambda_{\varepsilon}-a_{\varepsilon}(x)+i \tau}\,d\tau\\ \\ & + & \frac{1}{2\pi } \gamma(x)e^{s't}\int_{-\infty}^{\infty}\frac{g_{\varepsilon}(s'+i \tau)e^{i\tau t}}{s'+\lambda_{\varepsilon}-a_{\varepsilon}(x)+i \tau}\,d\tau\\ \\ & = & e^{-(\lambda_{\varepsilon}-a_{\varepsilon}(x))t}u_{0}(x)\\ \\ & + & \frac{1}{2\pi } \gamma(x)e^{s't}\int_{-\infty}^{\infty}\frac{g_{\varepsilon}(s'+i \tau)e^{i\tau t}}{s'+\lambda_{\varepsilon}-a_{\varepsilon}(x)+i \tau}\,d\tau , \end{array}\ ] ] where we used the estimate @xmath305 and the identity @xmath306 ( for @xmath307 ) .
we now would like to find a bound for @xmath308 .
+ we see that @xmath309 and @xmath310 since @xmath311 . for the numerator of
@xmath312 we can estimate @xmath313 we now find a lower bound for the denominator of @xmath312 .
we use the elementary estimate @xmath314 and we start with the real part .
@xmath315 where in the last inequality we used the estimates @xmath316 , @xmath317 .
we also used that , since @xmath318 is strictly positive and tends to zero when @xmath26 goes to zero , there exists @xmath319 such that @xmath320 we have @xmath321 .
+ in a similar way , for the imaginary part , @xmath322 defining @xmath323 we see that @xmath324 and then , using , and @xmath325 now , since @xmath326 and @xmath327 are strictly positive continuous functions , @xmath328 and @xmath329 tends to a positive limit when @xmath330 goes to @xmath331 , there exists a constant @xmath92 ( independent of @xmath26 ) such that @xmath332 . choosing @xmath333 , where @xmath334 we can write @xmath335 finally , going back to and using , we end up with @xmath336
we start here the proof of theorem [ thm : smallt ] . from now on , @xmath95 will designate a strictly positive constant depending only on some upper bounds on @xmath96 , @xmath97 , a lower bound on @xmath98 ( see remark [ rem : c ] ) , and on @xmath337 .
thanks to the variation of the constant formula , the solution @xmath246 of ( [ eqq0 ] ) satisfies : @xmath338 where @xmath339 obtaining a precise estimate on @xmath340 is the key to prove theorem [ thm : smallt ] .
if we differentiate @xmath345 with respect to @xmath0 , we get @xmath346 which implies , since @xmath347 , that @xmath348 thanks to , and the nonnegativity of @xmath349 , one gets @xmath350 while for some constants @xmath351 , @xmath352 for @xmath353 .
note that here we used a lower bound on @xmath354 around @xmath14 ( we have assumed that @xmath90 and that @xmath354 is continuous ) . thanks to this lower bound , becomes @xmath355 thanks to and , we can estimate the second term of as follows : @xmath356 in order to estimate @xmath357 , we proceed as follows : @xmath358 this estimate combined with implies that @xmath359 satisfies @xmath360 since @xmath89 , we can estimate @xmath361 where @xmath362 thanks to ( and the definition of @xmath124 and @xmath364 : see and respectively ) , we see that @xmath365 so that @xmath366 we will now estimate each of the terms on the right hand side of .
we start by estimating the third term on the right hand side , thanks to and an integration by parts : @xmath367\nonumber\\ & \leq & \frac{c}{1-\varepsilon}\left[\frac{e^{(1-\varepsilon)t}}{t}+t\max_{s\in [ 1,t]}\frac{e^{(1-\varepsilon)s}}{s^2}\right]\nonumber\\ & \leq & \frac{2c}{(1-\varepsilon)t}e^{(1-\varepsilon)t},\label{eq : esttruc}\end{aligned}\ ] ] provided @xmath368 is large enough , and @xmath37 is small enough ( to ensure that @xmath369}\frac{e^{(1-\varepsilon)s}}{s^2}=\frac{e^{(1-\varepsilon)t}}{t^2}$ ] ) .
we now estimate the second term on the right hand side of , using an integration by parts : @xmath370 and then , applying an estimate similar to the one used to obtain , we get , provided that @xmath368 is large enough , and that @xmath37 is small enough , @xmath371 finally , we estimate the last term of the right hand side of , thanks to estimates and : @xmath372 where we have used an integration by part to obtain the last inequality .
[ [ estimation - of - lefte1-varepsilont - int_0tmathcal - isds - fracsqrt - tf00-int_i - e - x2dxright ] ] estimation of @xmath376 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ using the bounds and , we can show that as soon as @xmath379 , @xmath382 so that identity leads to the bound @xmath383 notice also , as this is going to be useful further on , that for @xmath384 , thanks to and , @xmath385 in this last part of the proof , we consider times @xmath379 .
we estimate @xmath386 let us start by estimating the second term on the right hand side of , thanks to estimate : @xmath387 we now rewrite the first term on the right hand side of , using formula and : @xmath388 and then , thanks to , and , @xmath389 we estimate @xmath390 and then @xmath391 where we have used the fact that @xmath392 . thanks to and , becomes : @xmath393 theorem [ thm : smallt ] follows from this estimate . * acknowledgement * : the research leading to this paper was funded by the french `` anr blanche '' project kibord : anr-13-bs01 - 0004 , and `` anr jcjc '' project modevol anr-13-js01 - 0009 . a.c . and s.c .
were partially supported by the ministerio de ciencia e innovacin , grants mtm2011 - 27739-c04 - 02 and mtm2014 - 52402-c3 - 2-p .
w. arendt , a , grabosch , g. greiner , i. groh , h.p .
lotz , u. moustakas , r. nagel , f. neubrander , u. schlotterbeck .
one - parameter semigroups of positive operators .
lecture notes in mathematics , 1184 .
springer - verlag , berlin , 1986 .
d. i. bolnicke , p. amarasekare , m. s. arajo , r. brger , j. m. levine , m. novak , v. h. w. rudolf , s. j. schreiber , m. c. urban , d. a. vasseur , why intraspecific trait variation matters in community ecology .
_ trends ecol .
_ 26 , 183192 ( 2011 ) .
o. diekmann , j. a. p. heesterbeek , j. a. metz , on the definition and the computation of the basic reproduction ratio @xmath398 in models for infectious diseases in heterogeneous populations . j. math
. biol . * 28*(4 ) , 365 - 382 ( 1990 ) .
g. martin , r. aguile , j. ramsayer , o. kaltz , o. ronce , the probability of evolutionary rescue : towards a quantitative comparison between theory and evolution experiments .
b _ * 368 * , 20120088 ( 2012 ) . c. violle
, b. j. enquist , b. j. mcgill , l. jiang , c. h. albert , c. hulshof , v. jung , j. messier , the return of the variance : intraspecific variability in community ecology .
_ trends ecol .
_ * 27*(4 ) , 24452 ( 2012 ) . | in this paper , we study the asymptotic ( large time ) behavior of a selection - mutation - competition model for a population structured with respect to a phenotypic trait , when the rate of mutation is very small .
we assume that the reproduction is asexual , and that the mutations can be described by a linear integral operator .
we are interested in the interplay between the time variable @xmath0 and the rate @xmath1 of mutations .
we show that depending on @xmath2 , the limit @xmath3 with @xmath4 can lead to population number densities which are either gaussian - like ( when @xmath5 is small ) or cauchy - like ( when @xmath5 is large ) . |
the energy loss of partons is predicted to be a sensitive probe of the qcd matter created in high energy nucleus - nucleus collisions since its magnitude depends strongly on the color charge density of the matter traversed @xcite .
in particular , the understanding of the flavor dependent coupling of quarks and the modification of their fragmentation function give essential information on the properties of the hot qcd matter produced in such collisions @xcite . due to their large mass ( @xmath1 gev/@xmath2 ) ,
heavy quarks are believed to be produced predominantly in hard scattering processes in the early stage of the collision and , therefore , probe the complete space - time evolution of the expanding medium and their yields are sensitive to the initial gluon density @xcite .
heavy - quark production by initial state gluon fusion also dominates in heavy - ion collisions where many , in part overlapping nucleon - nucleon collisions occur .
it has been shown that charm production in the qcd medium might be significant at lhc energies @xcite .
the penetrating power of heavy quarks is much higher than for light quarks , providing a sensitive probe of the matter .
theoretical models based on perturbative qcd predict that the energy loss of heavy quarks in the medium is expected to be smaller compared to light - quarks and gluons due to the mass dependent suppression of the gluon radiation at small angles , the so - called dead - cone effect @xcite .
however , recent rhic results on single electron production in nucleus - nucleus collisions have shown that the yield at high transverse momentum is suppressed to the same level as observed for light - quark hadrons .
azimuthal angular correlations of heavy - quark particles allow for the study of the fragmentation function of charm and bottom quarks separately , where high-@xmath0 decay electrons are associated with open charmed mesons .
of single electrons in the 10% most central au+au collisions at @xmath4 = 200 gev measured by the phenix collaboration .
the data are compared with a collisional dissociation model ( purple curve ) . ] at rhic charm and bottom quarks are identified by assuming that isolated electrons in the event originate from semileptonic decays of heavy - quark mesons . at high transverse momentum , this mechanism of electron production is dominant enough to reliably subtract other sources of electrons such as @xmath5 conversions and @xmath6 dalitz decays .
star @xcite and phenix measurements @xcite in central au+au collisions have shown that the high-@xmath0 yield of electrons from semileptonic charm and bottom decays is suppressed relative to properly scaled proton - proton collisions , quantified in the nuclear modification factor @xmath3 @xcite ( cf .
figure [ fig:1 ] ) .
this factor exhibits an unexpectedly similar amount of suppression as observed for light - quark hadrons , suggesting substantial energy loss of heavy quarks in the produced medium .
this finding is not in line with expectations from the dead - cone effect .
energy - loss models incorporating contributions from charm and bottom do not describe the observed suppression sufficiently well ( see discussions in @xcite ) .
although it has been realized that energy loss of heavy quarks by elastic parton scattering causing collisional energy loss is probably of comparable importance to energy loss by gluon radiation , yet the quantitative description of the suppression is still not satisfying .
first attempts has been performed to distinguish between radiative and collisional energy loss @xcite .
furthermore , it has been shown that the in - medium fragmentation of heavy quarks into heavy mesons which are collisionally dissociated , a process that can happen multiple times in the medium , may be significant in heavy - ion collisions .
in addition , cold nuclear effects might explain part of the observed high-@xmath0 yield suppression @xcite .
collisions at @xmath7 = 200 gev obtained from electron hadron and electron d@xmath8 meson azimuthal correlations .
error bars are statistical and brackets are systematic uncertainties .
the solid curve illustrates the fonll calculation @xcite with its theoretical uncertainties ( dashed curve ) . ]
perturbative qcd calculations at fixed - order plus next - to - leading logarithm ( fonll ) level have shown that the bottom contribution to the heavy - quark decay electrons is significant at intermediate @xmath0 ( 4 - 5 gev / c ) although the uncertainties are relatively large @xcite .
therefore , measurements of the relative charm and bottom contribution to the single electrons is essential .
azimuthal angular correlations of single electrons and hadrons allow to exploit the different fragmentation of the associated jets @xcite .
heavy quarks have , in general , a harder fragmentation function than gluons and light quarks , making the near - side correlation more sensitive to the decay kinematics . for the same electron transverse momentum the near - side electron
hadron angular correlation distribution from b decays is much broader than that from d decays .
furthermore , azimuthal angular correlations of heavy - quark decay electrons and open charmed mesons yield important information about the charm and bottom contributions due to their different decay kinematics @xcite .
flavor conservation implies that heavy quarks are produced in quark anti - quark pairs .
a more detailed understanding of the underlying production process can be obtained from events in which both heavy - quark particles are detected . due to momentum conservation
, these heavy - quark pairs are correlated in relative azimuth ( @xmath9 ) in the plane perpendicular to the colliding beams , leading to the characteristic back - to - back orientated sprays of particles .
this correlation survives the fragmentation process to a large extent in @xmath10 collisions .
charm quarks predominantly ( @xmath1154% ) hadronize directly and bottom quarks via b decays into d@xmath8 mesons .
the branching fraction for charm and bottom decays into electrons is @xmath1110% @xcite .
while triggering on the so - called leading electron ( trigger side ) , the balancing heavy quark , which is identified by the d@xmath8 meson , can be used to identify the underlying production mechanisms ( probe side ) . requiring like - sign electron
kaon pairs , where the kaons stem from the d@xmath8 decays , the near - side correlation peak can be attributed to b decays only whereas the contribution to the away - side peak can mostly be attributed to charmed meson decays .
figure [ fig:2a ] depicts the relative bottom contribution to the single electrons from @xmath10 collisions at @xmath7 = 200 gev measured by the star experiment @xcite .
the phenix collaboration obtained similar results @xcite .
the data provide convincing evidence that the bottom contribution to the single electron yields is of the same magnitude than that of charm at and above @xmath0 = 5 gev / c .
the direct measurements of open charm and bottom mesons at rhic will be possible with the high precision @xmath12 vertex detectors in the star and phenix experiments .
collisions at rhic and lhc energies obtained from fonll calculations @xcite . ]
figure [ fig:2b ] illustrates the charm and bottom contribution to the single electrons at 0.2 , 10 and 14 tev @xcite .
it is remarkable that the crossing point , where bottom starts to dominate over charm seems to be at around the same @xmath0 for all three collision energies .
a possible explanation is that at the lhc energies more bottom is produced at small @xmath0 with respect to charm than at rhic energies , which would bring the crossing point closer to zero . on the other hand
the single spectra of bottom and charm are likely to be more similar at the lhc due to the fact that the quark mass is less relevant .
this would push the crossing point away from zero .
it seems that both mechanisms roughly compensate each other .
alice , a large ion collider experiment , is the dedicated detector to study hot qcd matter in lead - lead collisions at the tev scale at the cern - lhc . the initial energy density in the collision zone is expected to be about one order of magnitude higher than at the rhic facility ( @xmath11100 gev/@xmath13 ) .
the higher energy density allows thermal equilibrium to be reached more quickly and creates a relatively long - lived qgp phase .
therefore , it is expected that most of the in - medium effects will be enhanced .
the first pb+pb data taken in november 2010 at @xmath14 = 2.76 tev provided first exciting results @xcite . also first heavy - flavor results from @xmath10 collisions at @xmath7 = 7 tev have been presented at this workshop @xcite .
mesons from pythia simulations of @xmath10 interactions ( black histogram ) and from pyquen simulations of the 5% most central pb+pb collisions at @xmath14 = 5.52 tev ( red histogram ) .
right panel : di - hadron azimuthal correlation distribution from @xmath10 , @xmath15+au and au+au collisions at @xmath14 = 0.2 tev measured by the star experiment.,title="fig : " ] mesons from pythia simulations of @xmath10 interactions ( black histogram ) and from pyquen simulations of the 5% most central pb+pb collisions at @xmath14 = 5.52 tev ( red histogram ) .
right panel : di - hadron azimuthal correlation distribution from @xmath10 , @xmath15+au and au+au collisions at @xmath14 = 0.2 tev measured by the star experiment.,title="fig : " ] the feasibility of heavy - flavor correlations in pb+pb collisions at lhc energies were studied using the pyquen event generator @xcite .
the azimuthal angular correlation of single electrons and d@xmath8 mesons at mid - rapidity ( @xmath16 ) is studied in pythia simulations of @xmath10 interactions and in pyquen simulations of the 5% most central pb+pb collisions at @xmath14 = 5.52 tev .
the left panel of figure [ fig:3 ] shows both correlation distributions . for the @xmath10 collisions the correlation distribution exhibits a pronounced near - side peak , which can be attributed to b decays , and an away - side peak , which arises from charm decays . for pb+pb collisions
, however , the near - side peak is expected to be suppressed whereas the away - side correlation peak disappear .
a similar pattern have be found for di - hadron correlations with the first data from the rhic facility ( figure [ fig:3 ] , right panel ) @xcite , indicating promising results to come from the first lhc data sample .
the yield of the near - side peak for an electron trigger @xmath0 of @xmath17 gev / c was extracted from the integral of a gaussian fit to the data ( @xmath18 range around the peak position ) .
the width of the near - side peak from pyquen ( pb+pb case ) is slightly broader compared to pythia ( @xmath10 case ) .
the yield as a function of the associated @xmath0 of the d@xmath8 mesons is shown in the left panel of figure [ fig:4 ] .
the @xmath19 decreases and stays constant at around @xmath0 = 8 gev / c .
the magnitude of the yield suppression of the near - side peak in central pb+pb collisions at @xmath14 = 5.52 tev is similar to the nuclear modification factor of the single electrons , as illustrated in the right panel of figure [ fig:4 ] .
the single electron @xmath3 agrees with the alice measurement @xcite and previous theoretical model calculations @xcite . of the near - side peak ( left panel ) and @xmath3 of single electrons ( right panel ) from pyquen simulations of the 5% most central pb+pb collisions at @xmath14 = 5.52 tev .
the pyquen settings are specified in the right hand panel.,title="fig : " ] of the near - side peak ( left panel ) and @xmath3 of single electrons ( right panel ) from pyquen simulations of the 5% most central pb+pb collisions at @xmath14 = 5.52 tev .
the pyquen settings are specified in the right hand panel.,title="fig : " ]
in the previous sections is has been shown that the understanding of the heavy - flavor production mechanism is of considerable interest .
higher order sub - processes such as gluon splitting to @xmath20 pairs are expected to have a significant contribution to the total open charm yield at lhc energies @xcite .
the gluon splitting contribution is studied by identifying the _ charm content in jets_. by utilizing jet - finder algorithms and isolating the near side of the jet-@xmath21 azimuthal angular correlation , the charm ( @xmath21 ) content in a jet was measured . from the correlation strength on the near side , it is possible to estimate the amount of open charm production which results from gluon splitting .
the lep experiments @xcite have studied the d*@xmath22-meson content in jets and demonstrated that the production from z@xmath8 decays in @xmath23 collisions is dominated by d * mesons that carry large fractions of the jet momenta .
this finding is consistent with the jets being produced from primary c ( anti-)quarks . at higher collision energies
( @xmath24 at @xmath7 = 1.8 tev and 630 gev ) the cdf and ua1 experiments have observed d*@xmath22 mesons in the cone of jets with transverse energies larger than 40 gev @xcite .
their fractional momenta are found to be smaller , consistent with a different production mechanism in which the d * mesons originate from gluon splitting into @xmath25 pairs ( @xmath26 in the initial or final parton shower , with neither of the quarks from the @xmath25 pair participating in the hard qcd interaction ) @xcite . the star experiment measured a gluon splitting rate of about 10% in @xmath10 at @xmath7 = 200 gev @xcite .
the later three measurements are summarized in the left panel of figure [ fig:5 ] compared to perturbative qcd @xcite .
moreover , the star measurement is consistent with results from mc@nlo calculations @xcite , a qcd computation with a realistic parton shower model .
first performance studies have been conducted at lhc .
the right panel of figure [ fig:5 ] shows the charged @xmath21 signal reconstructed within the cone of jets with the alice experiment in @xmath10 collisions at @xmath7 = 7 tev @xcite .
this is an important step towards the determination of the charm content in jets at the tev scale .
signal reconstructed with the alice experiment within the cone of jets in @xmath10 collisions at @xmath7 = 7 tev @xcite .
see text for details.,title="fig : " ] signal reconstructed with the alice experiment within the cone of jets in @xmath10 collisions at @xmath7 = 7 tev @xcite .
see text for details.,title="fig : " ]
heavy quarks are particularly good probes to study the properties of hot qcd matter ( especially its transport properties ) .
rhic data exhibit a relatively large energy loss of heavy quarks in the medium .
azimuthal angular correlations of single electron
hadron and single electron
d@xmath8 mesons show that bottom decay yields are of comparable magnitude at and above @xmath115 gev / c , consistent with fonll calculations , indicating that in particular bottom is more strongly suppressed than expected . at the lhc
, heavy - flavor particle correlations will allow for the study of the possible modification of the fragmentation function in the medium .
next - to - leading - order qcd processes , such as gluon splitting , become important at lhc energies and their contribution can be studied through the measurement of the _ charm content in jets_. first heavy - ion collisions in november 2010 mark the start of the study of hot quark matter in a new energy domain .
the author would like to thank the organizers for the nice atmosphere and the stimulating discussions during the workshop .
the european research council has provided financial support under the european community s seventh framework programme ( fp7/2007 - 2013 ) / erc grant agreement no 210223 .
this work was supported in part by a vidi grant from the netherlands organisation for scientific research ( project number 680 - 47 - 232 ) . | measurements at the relativistic heavy ion collider ( rhic ) at brookhaven national laboratory have revealed strong modification of the jet structure in high - energy heavy - ion collisions , which can be attributed to the interaction of hard scattered partons with the hot and dense qcd matter .
the study of heavy - quark ( charm and bottom ) production in such collisions provides key tests of parton energy - loss models and , thus , yields profound insight into the properties of the produced matter
. the high-@xmath0 yield of heavy - flavor decay electrons exhibits an unexpected large suppression .
since those single electrons have contributions from charm and bottom decays an experimental method is needed to investigate them separately .
heavy - flavor particle correlations provide information about the underlying production mechanism . in this contribution , a review on recent measurements on azimuthal correlations of single electrons and open charmed mesons at rhic and perspectives of such measurements at the cern - large hadron collider ( lhc ) are presented .
moreover , it has been shown that next - to - leading - order ( nlo ) qcd processes , such as gluon splitting , become important at lhc energies .
it will be demonstrated how this contribution can be determined through the measurement of the _ charm content in jets_. |
the wheeler - dewitt ( wdw ) equation represents the quantum version of the classical constraint describing the invariance with respect to time reparametrization@xcite . in the context of the friedmann - lematre - robertson - walker space - time
, it assumes the simple form @xmath0 \psi\left ( a\right ) \nonumber\\ & = \left [ -\frac{\partial^{2}}{\partial a^{2}}-\frac{q}{a}\frac{\partial } { \partial a}+\frac{9\pi^{2}}{4g^{2}}\left ( a^{2}-\frac{\lambda}{3}% a^{4}\right ) \right ] \psi\left ( a\right ) = 0,\label{wdw_0}%\end{aligned}\ ] ] where we have introduced the following line element@xmath1 and where we have denoted with @xmath2 the metric on the three - sphere .
@xmath3 is the lapse function , @xmath4 is the scale factor , @xmath5 and @xmath6 are the newton s constant and the cosmological constant , respectively .
we have also introduced a factor order ambiguity @xmath7 . in eq.@xmath8
all the degrees of freedom except the scale factor @xmath9 have been integrated and matter fields have been excluded . if the wdw equation is interpreted as an eigenvalue equation , one simply finds@xmath10 namely a zero energy eigenvalue .
however , it appears that the wdw equation has also a hidden structure .
indeed eq.@xmath8 has the structure of a sturm - liouville eigenvalue problem with the cosmological constant interpreted as the associated eigenvalue .
we recall to the reader that a sturm - liouville differential equation is defined by@xmath11 and the normalization is defined by@xmath12 where the boundary conditions are momentarily suspended .
it is a standard procedure , to convert the sturm - liouville problem @xmath13 into a variational problem of the form@xmath14 = \frac{-\int_{a}^{b}dxy^{\ast}\left ( x\right ) \left [ \frac{d}{dx}\left ( p\left ( x\right ) \frac{d}{dx}\right ) + q\left ( x\right ) \right ] y\left ( x\right ) } { \int_{a}^{b}dxw\left ( x\right ) y^{\ast}\left ( x\right ) y\left ( x\right ) } \,,\label{funct}%\ ] ] with unspecified boundary condition . if @xmath15 is an eigenfunction of @xmath13 , then@xmath16 y\left ( x\right ) } { \int_{a}^{b}dxw\left ( x\right ) y^{\ast}\left ( x\right ) y\left ( x\right ) } \,,\ ] ] is the eigenvalue , otherwise@xmath17 y\left ( x\right ) } { \int_{a}^{b}dxw\left ( x\right ) y^{\ast}\left ( x\right ) y\left ( x\right ) } \,.\ ] ] * * * * the minimum of the functional @xmath18 $ ] corresponds to a solution of the sturm - liouville problem @xmath19 with the eigenvalue @xmath20 in the case of the flrw model we have the following correspondence@xmath21 since @xmath22 , the normalization becomes@xmath23 where it is understood that @xmath24 . in the mini - superspace approach with a flrw background , one finds@xmath25 \psi\left ( a\right ) } { \int \mathcal{d}aa^{q+4}\psi^{\ast}\left ( a\right ) \psi\left ( a\right ) } % = \frac{3\lambda\pi^{2}}{4g^{2}}.\label{wdw_1}%\ ] ] as a concrete case , fixing @xmath26 and taking as a trial wave function @xmath27 , one finds @xmath28 when @xmath29 .
then the only solution allowed is complex and therefore it must be discarded@xcite . of course , the general @xmath7 case is much more complicated@xcite .
note that the global energy eigenvalue is still vanishing .
what we can compute in the sturm - liouville formulation is the degree of degeneracy which is represented by the cosmological constant and the value of the cosmological constant itself . in the next section we give the general guidelines to build a sturm - liouville problem associated to the wdw equation when general relativity ( gr ) is distorted .
units in which @xmath30 are used throughout the paper .
in this section we will give some glimpses on how the wdw is modified when we distort gr . the reference metric will be the flrw line element and we will consider modifications coming from gravity s rainbow , hl gravity , vsl cosmology and gup deformation . we begin to consider gravity s rainbow and hl gravity
. gravity s rainbow is a distortion of the metric tensor around and beyond the planck scale@xcite .
the basic ingredient is the definition of two arbitrary functions @xmath31 and @xmath32 , which have the following property@xmath33 for a flrw metric , the deformation acts in the following way,@xmath34 where@xmath35 is the line element on the three - sphere .
of course , ordinary gravity is recovered when @xmath36 . when we generalize eq.@xmath37 to include @xmath31 and @xmath32 , the equation modifies in the following way@xmath38 \psi\left ( a\right ) = 0,\label{wdwgrw}%\ ] ] where we have set @xmath39 and defined the distorted potential as@xmath40 ^{2}a^{2}\left [ 1-\frac{a^{2}}{a_{0}^{2}g_{2}^{2}\left ( e / e_{\mathrm{pl}% } \right ) } \right ] ~,\label{u(a , e)}%\ ] ] with @xmath41 .
the potential @xmath42 has been used to discuss a possible alternative explanation of the inflation problem@xcite . moreover , always in the context of the flrw space - time , it is possible to build a bridge between gravity s rainbow and hl theory@xcite . indeed ,
if one considers@xmath43 with the choice@xmath44 and@xmath45 , \ ] ] one finds that the wdw equation becomes @xmath46 \psi\left ( a\right ) = 0.\label{hhl}%\ ] ] to obtain eq.@xmath47 , we have also defined@xmath48 \frac{de}{da}%\ ] ] and@xmath49 .\ ] ] moreover , by identifying@xmath50 it is immediate to recognize that eq.@xmath47 represents the wdw equation for the projectable version of a hoava - lifshitz gravity , without detailed balanced condition .
a vsl cosmology model is described by the following line element@xmath51 and where @xmath52 is an arbitrary function of time with the dimensions of a @xmath53 $ ] .
following @xcite , we assume that@xmath54 where @xmath55 is a reference length scale .
the form of the background is such that the shift function @xmath56 vanishes . in this case ,
when @xmath57 is promoted to an operator , we can write@xmath58 where we have introduced a factor ordering ambiguity .
thus the wdw equation
@xmath59 simply becomes@xmath60 where we have assumed that the factor ordering is not distorted by the presence of a vsl and we have set @xmath39 .
the quantum potential is defined as@xmath61 note that the potential @xmath62 vanishes in the same points where the original @xmath63 has its roots .
the wdw equation deformed by a gup can be derived modifying the original undeformed momentum conjugate to the scale factor @xmath9@xmath64 into@xcite @xmath65 where @xmath66 is a coefficient with inverse momentum dimensions , in order to reestablish the correct powers .
then the wdw equation for a flrw metric deformed by a gup becomes@xmath67 \psi\left ( a\right ) = 0.\ ] ] this deformation of the wdw equation prevents the existence of singularities @xcite .
this is because this deformation modifies the uncertainty principle as@xmath68 thus , we obtain a minimum value for the scale factor of the universe , @xmath69 .
note that the action of the gup is only on the kinetic term .
in this work , we have considered the wdw equation on a flrw background and we have shown how such an equation is modified when some deviations from gr are considered . even if we have only considered a mini - superspace approach with the scale factor @xmath9 as unique degree of freedom , one has not to think that this procedure works only on a flrw background .
for instance , in a series of papers@xcite , it has been shown that gravity s rainbow can keep under control uv divergences , at least to one loop .
this procedure has been widely tested on a spherically symmetric background .
the same procedure can be applied also to noncommutative geometries@xcite and it has been extended to include also a @xmath70 theory@xcite .
r. garattini , _ phys .
_ * b * * 685 * , 329 ( 2010 ) , arxiv : 0902.3927 [ gr - qc ] .
r. garattini and g. mandanici , _ phys . rev . _ * d 85 * , 023507 ( 2012 ) ; arxiv:1109.6563 [ gr - qc ] .
r. garattini , jcap * 1306 * , 017 ( 2013 ) , arxiv : 1210.7760 [ gr - qc ] .
r. garattini and f. s. n. lobo , _ phys .
_ * d * * 85 * , 024043 ( 2012 ) , arxiv:1111.5729 [ gr - qc ] .
r. garattini , _ int .
j. mod .
ser . _ * 14 * , 326 ( 2012 ) , arxiv:1112.1630 [ gr - qc ] .
r. garattini and f.s.n .
lobo , _ eur .
j. _ * c 74 * ( 2014 ) , arxiv:1303.5566 [ gr - qc ] .
r. garattini and b. majumder , _ nucl .
* b * * 883 * ( 2014 ) , arxiv:1305.3390 [ gr - qc].r .
garattini and b. majumder , _ nucl .
_ * b * * 884 , * ( 2014 ) , arxiv:1311.1747 [ gr - qc ] . | the wheeler - dewitt equation represents a tool to study quantum gravity and quantum cosmology .
its solution in a very general context is , of course , impossible . to this purpose
we consider some distortions of general relativity like gravity s rainbow , varying speed of light cosmology , generalized uncertainty principle deformations and hoava - lifshitz gravity which could allow the calculation of some observables like the cosmological constant . for simplicity
we consider only the mini - superspace approach related to a friedmann - lematre - robertson - walker space - time . |
for a long time , it was assumed that neutrinos have vanishing quantities : @xmath0 , and @xmath1 . among these the neutrino mass problem
has attracted the most attention , and finally we might have an evidence for nonzero neutrino mass @xcite . the other important neutrino property to be exploited
is the electromagnetic property , in particular the magnetc moment .
the reason for vanishing neutrino mass was very naive in 50 s and 60 s : the hypothesis of the @xmath2-invariance . under the @xmath2-invariance , @xmath3
, @xmath4 appears only in one chirality . in the standard model ( sm ) , this is encoded as no right - handed neutrino . in gauge theory models ,
the story changes because one can calculate the properties of the neutrino at high precision . in sm
, one can not write a mass term for @xmath4 in @xmath5 terms . to write a mass term for @xmath4
, one has to introduce @xmath6 terms , or introduce singlet neutrino(s ) .
the two - component neutrino we consider in the left - handed doublet can be weyl or majorana type .
if it is a weyl neutrino , it satisfies @xmath7 where @xmath8 or @xmath9 .
then the magnetic moment term is given by @xmath10 .
$ ] therefore , to have a nonvanishing magnetic moment ( or mass ) , we must require @xmath11 , i.e. the existence of right - handed singlet neutrino(s ) . for majorana neutrinos , @xmath12 ,
it is possible to write a mass term without introducing right - handed neutrino(s ) .
thus , it is possible to introduce neutrino masses and magnetic moments in the sm , by a slight extension of the model .
the question is how large they are . for the detection of neutrino masses and oscillations ,
there have been numerous studies from solar- , atmospheric- , reactor- , and accelerator - neutrino experiments .
the effect of neutrinos in cosmology was also used to get bounds on neutrino masses . on the other hand , for the neutrino magnetic moment astrophysical constraints gave useful bounds .
usually , the bound of the neutrino magnetic moment is given in units ( @xmath13 ) of bohr magneton ( @xmath14 ) , @xmath15
the first significant bound on magnetic moment of @xmath16 was given from astrophysics , @xmath17 , by bernstein et al .
@xcite . a better bound on @xmath18 was obtained from sn1987a , @xmath19 @xcite . for the muon neutrino
, the useful bound was obtained from the neutral current data @xcite , @xmath20 .
the first bound on the transition magnetic moment was given also from the neutral current experiment @xcite .
for the tau neutrino , @xmath21 has been obtained recently @xcite . for the theoretical side ,
it has been known from early days that it is possible to generate large magnetic moments for neutrinos @xcite .
note that the see - saw mass for neutrinos appear as in fig . 1 .
here , there does not exist any charged particle and hence there is no contribution to magnetic moment at this level .
thus to have a large neutrino magnetic moments , one needs a feynman diagram of fig .
2 type , where we introduced a heavy lepton @xmath22 coupling to @xmath23 via @xmath24 then one can easily estimate the magnetic moment of neutrinos as @xcite is the transition moment . ]
@xmath25 where @xmath26 , which is a function of mixing and feynman integral , is of order 1 .
one can also draw feynman diagrams with charged scalars in the loop with appropriate yukawa interactions introduced .
this kind of diagrams generally introduce transition magnetic moment of order@xmath27 where @xmath28 is the mass of the intermediate scalar or gauge boson .
note that without extra charged leptons @xmath29 should be neutrino mass , rendering an extremely small @xmath30 .
the super - k collaboration has reported the ratio @xmath34 which is consistent with 1 at present .
however , one may narrow down the experimental errors and can observe whether it is different from 1 or not .
one assumes that @xmath16 is not oscillated in the atmospheric neutrino data sample , and hence the mc electrons are estimated with the standard cc cross section .
the denominator is calculated assuming that the nc is the same for the cases with and without neutrino oscillation .
so @xmath35 is expected to be 1 if there is no oscillation of sm neutrinos to sterile neutrinos .
if there exist oscillation of @xmath36 to sterile neutrinos , then one expects that @xmath37 . however , in our recent work @xcite we pointed out that one should be careful to draw a firm conclusion on this matter because if a sizable transition magnetic moment of @xmath36 exists then one expects a different conclusion . for the study of nc ,
the single @xmath32 production is known to be very useful .
most dominant contribution to the single @xmath32 production at the atmospheric neutrino energy is through @xmath38 production , @xmath39 in this calculation , we used the form factors given in ref .
@xcite . for @xmath40 gev or the kinetic energy of recoil nucleon @xmath41 gev ,
the process @xmath42 is difficult to observe at super - k .
so the @xmath32 production is the cleanest way to detect nc interactions through cherenkov ring ( from @xmath32 decay ) at super - k . for transition
magnetic moment parametrized by @xmath43 , @xmath44 where @xmath45 , the single @xmath32 production cross section through @xmath38 production is given in ref . @xcite . in fig . 3
, we show the result ( the ratio of the neutrino magnetic moment contribution and the nc contribution ) as a function of neutrino energy .
note that the magnetic moment part is more important at low @xmath46 region due to the photon propagator . in principle
, one can distinguish neutrino magnetic moment interactions from the nc interactions . from fig . 3 ,
if we require @xmath47 , then we obtain a bound @xmath48 . in conclusion ,
the transition magnetic moment @xmath43 can be large . for @xmath49
heavy , it is not restricted by sn1987a bound . but atmospheric neutrinos of 110 gev can produce @xmath49 , and can mimick nc data @xcite . before interpreting nc effects from atmospheric neutrino data , one has to separate out the @xmath50 contribution .
before closing , we point out @xmath30 and solar neutrino problem . one possibility to reduce solar @xmath16 flux is to precess @xmath51 to @xmath52 with a large @xmath30 in a strong magnetic field @xcite .
but this idea seems to be ruled out by the nucleosynthesis argument @xcite , @xmath53 , and the sn1987a argument @xcite , @xmath54 .
the sn1987a bound is coming from the energy loss mechanism : if @xmath55 is created , it takes out energy out of the core .
but if it is trapped , then the bound does not apply .
are not restricted by these bounds for a heavy enough @xmath49 , but then the transition magnetic moment can not account for the solar neutrino deficit . ]
the solar neutrino problem requires @xmath56 which is considered to be large .
to use the idea of trapping , the oscillation is @xmath57 namely , we use the konopinski - mahmoud scheme where @xmath58 is the weak interaction partner of @xmath59 , implying the oscillated neutrino participates in weak interactions ; hence trapped in the supernova core .
this picture is very restrictive as suggested by models given in ref .
these models try to get a large neutrino magnetic moment while keeping the neutrino mass small .
the @xmath60 symmetry in which @xmath61 forms a doublet under @xmath60 is introduced for this purpose . in this case
, @xmath62-number minus electron - number is conserved .
the reason for vanishing mass is that @xmath63 is symmetric under @xmath60 , i.e. the mass term is a triplet under @xmath60 and hence is forbidden . on the other hand the magnetic moment term , @xmath64 , is a singlet and is allowed , as given in fig . 4 . but , _
why @xmath60 ? _
it is like the dilemma asked in any fermion mass ansatz problem .
super - kamiokande collaboration , phys .
lett . * 81 * , 1562 ( 1998 ) .
j. bernstein , m. ruderman and g. feinberg , phys . rev . * 132 * , 1227 ( 1963 ) .
j. m. lattimer and j. cooperstein , phys .
lett . * 61 * , 23 ( 1988 ) . v. s. mathur , s. okubo and j. e. kim , phys
. rev . * d10 * , 3648 ( 1974 ) .
j. e. kim , phys .
* 41 * , 360 ( 1978 ) .
s. n. gninenko , preprint hep - ph/9901289 .
j. e. kim , phys . rev . *
d14 * , 3000 ( 1976 ) .
s. k. kang , j. e. kim and j. s. lee , phys .
d , to appear.[hep - ph/9812531 ] g. l. fogli and g. nardulli , nucl . phys . *
b160 * , 116 ( 1979 ) ; _ ibid _ * b165 * , 162 ( 1980 ) . l. b. okun , m. b. voloshin , m. i. vysotsky , sov .
. phys . * 44 * , 440 ( 1986 ) .
j. a. morgan , phys .
b102 * , 247 ( 1981 ) .
m. b. voloshin , sov .
. phys . * 48 * , 512 ( 1988 ) . | i review the history on neutrino magnetic moments and apply the neurino magnetic moment idea to constrain its bound from super - kamiokande neutrino oscillation data . |
graphs are used to model infrastructure networks , the world wide web , computer traffic , molecular interactions , ecological systems , epidemics , citations , and social interactions , among others . despite the differences in the motivating applications ,
some topological structures have emerged to be important across all these domains .
triangles , which can be explained by homophily ( people become friends with those similar to themselves ) and transitivity ( friends of friends become friends ) .
this abundance of triangles , along with the information they reveal , motivates metrics such as the _ clustering coefficient _ and the _ _
ratio__s @xcite .
the triangle structure of a graph is commonly used in the social sciences for positing various theses on behavior @xcite .
triangles have also been used in graph mining applications such as spam detection and finding common topics on the www @xcite .
the authors earlier work used distribution of degree - wise clustering coefficients as the driving force for a new generative model , blocked two - level erds - rnyi @xcite . the authors have also observed that relationships among degrees of triangle vertices can be a descriptor of the underlying graph @xcite .
the information about triangles is usually summarized in terms of _ clustering coefficients_. let @xmath0 be a simple undirected graph with @xmath1 vertices and @xmath2 edges .
let @xmath3 denote the number of triangles in the graph and @xmath4 be the number of _ wedges _ ( a path of length @xmath5 ) .
the most common measure is the _ global clustering coefficient
_ @xmath6 , which measures how often friends of friends are also friends .
we show that we can achieve speed - ups of up to four orders of magnitude with extremely small errors ; see and .
our approach is not limited to clustering coefficients , however .
a per - vertex clustering coefficient , @xmath7 , is defined as the fraction of wedges centered at @xmath8 that participate in triangles .
the mean value of @xmath7 is called the _ local clustering coefficient _ , @xmath9 . a more nuanced view of triangles is given by the _ degree - wise clustering coefficients_. this is a set of values @xmath10 indexed by degree , where @xmath11 is the average clustering coefficient of degree @xmath12 vertices . in addition to degree distribution ,
many graphs are characterized by plotting the clustering coefficients , i.e. , @xmath11 versus @xmath12 .
we summarize our notation and give formal definitions in .
@lx@ & number of vertices + @xmath13 & number of vertices of degree @xmath12 + @xmath2 & number of edges + @xmath14 & degree of vertex @xmath8 + @xmath15 & set of degree-@xmath12 vertices + @xmath4 & total number of wedges + @xmath16 & number of wedges centered at vertex @xmath8 + @xmath3 & total number of triangles + @xmath17 & number of triangles incident to vertex @xmath8 + @xmath18 & number of triangles incident to a degree @xmath12 vertex + @lx@ @xmath6 & global clustering coefficient + @xmath19 & clustering coefficient of vertex @xmath8 + @xmath20 & local clustering coefficient + @xmath21 & degree - wise clustering coefficient + there has been significant work on enumeration of all triangles @xcite .
recent work by cohen @xcite and suri and vassilvitskii @xcite give parallel implementations of these algorithms .
arifuzzaman et al .
@xcite give a massively parallel algorithm for computing clustering coefficients .
enumeration algorithms however , can be very expensive , since graphs even of moderate size ( millions of vertices ) can have an extremely large number of triangles ( see e.g. , ) .
eigenvalue / trace based methods have been used by tsourakakis @xcite and avron @xcite to compute estimates of the total and per - degree number of triangles .
however , computing eigenvalues ( even just a few of them ) is a compute - intensive task and quickly becomes intractable on large graphs .
most relevant to our work are sampling mechanisms .
tsourakakis et al .
@xcite started the use of sparsification methods , the most important of which is doulion @xcite .
this method sparsifies the graph by keeping each edge with probability @xmath22 ; counts the triangles in the sparsified graph ; and multiplies this count by @xmath23 to predict the number of triangles in the original graph .
various theoretical analyses of this algorithm ( and its variants ) have been proposed @xcite .
one of the main benefits of doulion is that it reduces large graphs to smaller ones that can be loaded into memory . however
, their estimate can suffer from high variance @xcite .
alternative sampling mechanisms have been proposed for streaming and semi - streaming algorithms @xcite . yet
, all these fast sampling methods only estimate the number of triangles and give no information about other triadic measures . in this paper
, we introduce the simple yet powerful technique of _ wedge sampling _ for counting triangles .
wedge sampling is really an algorithmic template , as opposed to a single algorithm , as various algorithms can be obtained from the same basic structure .
some of the salient features of this method are : * versatility of wedge sampling : * we show how to use wedge sampling to approximate the various clustering coefficients : @xmath24 , @xmath9 , and @xmath10 . from these
, we can estimate the numbers of triangles : @xmath3 and @xmath25 .
moreover , our techniques can even be extended to find uniform random triangles , which is useful for computing triadic statistics .
other sampling methods are usually geared towards only @xmath3 and @xmath24 .
* provably good results with precise bounds : * the mathematical analysis of this method is a direct consequence of standard chernoff - hoeffding bounds .
we obtain explicit time - error - accuracy tradeoffs . in other words , if we want an estimate for @xmath24 with error at most 10% with probability
at least 99.9% ( say ) , we know we need only 380 wedge samples .
this estimate is _ independent of the size of the graph _
, though the preprocessing required by our method is linear in the number of edges ( to obtain the degree distribution ) . *
fast and scalable : * our estimates converge rapidly , well within the theoretical bounds provided .
although there is no other method that competes directly with wedge sampling , we compare with doulion @xcite .
our experimental studies show that our wedge sampling algorithm is far faster , when the variances of the two methods are similar ( see in the appendix ) .
we do not compare to eigenvalue - based approaches since they are much more expensive for larger graphs .
we present the general method of wedge sampling for estimating clustering coefficients . in later sections
, we instantiate this for different algorithms . we say a wedge is _ closed _ if it is part of a triangle ; otherwise , we say the wedge is _
open_. thus , in , -- is an open wedge , while -- is a closed wedge .
the middle vertex of a wedge is called its _ center _ , i.e. , wedges -- and -- are centered at .
wedge sampling is best understood in terms of the following thought experiment .
fix some distribution over wedges and let @xmath26 be a random wedge .
let @xmath27 be the indicator random variable that is @xmath28 if @xmath26 is closed and @xmath29 otherwise .
denote @xmath30 $ ] .
suppose we wish to estimate @xmath31 .
we simply generate @xmath32 independent random wedges @xmath33 , with associated random variables @xmath34 .
define @xmath35 as our estimate .
the chernoff - hoeffding bounds give guarantees on @xmath36 , as follows .
[ thm : hoeffding ] let @xmath37 be independent random variables with @xmath38 for all @xmath39 . define @xmath40 .
let @xmath41 $ ] .
then for @xmath42 , we have @xmath43 hence , if we set @xmath44 , then @xmath45 < \delta$ ] . in other words , with confidence at least @xmath46 , the error in our estimate is at most @xmath47 .
shows the number of samples needed for different error rates .
we show three different curves for difference confidence levels . increasing the confidence has minimal impact on the number of samples .
the number of samples is fairly low for error rates of 0.1 or 0.01 , but it increases with the inverse square of the desired error .
& & & & & & & ( secs ) + amazon0312 & 401k & 2350k & 69 m & 3686k & 0.160 & 0.421 & 0.261 + amazon0505 & 410k & 2439k & 73 m & 3951k & 0.162 & 0.427 & 0.269 + amazon0601 & 403k & 2443k & 72 m & 3987k & 0.166 & 0.430 & 0.268 + as - skitter & 1696k & 11095k & 16022 m & 28770k & 0.005 & 0.296 & 90.897 + cit - patents & 3775k & 16519k & 336 m & 7515k & 0.067 & 0.092 & 3.764 + roadnet - ca & 1965k & 2767k & 6 m & 121k & 0.060 & 0.055 & 0.095 + web - berkstan & 685k & 6649k & 27983 m & 64691k & 0.007 & 0.634 & 54.777 + web - google & 876k & 4322k & 727
m & 13392k & 0.055 & 0.624 & 0.894 + web - stanford & 282k & 1993k & 3944 m & 11329k & 0.009 & 0.629 & 6.987 + wiki - talk & 2394k & 4660k & 12594 m & 9204k & 0.002 & 0.201 & 20.572 + youtube & 1158k & 2990k & 1474
m & 3057k & 0.006 & 0.128 & 2.740 + flickr & 1861k & 15555k & 14670 m & 548659k & 0.112 & 0.375 & 567.160 + livejournal & 5284k & 48710k & 7519
m & 310877k & 0.124 & 0.345 & 102.142 + we use the wedge sampling scheme to estimate the global clustering coefficient , @xmath24 . consider the uniform distribution on wedges .
we can interpret @xmath48 $ ] as the probability that a uniform random wedge is closed or , alternately , the fraction of closed wedges . to generate a uniform random wedge , note that the number of wedges centered at vertex @xmath8 is @xmath49 and @xmath50 .
we set @xmath51 to get a distribution over the vertices .
note that the probability of picking @xmath8 is proportional to the number of wedges centered at @xmath8 .
a uniform random wedge centered at @xmath8 can be generated by choosing two random neighbors of @xmath8 ( without replacement ) .
[ clm : random ] suppose we choose vertex @xmath8 with probability @xmath52 and take a uniform random pair of neighbors of @xmath8 .
this generates a uniform random wedge .
consider some wedge @xmath26 centered at vertex @xmath8 .
the probability that @xmath8 is selected is @xmath53 .
the random pair has probability of @xmath54 .
hence , the net probability of @xmath26 is @xmath55 .
@xmath56 compute @xmath52 for all vertices select @xmath32 random vertices ( with replacement ) according to probability distribution defined by @xmath57 .
for each selected vertex @xmath8 , choose a uniform random pair of neighbors of @xmath8 to generate a wedge .
output fraction of closed wedges as estimate of @xmath24 . shows the randomized algorithm @xmath58-wedge sampler for estimating @xmath24 in a graph @xmath0 . combining the bound of with
, we get the following theorem .
[ thm : kappa ] set @xmath44 .
the algorithm @xmath58-wedge sampler outputs an estimate @xmath36 for the clustering coefficient @xmath24 such that @xmath59 with probability greater than @xmath60 .
note that the number of samples required is _ independent of the graph size _ , but computing @xmath52 does depend on the number of edges , @xmath2 . to get an estimate on @xmath3 , the number of triangles , we output @xmath61 , which is guaranteed to be within @xmath62 of @xmath3 with probability greater than @xmath46 .
we implemented our algorithms in c and ran our experiments on a computer equipped with a 2.3ghz intel core i7 processor with 4 cores and 256 kb l2 cache ( per core ) , 8 mb l3 cache , and an 8 gb memory .
we performed our experiments on 13 graphs from snap @xcite and per private communication with the authors of @xcite . in all cases ,
directionality is ignored , and repeated and self - edges are omitted .
the properties of these matrices are presented in , where @xmath1 , @xmath2 , @xmath4 , and @xmath3 are the numbers of vertices , edges , wedges , and triangles , respectively .
and @xmath24 and @xmath9 correspond to the global and local clustering coefficients .
the last column reports the times for the enumeration algorithm .
our enumeration algorithm is based on the principles of @xcite , such that each edge is assigned to its vertex with a smaller degree ( using the vertex numbering as a tie - breaker ) , and then vertices only check wedges formed by edges assigned to them for closure . as seen in wedge sampling
works orders of magnitude faster than the enumeration algorithm .
detailed results on times can be seen in in the appendix .
the timing results show tremendous savings ; for instance , wedge sampling only takes 0.015 seconds on ` as - skitter ` while full enumeration takes 90 seconds .
show the accuracy of the wedge sampling algorithm .
again detailed results on times can be seen in in the appendix . at 99.9% confidence ( @xmath63 ) , the upper bound on the error we expect for 2k , 8k , and 32k samples is .043 , .022 , and .011 , respectively .
most of the errors are much less than the bounds would suggest .
for instance , the maximum error for 2k samples is .007 , much less than that 0.43 given by the upper bound . shows the fast convergence of the clustering coefficient estimate ( for the graph amazon0505 ) as the number of samples increases .
the dashed line shows the error bars at 99.9% confidence . in all our experiments ,
the real error is always much smaller than what is indicated by .
we now demonstrate how a small change to the underlying distribution on wedges allows us to compute the local clustering coefficient , @xmath9 .
shows the procedure @xmath9-wedge sampler .
the only difference between @xmath9-wedge sampler and @xmath58-wedge sampler is in picking random vertex centers .
vertices are picked uniformly instead of from the distribution @xmath57 .
pick @xmath32 uniform random vertices . for each selected vertex @xmath8 , choose a uniform random pair of neighbors of @xmath8 to generate a wedge .
output the fraction of closed wedges as an estimate for @xmath9 .
[ thm : lcc ] set @xmath44 .
the algorithm @xmath9-wedge sampler outputs an estimate @xmath36 for the clustering coefficient @xmath9 such that @xmath64 with probability greater than @xmath60 .
let us consider a single sampled wedge @xmath26 , and let @xmath65 be the indicator random variable for the wedge being closed .
let @xmath66 be the uniform distribution on wedges .
for any vertex @xmath8 , let @xmath67 be the uniform distribution on pairs of neighbors of @xmath8 . observe that @xmath68 }
= \pr_{v \sim { { \cal v}}}[\pr_{(u , u ' ) \sim { { \cal n}}_v}[\textrm{wedge $ \{(u , v),(u',v)\}$ is closed}]]\ ] ] we will show that this is exactly @xmath9 .
@xmath69 \\ & = & { \hbox{\bf e}}_{v \sim { { \cal v } } } [ \textrm{frac . of closed wedges centered at $ v$ } ] \\ & = & { \hbox{\bf e}}_{v \sim { { \cal v } } } [ { \hbox{\bf e}}_{(u , u ' ) \sim { { \cal n}}_v}[x(\{(u , v),(u',v)\ } ) ] ] \\ & = & \pr_{v \sim { { \cal v}}}[\pr_{(u , u ' ) \sim { { \cal n}}_v}[\textrm{$\{(u , v),(u',v)\}$ is closed } ] ] \\ & = & { \hbox{\bf e}}[x]\end{aligned}\ ] ] for a single sample , the probability that the wedge is closed is exactly @xmath9 .
the bound of completes the proof .
@xmath56 and present the results of our experiments for computing the local clustering coefficients .
more detailed results can be found in the appendix . experimental setup and
the notation are the same as in .
the results again show that wedge sampling provides accurate estimations with tremendous improvements in runtime . in this case , we come closer to the theoretical error bounds . for instance , the largest different in the 2k sample case is 0.017 , which is much closer to the theoretical error bound of 0.043 .
we demonstrate the power of wedge sampling by estimating the degree - wise clustering coefficients @xmath10 . we also provide a sampling algorithm to estimate @xmath18 , the number of triangles incident to degree-@xmath12 vertices .
shows procedure @xmath11-wedge sampler .
pick @xmath32 uniform random vertices of degree @xmath12 . for each
selected vertex @xmath8 , choose a uniform random pair of neighbors of @xmath8 to generate a wedge .
output the fraction of closed wedges as an estimate for @xmath11 .
[ thm : dcc ] set @xmath44 .
the algorithm @xmath11-wedge sampler outputs an estimate @xmath36 for the clustering coefficient @xmath11 such that @xmath70 with probability greater than @xmath60 .
the proof of the following is similar to that of .
since @xmath11 is the average clustering coefficient of a degree @xmath12 vertex , we can apply the same arguments as in .
@xmath56 by modifying the template given in , we can also get estimates for @xmath18 .
now , instead of simply counting the fraction of closed wedges , we take a weighted sum .
describes the procedure @xmath18-wedge sampler .
we let @xmath71 denote the total number of wedges centered at degree @xmath12 vertices .
pick @xmath32 uniform random vertices of degree @xmath12 . for each
selected vertex @xmath8 , choose a uniform random pair of neighbors of @xmath8 to generate a wedge . for each wedge
@xmath72 generated , let @xmath73 be the associated random variable such that @xmath74 @xmath75 .
output @xmath76 as the estimate for @xmath18 .
[ thm : dcc2 ] set @xmath44 .
the algorithm @xmath18-wedge sampleroutputs an estimate @xmath77 for the @xmath18 with the following guarantee : @xmath78 with probability greater than @xmath46 . for a single sampled wedge @xmath72
, we define @xmath73 .
we will argue that the expected value of @xmath79 $ ] is exactly @xmath80 below .
once we have that , an application of the hoeffding bound of shows that @xmath81 with probability greater than @xmath46 . multiplying this inequality by @xmath82 , we get @xmath78 , completing the proof . to show @xmath79 = t_d / w_d$ ] , partition the set @xmath83 of wedges centered on degree @xmath12 vertices into four sets @xmath84 .
the set @xmath85 ( @xmath86 ) contains all closed wedges containing exactly @xmath87 degree-@xmath12 vertices .
the remaining open wedges go into @xmath88 . for a sampled wedge @xmath26 , if @xmath89 , then @xmath90 . if @xmath91 , then nothing is added .
the wedge @xmath26 is a uniform random wedge from those centered on degree-@xmath12 vertices .
hence , @xmath79 = |s|^{-1}(|s_1| + |s_2|/2 + |s_3|/3)$ ] .
now partition the set of triangles involving degree @xmath12 vertices into three sets @xmath92 , where @xmath93 is the set of triangles with @xmath87 degree @xmath12 vertices .
observe that @xmath94 .
if a triangle has @xmath87 vertices of degree @xmath12 , then there are exactly @xmath87 wedges centered in degree @xmath12 vertices ( in that triangle ) .
so , @xmath95 @xmath96 @xmath97
. therefore , @xmath79 = t_d / w_d$ ] .
@xmath56 algorithms in the previous section present how to compute the clustering coefficient of vertices of a given degree . in practice
, it may be sufficient to compute clustering coefficients over bins of degrees .
wedge sampling algorithms can still be extended for bins of degrees by a small adjustment of the sampling procedure . within each bin , we weight each vertex according to the number of wedges it produces .
this guarantees that each wedge in the bin is equally likely to be selected
. shows results on three graphs for clustering coefficients ; the remaining figures are shown in the in the appendix .
the data is grouped in logarithmic bins of degrees , i.e. , @xmath98 .
in other words , @xmath99 form the @xmath87-th bin .
we show the estimates with increasing number of samples . at 8k samples ,
the error is expected to be less than 0.02 , which is apparent in the plots .
observe that even 500 samples yields a reasonable estimate in terms of the differences by degree .
shows the time to calculate all @xmath100 values , showing a tremendous savings in runtime as a result of using wedge sampling . in this figure , runtimes are normalized with respect to the runtimes of full enumeration .
as the figure shows , wedge sampling takes only a tiny fraction of the time of full enumeration especially for large graphs .
while most triadic measures focus on the number of triangles and their distribution , the triangles themselves , not only their count , can reveal a lot of information about the graph .
the authors recent work @xcite has looked at the relations among the degrees of the vertices of triangles . in these experiments ,
a full enumeration of the triangles was used , which limited the sizes of the graphs we could use . to avoid this computational burden , a uniform sampling of the triangles can be used . _
wedge sampling provides a uniform sampling the triangles of a graph_. consider the uniform wedge sampling of .
some of these wedges will be closed , so we generate a random set of triangles .
each such triangle is an independent , uniform random triangle .
this is because wedges are chosen from the uniform distribution , and every triangles contains exactly @xmath101 closed wedges .
presents the results using triangle sampling to estimate the percentage of triangles where the maximum to minimum degree ratio is @xmath102 , which is motivated by an experiment in @xcite .
the figure shows that accurate results can be achieved by using only 500 triangles and that wedge sampling provides an unbiased selection of triangles .
the expected number of wedges to be sampled to generate @xmath103 triangles is @xmath104 , which means the method will be effective unless the clustering coefficient is extremely small .
doulion @xcite is an alternative sampling mechanism for estimating the number of triangles in a graph .
it has a single parameter @xmath22 .
each edge is chosen independently at random with probability @xmath22 , leading to a subgraph of expected size @xmath105 ( @xmath2 is the total number of edges ) .
we count the number of triangles @xmath106 in this subgraph , and estimate the total number of triangles by @xmath107 .
it is not hard to verify that this is correct in expectation , but bounding the variance requires a lot more work .
some concentration bounds for this estimate are known @xcite , but they depend on the maximum number of triangles incident to an edge in the graph .
so they do not have the direct form of .
some bad examples for doulion have been observed @xcite .
doulion is extremely elegant and simple , and leads to an overall reduction in graph size ( so a massive graph can now fit into memory ) .
common values of @xmath22 used are @xmath108 and @xmath109 . we show that wedge sampling performs at least as good as doulion in terms of accuracy , and has better runtimes .
we run wedge sampling with 32k samples .
we start with setting the doulion parameter @xmath110 , which has been used in the literature .
note that the size of the doulion sample is @xmath111 , which is much larger than 32k .
( for amazon0312 , one of the smaller graphs we consider , @xmath112 . )
we run each algorithm 100 times .
the ( average ) runtimes are compared in , where we see that wedge sampling is always competitive .
the accuracy of the estimate is comparable for both algorithms . in in the appendix
, we present the minimum , maximum , and standard deviation for the 100 runs .
this also shows the good convergence properties of both algorithms , since even the minimum and maximum values are fairly close to the true clustering coefficient .
we also try @xmath113 and note that wedge sampling with 32k samples is still faster in terms of runtime while offering somewhat better accuracy . by setting @xmath114 ,
the sample size of doulion becomes comparable to 32k , but doulion is quite inaccurate ( the range between the maximum and the minimum is large ) . in of the appendix , we also compare results for the local clustering coefficient .
we have proposed a series of wedge - based algorithms for computing various triadic measures on graphs .
our algorithms come with theoretical guarantees in the form of specific error and confidence bounds .
we want to stress that the number of samples required to achieve a specified error and confidence bound is independent of the graph size .
for instance , 38,000 samples guarantees and error less than 0.1% with 99.9% confidence _ for any graph _ , which gives our algorithms an incredible potential for scalability .
the limiting factors have to do with determining the sampling proportions ; for instance , we need to know the degree of each vertex and the overall degree distribution to compute the global clustering coefficient .
the flexibility of wedge sampling along with the hard error bounds essentially redefines what is feasible in terms of graph analysis .
the very expensive computation of clustering coefficient is now much faster ( enabling more near - real - time analysis of networks ) and we can consider much larger graphs than before . in an extension of this work , we are pursuing a mapreduce implementation of this method that scales to o(100 m ) nodes and o(1b ) edges , needing only a few minutes of compute time ( based on preliminary results ) . with triadic analysis
no long being a computational burden , we can extend our horizons into new territories and look at directed triangles , attributed triangles ( e.g. , we might compare the clustering coefficient for `` male '' and `` female '' nodes in a social network ) , evolution of triadic connections , higher - order structures such a 4-cycles and 4-cliques , and so on .
10 , _ patric : a parallel algorithm for counting triangles and computing clustering coefficients in massive networks _ , tech .
report 12 - 042 , ndssl , 2012 .
, _ counting triangles in large graphs using randomized matrix trace estimation _ , in kdd - ldmta10 , 2010 .
, http://dl.acm.org/citation.cfm?id=545381.545464[_reductions in streaming algorithms , with an application to counting triangles in graphs _ ] , in soda02 , 2002 , pp . 623632 . , http://dx.doi.org/10.1145/1401890.1401898[_efficient semi - streaming algorithms for local triangle counting in massive graphs _ ] , in kdd08 , 2008 , pp . 1624 . , http://ngc.sandia.gov/assets/documents/berry-soda11_sand2010-4474c.pdf [ _ listing triangles in expected linear time on power law graphs with exponent at least @xmath115 _ ] , tech .
report sand2010 - 4474c , sandia national laboratories , 2011 . , http://dx.doi.org/10.1145/1142351.1142388[_counting triangles in data streams _ ] , in pods06 , 2006 , pp . 253262 . , http://www.jstor.org/stable/10.1086/421787 [ _ structural holes and good ideas _ ] , american journal of sociology , 110 ( 2004 ) , pp .
349399 . , http://dx.doi.org/10.1137/0214017 [ _ arboricity and subgraph listing algorithms _ ] , siam j. comput . , 14 ( 1985 ) ,
210223 . ,
_ triangle listing in massive networks and its applications _ , in kdd11 , 2011 , pp . 672680 .
, http://dx.doi.org/10.1109/mcse.2009.120[_graph twiddling in a mapreduce world _ ] , computing in science & engineering , 11 ( 2009 ) , pp .
2941 . , http://www.jstor.org/stable/2780243[_social capital in the creation of human capital _ ] , american journal of sociology , 94 ( 1988 ) , pp .
s95s120 . ,
_ degree relations of triangles in real - world networks and graph models _ , in cikm12 , 2012 .
, http://dx.doi.org/10.1073/pnas.032093399[_curvature of co - links uncovers hidden thematic layers in the world wide web _ ] , pnas , 99 ( 2002 ) , pp .
58255829 .
, http://dx.doi.org/10.1145/1753846.1754097[_is a friend a friend ?
: investigating the structure of friendship networks in virtual worlds _ ] , in chi - ea10 , 2010 , pp . 40274032 . , http://www.jstor.org/stable/2282952[_probability inequalities for sums of bounded random variables _ ] , j. american statistical association , 58 ( 1963 ) , pp
. 1330 . , _ new streaming algorithms for counting triangles in graphs _ , in cocoon05 , 2005
, pp . 710716 . , _ efficient triangle counting in large graphs via degree - based vertex partitioning _ , in waw10 , 2010 . , http://dx.doi.org/10.1016/j.tcs.2008.07.017 [ _ main - memory triangle computations for very large ( sparse ( power - law ) ) graphs _ ] , theoretical computer science , 407 ( 2008 ) , pp
. 458473 . , http://dx.doi.org/10.1145/1298306.1298311[_measurement and analysis of online social networks _ ] , in imc07 , acm , 2007 , pp .
2942 . , _ colorful triangle counting and a mapreduce implementation _ , information processing letters , 112 ( 2012 ) , pp
277281 . , http://dx.doi.org/10.1146/annurev.soc.24.1.1 [ _ social capital : its origins and applications in modern sociology _
] , annual review of sociology , 24 ( 1998 ) , pp . 124 . , http://dx.doi.org/10.1007/11427186_54[_finding , counting and listing all triangles in large graphs , an experimental study _ ] , in experimental and efficient algorithms , springer berlin / heidelberg , 2005 , pp . 606609 . , http://dx.doi.org/10.1103/physreve.85.056109[_community structure and scale - free collections of erds - rnyi graphs _ ] , physical review e , 85 ( 2012 ) , p. 056109 . , http://dx.doi.org/10.1145/1963405.1963491[_counting triangles and the curse of the last reducer _ ] , in www11 , 2011 , pp .
607614 . , http://dx.doi.org/10.1109/icdm.2008.72[_fast counting of triangles in large real networks without counting : algorithms and laws _ ] , in icdm08 , 2008 , pp .
608617 . ,
_ spectral counting of triangles in power - law networks via element - wise sparsification _
, in asonam09 , 2009 , pp . 6671 . , _ triangle sparsifiers _ , j. graph algorithms and applications , 15 ( 2011 ) , pp .
703726 . , http://dx.doi.org/10.1145/1557019.1557111[_doulion : counting triangles in massive graphs with a coin _ ] , in kdd 09 , 2009 , pp .
837846 . ,
_ social network analysis : methods and applications _ , cambridge university press , 1994 .
, http://dx.doi.org/10.1038/30918 [ _ collective dynamics of ` small - world ' networks _ ] , nature , 393 ( 1998 ) , pp .
440442 . , http://dx.doi.org/10.1007/978-3-642-24082-9_83[_improved sampling for triangle counting with mapreduce _ ] , in convergence and hybrid information technology , g. lee , d. howard , and d. slezak , eds .
6935 of lecture notes in computer science , springer berlin / heidelberg , 2011 , pp . 685689 . .
available at http://snap.stanford.edu/.
various detailed results are shown in the appendix .
we present the runtime results for global clustering coefficient computations in .
as mentioned earlier , we perform 100 runs of each algorithm and show the minimum , maximum , and standard deviations of the output estimates .
we also show the relative speedup of wedge sampling over doulion with @xmath113 .
we also used doulion to compute the local clustering coefficient . for this purpose
, we predicted the number of triangles incident to a vertex by counting the triangles in the sparsified graph and then dividing this number by @xmath116 .
the results of our experiments are presented in , which shows that doulion fails in accuracy even for @xmath117 .
this is because local clustering coefficient is a finer level statistic , which becomes a challenge for doulion .
wedge sampling on the other hand , keeps its accurate estimations with low variance . & & & & & & & & & & & & + amazon0312 & 401k & 2350k & 69 m & 3686k & 0.160 & 0.163 & 0.161 & 0.160 & 0.261 & 0.004 & 0.007 & 0.016 + amazon0505 & 410k & 2439k & 73 m & 3951k & 0.162 & 0.158 & 0.165 & 0.163 & 0.269 & 0.005 & 0.007 & 0.016 + amazon0601 & 403k & 2443k & 72 m & 3987k & 0.166 & 0.161 & 0.164 & 0.167 & 0.268 & 0.004 & 0.007 & 0.017 + as - skitter & 1696k & 11095k & 16022 m & 28770k & 0.005 & 0.006 & 0.006 & 0.006 & 90.897 & 0.015 & 0.019 & 0.026 + cit - patents & 3775k & 16519k & 336 m & 7515k & 0.067 & 0.064 & 0.067 & 0.068 & 3.764 & 0.035 & 0.040 & 0.056 + roadnet - ca & 1965k & 2767k & 6 m & 121k & 0.060 & 0.061 & 0.058 & 0.058 & 0.095 & 0.018 & 0.022 & 0.037 + web - berkstan & 685k & 6649k & 27983 m & 64691k & 0.007 & 0.005 & 0.006 & 0.007 & 54.777 & 0.007 & 0.009 & 0.016 + web - google & 876k & 4322k & 727
m & 13392k & 0.055 & 0.055 & 0.054 & 0.056 & 0.894 & 0.009 & 0.011 & 0.020 + web - stanford & 282k & 1993k & 3944 m & 11329k & 0.009 & 0.013 & 0.008 & 0.009 & 6.987 & 0.003 & 0.005 & 0.011 + wiki - talk & 2394k & 4660k & 12594 m & 9204k & 0.002 & 0.004 & 0.003 & 0.002 & 20.572 & 0.021 & 0.024 & 0.033 + youtube & 1158k & 2990k & 1474 m & 3057k & 0.006 & 0.005 & 0.006&0.006 & 2.740 & 0.011 & 0.013 & 0.021 + flickr & 1861k & 15555k & 14670 m & 548659k & 0.112 & 0.110 & 0.113 & 0.112 & 567.160&0.019 & 0.026 & 0.051 + livejournal & 5284k & 48710k & 7519 m & 310877k & 0.124 & 0.127 & 0.126 & 0.124 & 102.142 & 0.048 & 0.051 & 0.073 + & & & & & & & & + amazon0312 & 0.421 & 0.427 & 0.417 & 0.420 & 0.301 & 0.001 & 0.002 & 0.008 + amazon0505 & 0.427 & 0.422 & 0.423 & 0.426 & 0.310 & 0.001 & 0.002 & 0.008 + amazon0601 & 0.430 & 0.435 & 0.421 & 0.430 & 0.314 & 0.001 & 0.002 & 0.007 + as - skitter & 0.296 & 0.280 & 0.288 & 0.297 & 88.290 & 0.002 & 0.009 & 0.036 + cit - patents & 0.092 & 0.089 & 0.094 & 0.091 & 4.081 & 0.001 & 0.003 & 0.012 + roadnet - ca & 0.055 & 0.049 & 0.059 & 0.054 & 0.112 & 0.000 & 0.002 & 0.006 + web - berkstan & 0.634 & 0.629 & 0.639 & 0.633 & 53.892 & 0.006 & 0.021 &
0.085 + web - google & 0.624 & 0.624 & 0.619 & 0.628 & 0.996 & 0.001 & 0.004 & 0.013 + web - stanford & 0.629 & 0.612 & 0.635 & 0.633 & 6.868 & 0.002 & 0.010 & 0.038 + wiki - talk & 0.201 & 0.199 & 0.194 & 0.199 & 20.254 & 0.007 & 0.028 & 0.108 + youtube & 0.128 & 0.130&0.132 & 0.131&18.948 & 0.002&0.008 & 0.031 + flickr & 0.375 & 0.369 & 0.371 & 0.377 & 575.493 & 0.001 & 0.006 & 0.021 + livejournal & 0.345 & 0.338 & 0.348 & 0.345 & 102.142 & 0.001 & 0.004&0.015 + & & & & & & & & & & & & & & + amaz0312 & .160 & .155 & .166 & .002 & .
.137 & .188 & .011 & .093 & .234 & .031 & .000 & .392 & .080 & 7.06 + amaz0505 & .162 & .159 & .167 & .002 & .133 & .193 & .011 & .098 & .241 & .028 & .000 & .370 & .081 & 7.37 + amaz0601 & .166 & .162 & .172 & .002 & .140 & .193 & .010 & .088 & .260 & .028 & .000 & .457 & .078 & 7.30 + skitter & .005 & .005 & .007 & .000 & .005 & .006 & .000 & .004 & .007 & .001 & .002 & .008 & .001 & 17.80 + cit - pat & .067 & .064 & .071 & .001 & .060 & .077 & .003 & .042 & .099 & .010 & .027 & .116 & .022 & 15.76 + road - ca & .060 & .058 & .065 & .001 & .000 & .133 & .023 & .000 & .250 & .062 & .000 & 1.001 & .193 & 5.84 + w - bersta & .007 & .006 & .008 & .001 & .006 & .008 & .000 & .006 & .008 & .000 & .004 & .011 & .001 & 16.28 + w - google & .055 & .052 & .059 & .001 & .051 & .060 & .002 & .044 & .071 & .005 & .021 & .107 & .016 & 1.20 + w - stan & .009 & .007 & .010 & .001 & .007 & .010 & .001 & .006 & .012 & .001 & .003 & .018 & .003 & 7.18 + wiki - t & .002 & .002 & .003 & .000 & .002 & .002 & .000 & .002 & .003 & .000 & .001 & .004 & .001 & 8.52 + youtube & .006 & .005 & .007 & .000 & .005 & .007 & .001 & .004 & .010 & .001 & .000 & .018 & .004 & 7.55 + flickr & .112 & .108 & .117 & .002 & .110 & .115 & .001 & .107 & .118 & .002 & .098 & .125 & .005 & 11.38 + livejour & .124 & .120 & .128 & .002 & .121 & .128 & .001 & .116 & .130 & .003 & .105 & .143 & .007 & 3.49 + & & & & & & & & & & & & & + amaz0312 & .421 & .415 & .426 & .003 & .395 & .463 & .014 & .318 & .558 & .047 & .195 & .869 & .140 + amaz0505 & .427 & .421 & .432 & .003 & .395 & .463 & .014 & .338 & .568 & .056 & .195 & .869 & .140 + amaz0601 & .430 & .423 & .437 & .003 & .403 & .466 & .013 & .329 & .633 & .058 & .239 & .819 & .120 + skitter & .296 & .288 & .303 & .003 & .272 & .322 & .011 & .206 & .384 & .039 & .122 & .667 & .105 + cit - pat & .092 & .088 & .096 & .002 & .085 & .099 & .003 & .066 & .126 & .011 & .028 & .199 & .035 + road - ca & .055 & .052 & .058 & .001 & .038 & .071 & .006 & .000 & .118 & .022 & .000 & .279 & .062 + w - bersta & .634 & .627 & .641 & .003 & .586 & .700 & .024 & .532 & .808 & .057 & .339 & 1.273 & .165 + w - google & .624 & .615 & .630 & .003 & .580 & .688 & .021 & .471 & .772 & .063 & .335 & 1.276 & .183 + w - stan & .629 & .622 & .636 & .003 & .532 & .737 & .038 & .441 & .982 & .109 & .218 & 1.218 & .230 + wiki - t & .201 & .195 & .206 & .002 & .171 & .240 & .015 & .055 & .379 & .067 & .006 & .609 & .160 + youtube & .128 & .167 & .179 & .002 & .128 & .218 & .015 & .055 & .294 & .057 & .007 & .767 & .182 + flickr & .375 & .368 & .381 & .003 & .316 & .411 & .016 & .212 & .553 & .066 & .086 & .774 & .154 + livejour & .345 & .337 & .355 & .003 & .330 & .359 & .006 & .296 & .401 & .023 & .214 & .500 & .060 + | graphs are used to model interactions in a variety of contexts , and there is a growing need to quickly assess the structure of a graph .
some of the most useful graph metrics , especially those measuring social cohesion , are based on _
triangles_. despite the importance of these triadic measures , associated algorithms can be extremely expensive .
we propose a new method based on _
wedge sampling_. this versatile technique allows for the fast and accurate approximation of all current variants of clustering coefficients and enables rapid uniform sampling of the triangles of a graph .
our methods come with _
provable _ and practical time - approximation tradeoffs for all computations .
we provide extensive results that show our methods are orders of magnitude faster than the state - of - the - art , while providing nearly the accuracy of full enumeration .
our results will enable more wide - scale adoption of triadic measures for analysis of extremely large graphs , as demonstrated on several real - world examples . |
atomic motion is an unavoidable element in the consideration of any amo system and an integral part of experimental designs in atom trapping devices . at issue here is the interaction between the internal degrees of freedom of an atom , assumed to contain an effective two - level system ( qubit ) , and the electromagnetic field ( emf ) , modified by the atom s quantal motional degree of freedom .
this problem has two aspects : 1 ) how does the two level activity affect the atomic motion ? and 2 ) how does atomic motion affect the two level activity ?
the first aspect is the basis for laser cooling and atom trapping , which have been studied in great detail and successfully implemented by well - known experiments ( for reviews see @xcite ) .
this paper is aimed at the second aspect , specifically , how quantized motion affects the qubit - emf system dynamics , which is of interest in the design of quantum computers based on atomic qubits ( in the form of a neutral atom @xcite or ion @xcite ) in a qed cavity or optical potential .
effects on internal dynamics due to quantized center of mass ( com ) motion have previously been studied in the situations of an atom in free space @xcite , in a cavity @xcite , and when the atom s qubit and com degrees of freedom are entangled @xcite . however ,
all have focused on spontaneous emission rather than decoherence .
the present work probes the non - markovian regime of atom - emf interaction , under the modest aim of explicitly computing how entanglement with quantized motion through recoil affects the decoherence and relaxation rates of an atomic qubit in free space . in order to achieve that end ,
we first discuss two issues of importance in computations of _ coherent _ reduced dynamics , using path integral methods . + * the importance of including back - action * + it is well - known that the interaction between a two - level system ( 2ls , or qubit ) and the emf is the primary source of its relaxation and decoherence , while effects associated with the atom s motional degrees of freedom are usually relegated to the background . assuming that the atom moves adiabatically limits one s consideration to those circumstances wherein the external degrees of freedom act merely as a passive parameter in the environment ( here comprised of the emf and atomic motion ) of our system ( the qubit ) , with no dynamical interplay . in technical terms , this amounts to a ` test - field ' approximation that the qubit lives in a fixed environment defined by a set of parameters , amongst them the adiabatic motion .
the test field approximation leaves out effects of changes in the environment on the system . to include the effects of the environmental variables
_ dynamically _ it is essential to perform a self - consistent back - action calculation .
this was done for the effect of a cavity emf on the 2la in ref .
@xcite . + * full coherence requires self - consistent treatment * + in tackling problems where many factors enter , it is useful to isolate one factor after another so that the remaining factors of interest to us can be simplified enough to yield some solution . for quantum coherence and entanglement such simplifications can lead to erroneous results , since phase information is lost if one artificially isolates the linking components of the complete quantum system
this brings up the necessity of self - consistency in any treatment of quantum coherence and entanglement issues .
in the present case of a qubit in an emf this requires that the fully entangled system of atomic 2ls ( internal ) , the emf , and the center of mass ( external ) degrees of freedom be treated coherently as a whole and each factor involved be allowed to evolve under the influence of the others in a self - consistent manner .
this self - consistency requirement leads to non - markovian dynamics since memory effects arise naturally and are necessary to preserve maximal coherence during the evolution . + * grassmannian and coherent state representation of influence functional * + a theoretical scheme we found satisfactory in meeting these requirements is the influence functional ( if ) formalism of feynman and vernon @xcite or the related closed - time - path ( ctp ) effective action of schwinger and keldysh @xcite .
the influence of the environmental variables on the system of interest is incorporated in the if ( or effective action ) in such a way that the equations of motion obtained for the system will already have included the back - action of the environmental variables on the system in a dynamically self - consistent manner .
this scheme has been applied to a two - level atom ( 2la ) interacting with an electromagnetic field ( emf ) in reference @xcite . there , a first - principles derivation of the general master equations is given and applied to the study of the decoherence of a 2la in an emf , for the cases of a free quantum field and a cavity field in the vacuum at zero temperature .
the authors found that for the standard resonant type of coupling characteristic of such systems the decoherence time is close to the relaxation time .
here we use the influence functional method for the treatment of the back - action of the quantum field and the quantal motion of the atom on the qubit . in section
ii we compute the transition amplitude between an initial and final state using a coherent state label for the ( bosonic ) states of the emf and a grassmannian for the ( fermionic ) 2ls .
the coherent state basis allows us to identify the hilbert space of states with a space of coherent states .
the sum over all quantum evolutions is then a sum over all paths in this space .
once the transition amplitude is computed in some sufficiently simplified form , forward and backward versions can be combined and reduced to form the reduced density matrix evolutionary operator . in section iii we calculate the evolutionary operator for the reduced density matrix when the emf and motional degrees of freedom are integrated over .
we derive an equation describing the evolution of the on and off - diagonal elements , the latter is the coherence function we seek .
we end in section iv with a discussion of our results and comments on possible further developments on this subject .
our system is a 2-level atom interacting with its own center of mass ( com ) motion and the emf .
we begin with a modified multi - mode jaynes - cummings type hamiltonian ( e.g. dipole and rotating wave approximation , see appendix a of ref@xcite ) , @xmath0.\ ] ] the first term in the hamiltonian is the com kinetic energy .
the next two terms are the qubit and emf energies , respectively .
the last two terms are the interaction between the qubit , emf , and the atom s com degree of freedom .
note that @xmath1 and @xmath2 are both operators .
coupling of the qubit to its com motion is through the spatial mode functions of the emf .
we shall restrict our consideration to an initial vacuum emf at zero temperature .
the result of this calculation will thus be the modification of the vacuum decoherence and relaxation rates of a qubit when the effects of quantized atomic motion are included . the first step towards obtaining the reduced system dynamics while retaining the full system s coherence is to compute the transition amplitudes between the initial and final states which are the matrix elements of the evolution operator of the full system .
we do this with coherent state path integrals . for the emf
we use a bosonic coherent state representation and for the 2-level system ( qubit ) degree of freedom we use the grassmannian coherent states @xcite .
coherent states are by definition generated by the exponentiated operation of the creation operator and a suitable label on a chosen fiducial state : @xmath3 in the case of bosonic coherent states defined in eq .
( [ emcoherentstates ] ) the label , @xmath4 , is a complex number , and in the case of the grassmann coherent states defined in eq .
( [ gmcoherentstates ] ) the label , @xmath5 , is an anti - commuting number .
the chosen fiducial states are the emf vacuum and the lower 2-level state , respectively . in order for any set of states to be useful for the decomposition of the transition matrix they must have a resolution of unity .
the emf and grassmannian coherent states have the following decompositions of unity @xmath6 with the measures @xmath7 grassmann coherent states also share other well known properties of coherent states such as being non - orthogonal and eigenstates of the annihilator : @xmath8 the center of mass or external degree of freedom can be represented in either the position or momentum basis . in the coherent state basis
the hamiltonian eq .
( [ hamiltonian ] ) can be written in its q - representation @xcite as [ cf eq .
( 2.8 ) of @xcite ] @xmath9.\end{aligned}\ ] ] the transition matrix elements between the initial and final coherent states are then @xmath10 using the completeness property of the ( emf and grassmann ) coherent state basis to facilitate time - discretization of the transition matrix @xcite puts the transition matrix elements in a coherent state path integral representation . after inserting the q - representation , the transition elements transform into a sum over paths in the coherent state labels .
having done the above the transition matrix becomes a triple functional integral : @xmath11 \mbox { } { { \rm e}}^{{{{\rm i } } } { \omega}_o t/2 } \nonumber\\ \times \exp\bigg[-\int_0^t \big({\bar{\eta}}\dot{\eta } + { { { \rm i}}}{\omega}_o { \bar{\eta}}\eta & + & \sum_{{\bf k}}{\bar{z}}_{{\bf k}}\dot{z}_{{\bf k}}+{{{\rm i}}}\sum_{{\bf k}}{\omega}_{{\bf k}}{\bar{z}}_{{\bf k}}z_{{\bf k}}{{{\rm i}}}\sum_{{\bf k}}{\bar{\eta}}{{\rm g}}_{{\bf k}}(\bx ) z_{{\bf k}}+{{{\rm i}}}\sum_{{\bf k}}{\bar{z}}_{{\bf k}}{\bar{\rm g}}_{{\bf k}}(\bx ) \eta \big ) { { { \rm d } } } s \bigg]\end{aligned}\ ] ] in this form the transition matrix elements can be evaluated exactly by a combination of stationary phase and correlation function methods which exploit the truncating properties of grassmann variables .
the order of evaluation will be the emf , com , and then grassmann functional integrals .
the details follow .
first , the emf coherent state part of the triple path integral can be evaluated by the stationary phase method @xcite .
the variational equations of motion for the electromagnetic field variables in eq .
( [ transamp1 ] ) are @xmath12 which have integral solutions [ cf eq .
( 2.14 ) of @xcite ] @xmath13 the transition amplitude from an initial emf vacuum ( @xmath14 ) to an arbitrary final state becomes @xmath15 { { \rm e}}^{{{{\rm i } } } { \omega}_o t/2 } \nonumber\\ & \times&\prod_{{\bf k}}\exp\bigg[-{{{\rm i}}}\int_0^t { { { \rm d } } } s { \bar{\rm g}}_{{\bf k}}(\bx(s ) ) e^{-{{{\rm i}}}{\omega}_{{\bf k}}(t - s ) } { \bar{z}}_{f{{\bf k}}}\mbox { } \eta(s ) -\int_0^t { { { \rm d } } } s \int_0^s { { { \rm d } } } r { { \rm g}}_{{\bf k}}(\bx(s ) ) { \bar{\rm g}}_{{\bf k}}(\bx(r ) ) e^{-{{{\rm i}}}{\omega}_{{\bf k}}(s - r ) } { \bar{\eta}}(s)\eta(r)\bigg].\end{aligned}\ ] ] the path integral for the emf degrees of freedom is now complete .
second , the position path integral can be evaluated as a set of 0 , 1 and 2 point functions .
note in the transition amplitude of eq .
( [ transamp2 ] ) that since the emf is taken to be in an initial vacuum , any source term for @xmath16 will be proportional to @xmath17 .
the variational equation of motion derived from eq .
( [ transamp2 ] ) for @xmath16 yields @xmath18 with the boundary condition @xmath19 .
therefore @xmath20 .
we use this to expand the exponent in the transition amplitude of eq .
( [ transamp2 ] ) . due to the nilpotency of the grassmann variables
( i.e. @xmath21 ) it will truncate after the first term in the expansion . after expanding and truncating the integrand ,
the position path integral is @xmath22 \\ \times\bigg[1 & -&{{{\rm i}}}\int_0^t { { { \rm d } } } s \sum_{{\bf k}}{\bar{\rm g}}_{{\bf k}}(\bx(s ) ) e^{-{{{\rm i}}}{\omega}_{{\bf k}}(t - s ) } { \bar{z}}_{f{{\bf k}}}\mbox { } \eta(s ) -\int_0^t { { { \rm d } } } s \int_0^s { { { \rm d } } } r \sum_{{\bf k}}{{\rm g}}_{{\bf k}}(\bx(s ) ) { \bar{\rm g}}_{{\bf k}}(\bx(r ) ) e^{-{{{\rm i}}}{\omega}_{{\bf k}}(s - r ) } { \bar{\eta}}(s)\eta(r)\bigg ] .
\nonumber\end{aligned}\ ] ] there are thus three correlation functions which need to be computed . first the spatial mode functions must be chosen in order to specify the targeted correlation functions .
for an electromagnetic field in free space ( no cavity or boundaries ) @xmath23 the correlations functions are computed in appendix a. substituting these expressions back into the eq .
( [ transamp2 ] ) gives for the transition amplitude @xmath24 \bigg(\frac{m}{2\pi{{{\rm i } } } t}\bigg)^{3/2 } { { \rm e}}^{{{{\rm i } } } { \omega}_o t/2 } e^{-{{{\rm i}}}{\omega}_{{\bf k}}(s - r ) } \nonumber\\ \times\bigg\{1 & -&{{{\rm i}}}\int_0^t { { { \rm d } } } s \sum_{{\bf k}}\frac{{\lambda}}{\sqrt{{\omega}_{{\bf k } } } } \exp\bigg[{{{\rm i}}}\frac{s}{t}{{\bf k}}\cdot(\bx_f-\bx_i ) -\frac{{{{\rm i}}}}{2 m } \frac{s(t - s)}{t } { { \bf k}}^2 \bigg ] e^{-{{{\rm i}}}{\omega}_{{\bf k}}(t - s ) } { \bar{z}}_{f{{\bf k}}}\mbox { } \eta(s ) \\
& -&\int_0^t { { { \rm d } } } s \int_0^s { { { \rm d } } } r \sum_{{\bf k}}\frac{{\lambda}^2}{{\omega}_{{\bf k } } } \exp\bigg[-{{{\rm i}}}\frac{s - r}{t}{{\bf k}}\cdot(\bx_f-\bx_i ) -\frac{{{{\rm i}}}}{2m}\frac{(t-(s - r))(s - r)}{t}{{\bf k}}^2\bigg ] { \bar{\eta}}(s)\eta(r)\bigg\}. \nonumber\end{aligned}\ ] ] the path integral for the external degrees of freedom is now complete .
finally , the grassmann variable path integral can be evaluated along its stationary path .
the variational equation of motion for the grassmann field variable in eq .
( [ transamp3 ] ) is @xmath25 with the definition : @xmath26.\ ] ] note that the final time @xmath27 enters as a parameter in the variational equation of motion just as the mass or position do .
the reason for this is that the above variational equation of motion is for the evolution of the atom from an initial time to a final time , so the time is an explicit parameter . rewriting the above variational equation in laplace space allows the non - local integral part to be transformed with the convolution theorem .
the solution is in terms of an inverse laplace transform , @xmath28 with the definition : @xmath29 { { { \rm d } } } z.\ ] ] the solution thus becomes a contour integral .
the pole of the denominator in eq .
( [ soln1 ] ) can be found to @xmath30 @xmath31 finding the pole to order @xmath30 gives a solution to the same order : @xmath32 \bigg\}.\end{aligned}\ ] ] evaluating the transition amplitude along its stationary path with the second order pole approximation yields an expression for the transition matrix that is second order in its action : @xmath33 \\ \exp\bigg[{\bar{\eta}}_f \eta_t(t ) & -&{{{\rm i}}}\int_0^t { { { \rm d } } } s \sum_{{\bf k}}\frac{{\lambda}}{\sqrt{{\omega}_{{\bf k } } } } \exp\big[-{{{\rm i}}}{\omega}_{{\bf k}}(t - s ) + { { { \rm i}}}\frac{s}{t}{{\bf k}}\cdot(\bx_f-\bx_i ) -\frac{{{{\rm i } } } s(t - s)}{2mt}{{\bf k}}^2\big ] { \bar{z}}_{f{{\bf k}}}\mbox { } \eta_t(s)\bigg ] .
\nonumber\end{aligned}\ ] ] all three functional integrals are now evaluated . in the next section
we proceed to derive the evolutionary operator for the density matrix by combining the transition amplitudes into a closed loop .
at this point the expression of eq .
( [ transamp4 ] ) for the transition amplitude can be combined with its counterpart propagating backwards in time and traced over all final emf states .
the result gives the evolutionary operator for the reduced density matrix ( we may call it the reduced propagator , for short ) , @xmath34 and is formed by integrating out the environmental variables which in our case are the emf and the atom s motional degrees of freedom .
the evolution of the qubit density matrix elements with back - action from the emf and the atomic motion can be calculated from the reduced propagator @xmath35 the functions @xmath36 and @xmath37 are initial states for the 2-level atomic and external degrees of freedom , respectively : @xmath38 @xmath39 the function @xmath40 is the initial ( external ) center of mass wavefunction of the atom . from eq .
( [ evolved1 ] ) the on and off - diagonal components of the reduced density matrix elements evolved to time @xmath27 are given by @xmath41 @xmath42 the emf , as previously stated , is in a vacuum state , but the choice of an initial center of mass wavefunction has not yet been made . to closely model an atom with fixed position and momentum
, we use a minimum uncertainty gaussian wavefunction centered at @xmath43 .
@xmath44\end{aligned}\ ] ] such an initial wavefunction simplifies the expressions for the diagonal and off - diagonal matrix elements of the qubit .
the result for the off - diagonal components which measures the coherence of the qubit under such conditions is shown here : @xmath45\end{aligned}\ ] ] the function @xmath46 is given by eq .
( [ ufunction ] ) with @xmath47 .
the evolution of the coherence function is found to follow an exponential decay with a decay rate slightly faster than in the infinite mass case .
the percentage change in the decoherence rate of the off - diagonal versus the the stationary qubit case is plotted in fig .
the decay rate increases with decreasing mass and matches the stationary qubit result given by @xcite in the limit of infinite mass .
we expect that a qubit in a smaller mass object is more affected by recoil than a qubit in heavy mass .
the variation in the decoherence rate with changes in the external wavefunction size is relatively flat and can not reliably be resolved with the available computing power and machine accuracy .
we find that so long as the resonant frequency is small enough or the mass large enough that the atomic recoil velocity is non - relativistic , which is where this theory is valid , then the motional decoherence will contribute negligibly to the decay of the qubit . a plot of the percentage increase in the decoherence of the off - diagonal matrix elements of the reduced density matrix versus the non - dimensionalized mass ( @xmath48 ) .
the decoherence rate increases as the mass of the atom containing the qubit is decreased . as the mass
is increased the decoherence rate asymptotes to the value of a stationary atom obtained by anastopoulos and hu @xcite .
this is consistent with a smaller mass qubit being more affected by its recoil than a heavy mass qubit .
typical experimental parameters fall to the right end of the shown plot . ]
often , one may separate the dynamics of an atom s motion from those of its internal degrees of freedom by arguing that the time scales associated with the motion of the atom are much longer than those of the two level activity .
this is the rationale behind the adiabatic approximation adopted for most considerations of the atomic dynamics .
however , coherence requirements in quantum computing implementations may prompt one to question this assumption .
one aim of our investigation is to test for non - adiabatic effects in atomic quantum computing schemes .
another is to describe the effect of recoil from the emission and re - absorption of virtual particles in the atom - emf interaction upon the center of mass motion .
these two problems correspond to the two aspects described in the introduction . here
we consider the second aspect mentioned above , aiming at the effect of quantum motional decoherence of the qubit , i.e. , the back - action of atomic motion on a two level system in free space as mediated by the emf .
we find that the inclusion of the external degrees of freedom only slightly alters the decoherence and relaxation rates as compared to a stationary atom .
typical experimental parameters fall to the right end in the plot of fig .
( 1 ) . a rubidium atom used as
a qubit would have a non - dimensionalized mass of approximately @xmath49 , which places it in a regime in which the effect of motion - induced decoherence is negligible .
for optical qubit transition frequencies in general , motion - induced decoherence will not be a factor unless the mass of the qubit is four to five orders of magnitude smaller than the mass of a typical alkali atom .
one can conclude tentatively that in general amo implementations , motion - induced decoherence of a free qubit is negligibly small . since the calculation done here is coherent and non - markovian , one can view our result as confirming the validity of the adiabatic approximation in alkali atom qubits .
although the result of the calculation is the expected one , the technique described here is the first able to compute the decoherence of a qubit coupled to its own quantized com without any form of markovian approximation , while allowing the qubit - emf coupling to be the non - linear form derived from the emf spatial mode functions .
useful applications of this method will include any situations in which the com motion of an atom back - acts onto its internal qubit dynamics and the full multi - mode structure of the emf is relevant .
two such examples , as drawn from the references cited in the introduction , are an atomic qubit in a cavity and an atom with entangled qubit and emf degrees of freedom . in the former ,
the presence of the cavity walls increases the cavity mode recoils on the atom @xcite .
the latter is at the center of certain two qubit gate implementations @xcite , with the question there being how well coherence is maintained when a qubit is entangled both internally and externally .
calculation in that case can provide an important feasibility test of quantum computing applications which utilize such entanglement . + * acknowledgements * this work is supported in part by nsf grant phy98 - 00967 and by a contract under arda .
the position path integral which needs to be evaluated is : @xmath22 \\
\times\exp\bigg[&-&{{{\rm i}}}\int_0^t { { { \rm d } } } s \sum_{{\bf k}}{\bar{\rm g}}_{{\bf k}}(\bx(s ) ) e^{-{{{\rm i}}}{\omega}_{{\bf k}}(t - s ) } { \bar{z}}_{f{{\bf k}}}\mbox { } \eta(s ) -\int_0^t { { { \rm d } } } s \int_0^s { { { \rm d } } } r \sum_{{\bf k}}{{\rm g}}_{{\bf k}}(\bx(s ) ) { \bar{\rm g}}_{{\bf k}}(\bx(r ) ) e^{-{{{\rm i}}}{\omega}_{{\bf k}}(s - r ) } { \bar{\eta}}(s)\eta(r)\bigg ] \nonumber\end{aligned}\ ] ] which can be expanded and truncated to : @xmath22 \\ \times\bigg[1 & -&{{{\rm i}}}\int_0^t { { { \rm d } } } s \sum_{{\bf k}}{\bar{\rm g}}_{{\bf k}}(\bx(s ) ) e^{-{{{\rm i}}}{\omega}_{{\bf k}}(t - s ) } { \bar{z}}_{f{{\bf k}}}\mbox { } \eta(s ) -\int_0^t { { { \rm d } } } s \int_0^s { { { \rm d } } } r \sum_{{\bf k}}{{\rm g}}_{{\bf k}}(\bx(s ) ) { \bar{\rm g}}_{{\bf k}}(\bx(r ) ) e^{-{{{\rm i}}}{\omega}_{{\bf k}}(s - r ) } { \bar{\eta}}(s)\eta(r)\bigg ] \nonumber\end{aligned}\ ] ] there are thus three correlation functions which need to be computed . first the spatial mode functions must be chosen in order to specify the targeted correlation functions . for an electromagnetic field in free space ( no cavity or boundaries ) @xmath23 the correlations functions are : @xmath50 = \bigg(\frac{m}{2\pi{{{\rm i } } } t}\bigg)^{3/2 } \exp\bigg[\frac{{{{\rm i } } } m}{2 t } ( \bx_f -\bx_i)^2\bigg]\ ] ] @xmath51 = \bigg(\frac{m}{2\pi{{{\rm i } } } t}\bigg)^{3/2 } \exp\bigg[\frac{{{{\rm i } } } m}{2 t } ( \bx_f -\bx_i)^2 + { { { \rm i}}}\frac{s}{t}{{\bf k}}\cdot(\bx_f-\bx_i ) -\frac{{{{\rm i}}}}{2m}\frac{s(t - s)}{t}{{\bf k}}^2\bigg ] \\\ ] ] @xmath52 \\ & = & \bigg(\frac{m}{2\pi{{{\rm i } } } t}\bigg)^{3/2 } \exp\bigg[\frac{{{{\rm i } } } m}{2 t } ( \bx_f -\bx_i)^2 -{{{\rm i}}}\frac{s - r}{t}{{\bf k}}\cdot(\bx_f-\bx_i ) -\frac{{{{\rm i}}}}{2m}\frac{(t-(s - r))(s - r)}{t}{{\bf k}}^2\bigg ] \nonumber\end{aligned}\ ] ] substituting these expressions back into the eq .
( [ transamp2 ] ) gives for the transition amplitude : @xmath53 { { \rm e}}^{{{{\rm i } } } { \omega}_o t/2 } \bigg(\frac{m}{2\pi{{{\rm i } } } t}\bigg)^{3/2 } \exp\bigg[\frac{{{{\rm i } } } m}{2 t } ( \bx_f -\bx_i)^2\bigg]e^{-{{{\rm i}}}{\omega}_{{\bf k}}(s - r ) } \nonumber\\ \times\bigg[1 & -&{{{\rm i}}}\int_0^t { { { \rm d } } } s \sum_{{\bf k}}\frac{{\lambda}}{\sqrt{{\omega}_{{\bf k } } } } \exp\bigg[{{{\rm i}}}\frac{s}{t}{{\bf k}}\cdot(\bx_f-\bx_i ) -\frac{{{{\rm i}}}}{2m}\frac{s(t - s)}{t}{{\bf k}}^2\bigg ] e^{-{{{\rm i}}}{\omega}_{{\bf k}}(t - s ) } { \bar{z}}_{f{{\bf k}}}\mbox { } \eta(s ) \\ & -&\int_0^t { { { \rm d } } } s \int_0^s { { { \rm d } } } r \sum_{{\bf k}}\frac{{\lambda}^2}{{\omega}_{{\bf k } } } \exp\bigg[-{{{\rm i}}}\frac{s - r}{t}{{\bf k}}\cdot(\bx_f-\bx_i ) -\frac{{{{\rm i}}}}{2m}\frac{(t-(s - r))(s - r)}{t}{{\bf k}}^2\bigg ] { \bar{\eta}}(s)\eta(r)\bigg ] \nonumber\end{aligned}\ ] ] + j. schwinger , j. math .
* 2 * ( 1961 ) 407 ; l. v. keldysh , zh . eksp .
fiz . * 47 * , 1515 ( 1964 ) [ engl
. trans.sov .
jept * 20 * , 1018 ( 1965 ) ] .
g. zhou , z. b. su , b. hao and l. yu , phys .
rep . * 118 * , 1 ( 1985 ) ; z. b. su , l. y. chen , x. t. yu and k. c. chou , phys . rev . * b37 * , 9810 ( 1988 ) .
e. calzetta and b. l. hu , phys .
rev . * d40 * , 656 ( 1989 ) .
ohnuki y , kashiwa t 1978 coherent states of fermi operators and the path integral _ coherent states : applications in physics and mathematical physics
_ eds j klauder and b skagerstam ( singapore : world scientific ) 449 - 465 . | the reduced dynamics of an atomic qubit coupled both to its own quantized center of mass motion through the spatial mode functions of the electromagnetic field , as well as the vacuum modes , is calculated in the influence functional formalism .
the formalism chosen can describe the entangled non - markovian evolution of the system with a full account of the coherent back - action of the environment on the qubit .
we find a slight increase in the decoherence due to the quantized center of mass motion and give a condition on the mass and qubit resonant frequency for which the effect is important . in optically resonant alkali atom systems we find the effect is negligibly small .
the framework presented here can nevertheless be used for general considerations of the coherent evolution of qubits in moving atoms in an electromagnetic field . |
the nearby radio galaxy m87 accompanies one of the best studied agn jets .
its proximity ( 16.7 mpc ) and brightness have enabled detailed studies of this jet over decades through radio , optical and to x - ray at tens of parsec scale resolutions .
furthermore , the inferred very massive black hole ( @xmath2 ) yields a linear resolution down to @xmath3 schwarzschild radii @xmath4 ( for @xmath5 ) , making this source an ideal case to probe the relativistic - jet formation at an unprecedented compact scale with very - long - baseline - interferometer ( vlbi ) observations ( e.g. , ly et al .
2007 ; kovalev et al .
2007 ; hada et al . 2011 ; asada & nakamura 2012 ; doeleman et al .
2012 ; hada et al .
m87 is now widely known to show @xmath0-ray emission up to the very - high - energy ( vhe ; @xmath6 gev ) regime , where this source often exhibits active flaring episodes . the location and
the physical processes of such emission have been a matter of debate over the past years , and there are two candidate sites which can be responsible for the vhe @xmath0-ray production .
one is a very active knot hst-1 which is located at more than 100 pc from the nucleus ( stawarz et al .
2006 ; cheung et al .
2007 ; harris et al . 2009 ) .
this argument is based on the famous vhe flare event in 2005 , where hst-1 underwent a large radio - to - x - ray outburst jointly with a vhe flare .
in contrast , the other candidate is the core / jet base , which is very close to the central black hole .
this argument is based on the vhe event in 2008 , where the core / vhe showed a remarkable correlation in the light curves .
there was another vhe event in 2010 , but this is rather elusive .
coincident with the vhe event , chandra detected an enhanced flux from the x - ray core ( harris et al .
2011 ; abramowski et al .
2012 ) , and vlba observations also suggested a possible increase of the radio core flux ( hada et al .
however , giroletti et al .
( 2012 ) found the emergence of a superluminal component in the hst-1 complex near the epoch of this event , which is reminiscent of the 2005 case .
recently , the veritas collaboration has reported new vhe @xmath0-ray activity from m87 in early 2012 ( beilicke et al .
2012 ) . while there were no remarkable flares like those in the previous episodes , the vhe flux in 2012 clearly exhibits an elevated state at a level of @xmath7@xmath8 ( @xmath9(0.20.3)@xmath10 photons @xmath11 s@xmath12 ) over the consecutive two months from february to march 2012 .
the observed flux is a factor of @xmath72 brighter than that in the neighboring quiescent periods .
therefore , this event provides another good opportunity for exploring the location of the vhe emission site by jointly using high - resolution instruments .
here we report a multi - wavelength radio and mev / gev study of the m87 jet during this period using the vlbi exploration radio astrometry ( vera , fegure 1 ) , the european vlbi network ( evn ) , the submillimeter array ( sma ) and the fermi - lat .
we especially focus on the vlbi data in the radio bands ; with vera , we obtained the high - angular - resolution , dense - sampling - interval , phase - referencing data set at 22 and 43 ghz during the vhe activity in 2012 : with the supportive evn monitoring , we obtained a complementary data set at 5 ghz , which enables a high - sensitivity imaging of the m87 jet .
a collective set of these radio data allows us to probe the detailed physical status and structural evolutions of m87 by pinpointing the candidate sites of the @xmath0-ray emission i.e. , the core and hst-1 . for more details regarding the radio data analysis ,
see hada et al .
( 2014 ) .
the lat data reported here were collected from 2011 february 1 ( mjd 55593 ) to 2012 september 30 ( mjd 56200 ) . during this time ,
the fermi observatory operated almost entirely in survey mode .
the analysis was performed with the ` sciencetools ` software package version v9r32p5 .
the lat data were extracted within a @xmath13 region of interest centred at the radio location of m87 .
only events belonging to the ` source ' class were used .
the time intervals when the rocking angle of the lat was greater than 52@xmath14 were rejected .
in addition , a cut on the zenith angle ( @xmath15 ) was applied to reduce contamination from the earth limb @xmath0 rays , which are produced by cosmic rays interacting with the upper atmosphere .
the spectral analysis was performed with the instrument response functions ` p7rep_source_v15 ` using an unbinned maximum - likelihood method implemented in the science tool ` gtlike ` .
a galactic diffuse emission model and isotropic component , which is the sum of an extragalactic and residual cosmic ray background , were used to model the background .
the normalizations of both components in the background model were allowed to vary freely during the spectral fitting .
we evaluated the significance of the @xmath0-ray signal from the sources by means of the maximum - likelihood test statistic ts = 2@xmath16log(likelihood ) between models with and without a point source at the position of m87 ( mattox et al .
the source model used in ` gtlike ` includes all of the point sources from the second fermi - lat catalog ( 2fgl ; nolan et al .
2012 ) that fall within @xmath17 of the source .
the spectra of these sources were parametrized by power - law functions , except for 2fglj1224.9 + 2122 ( 4c21.35 ) and 2fglj1229.1 + 0202 ( 3c273 ) , for which we used a log - parabola as in the 2fgl catalogue .
a first maximum - likelihood analysis was performed to remove from the model the sources having ts @xmath18 25 and/or the predicted number of counts based on the fitted model @xmath19 .
a second maximum - likelihood analysis was performed on the updated source model . in the fitting procedure ,
the normalization factors and the photon indices of the sources lying within 10@xmath14 of m87 were left as free parameters .
for the sources located between 10@xmath14 and 15@xmath14 , we kept the normalization and the photon index fixed to the values from the 2fgl catalogue . integrating over the period from 2011 february 1 to 2012 september 30 ( mjd 5559356200 ) , the fit with a power - law model in the 0.1100 gev energy range results in a ts = 134 , with an integrated average flux of ( 2.22 @xmath20 0.43 ) @xmath2110@xmath22 ph @xmath11 s@xmath12 and a photon index of @xmath23 = 2.25 @xmath20 0.10 .
taking into account the detection significance over the whole analysed period , we produced the @xmath0-ray light curves with 1-month and 2-month time bins .
this choice of binning is compatible with those adopted in the previous m87 studies with lat data ( abdo et al .
2009 ; abramowski et al .
2012 ) , and also reasonable for a comparison with the observed month - scale vhe activity in 2012 . for each time bin ,
the spectral parameters for m87 and for all the sources within 10@xmath14 from it were frozen to the value resulting from the likelihood analysis over the entire period . in the light curve with the 2-month time bins , if ts @xmath18 10 , 2@xmath24 upper limits were evaluated , while only bins with ts @xmath1 10 are selected in the light curve with the 1-month time bins .
we describe the results of the lat light curves in section 4.2 . dividing the 1-month bins with higher flux in 5-day sub - bins , the highest flux of ( [email protected])@xmath2110@xmath22 and (
[email protected])@xmath2110@xmath22 ph @xmath11 s@xmath12 was detected on 2011 october 12 - 16 and 2012 january 16 - 20 , respectively ( these sub - bin data also show ts@xmath110 ) . by means of the ` gtsrcprob ` tool
, we estimated that the highest energy photon emitted from m87 ( with probability @xmath1 90% of being associated with the source ) was observed by lat on 2011 april 7 , at a distance of 0.09@xmath14 from the source and with an energy of 254.0 gev , extending into the vhe range . in figure
2 we show a combined set of light curves of m87 from radio to mev / gev @xmath0-ray between mjd 55400 and mjd 56280 .
thanks to the dense , complementary coverages of vera and evn , we revealed the detailed evolutions of the radio light curves for both the core and hst-1 .
the most remarkable finding in these plots is a strong enhancement of the radio core flux at vera 22 and 43 ghz , which occurred coincidentally with the elevated vhe state . at 22 ghz
, we further detected a subsequent decay stage of the brightness at the last three epochs . also at 43 ghz , we detected possible saturation of the flux increase near the last epoch .
meanwhile , the evn monitoring confirmed a constant decrease of the hst-1 luminosity .
figure 3 describes vera 43-ghz images during the vhe active period , which indicate the flux enhancement within the central resolution element of 0.4 mas , corresponding to a linear scale of 0.03 pc or 56 @xmath25 .
we also note that the sma data at 230 ghz also appear to show a local maximum in its light curve during the period of the elevated vhe state .
another notable finding is a frequency - dependent evolution of the radio core flare .
the vera light curves clearly indicate that the radio core brightens more rapidly with a larger amplitude as frequency increases . at 43 ghz ,
the flux increased up to @xmath770% for the subsequent 2 months at an averaged rate of @xmath735%/month , and afterward the growth seems to be saturated . on the other hand ,
the core flux at 22-ghz progressively increased up to @xmath750% for the subsequent 4 months at a slower rate of @xmath712%/month . at 5 ghz ,
by contrast , the core remained virtually stable within the adopted error of 10% .
this is the first time that such a frequency - dependent nature of the radio flare is clearly confirmed in the m87 jet .
we also detected a core - shift between 22 and 43 ghz by using the vera dual - beam astrometry technique ( see hada et al .
2014 ) , where the amount of the shift was similar to the value obtained in the previous core - shift measurement ( hada et al .
2011 ) .
regarding the mev / gev regime , the lat light curves were stable up to february 2012 , and we did not find any significant flux enhancement during the period of the vhe activity . after march 2012 , however , no significant emission was detected for the subsequent 6 months in the 1- and 2-month binned data , suggesting a change in the he state after the vhe event .
this indicates a decrease in the he flux ( by a factor of @xmath26 ) after the vhe event , in agreement with the level of decrease observed at vhe in 2012 april - may ( beilicke et al .
following the 2008 episode this is the second time where a vhe event accompanied a remarkable radio flare from the core .
meanwhile , the radio luminosity of the hst-1 region was continuously decreasing , and we did not find any hints of the emergence of new components from hst-1 as seen in 2005 and 2010 .
these results strongly suggest that the vhe activity in 2012 is associated with the core at the jet base , while hst-1 is an unlikely source .
we note that these remarkable flares are very rare also in radio bands ( acciari et al .
2009 ) , so it is unlikely that an observed joint radio / vhe correlation is a chance coincidence , while the low statistics of the lat light curves still do not allow conclusive results on the he / vhe connection .
what kinds of mechanisms are responsible for the vhe production in the m87 core ? some of the existing models ascribe the vhe production to extremely compact regions near the central black hole ( e.g. , neronov & aharonian 2007 ; lenain et al .
2008 ; giannios et al . 2010 ; barkov et al .
these models well explain the rapid ( a few days ) variability observed in the previous vhe flares in 2005 , 2008 and 2010 .
however as far as we consider the case in 2012 , the size of the associated region expected from these models seems to be smaller than that suggested by vlbi and the observed longer timescale of the vhe variability .
indeed , a contemporaneous mm - vlbi observation at 230 ghz during the 2012 event also suggests the possible extended nature for the flaring region ( @xmath270.3 mas ; akiyama et al .
submitted ) .
another popular scenario for the m87 vhe production comes from a blazar - type , two - zone emission model where the vhe emission originates in the upstream part of a decelerating jet ( georganopoulos et al .
2005 ) or in the layer part of the spine - sheath structure ( tavecchio & ghisellini 2008 ) .
however in their steady state models , whether the models can explain the observed simultaneous radio / vhe correlation or not has not been well investigated yet because the emission regions associated with radio and vhe are spatially separated from each other . in this respect , a simple , homogeneous one - zone synchrotron self - compton jet model examined by abdo et al .
( 2009 ) would be interesting to note since one can in principle accept coincident radio / vhe correlations .
our multi - frequency radio monitoring additionally revealed a frequency - dependent evolution of the radio light curves for the m87 core .
such a behavior is often explained by the creation of a plasma condensation , which subsequently expands and propagates down the jet under the effect of synchrotron - self - absorption ( ssa ) .
the stronger ssa opacity at the jet base causes a delayed brightening at lower frequencies , and the light curve at each frequency reaches its maximum when the newborn component passes through the @xmath28 surface ( i.e. , the radio core at the corresponding frequency ) . in this context , by jointly using the observed time - lag ( @xmath29 ) and core - shift ( @xmath30 ) , we can estimate an apparent speed of the propagating component such that @xmath31 .
this results in a speed about @[email protected]@xmath32 , suggesting that the newborn component is sub - relativistic .
this is significantly smaller than the super - luminal features appeared from the core during the previous vhe event in 2008 ( 1.1@xmath32 ; acciari et al .
2009 ) , where the peak vhe flux is @xmath275 times higher than that in 2012 .
if we assume that propagating shocks or component motions seen in radio observations reflect the bulk velocity flow , this may suggest that the stronger vhe activity is associated with the production of the higher lorentz factor jet .
we are currently upgrading our m87 monitoring project by using the kvn and vera array ( kava ; niinuma et al .
2014 ) , which dramatically improves jet imaging capability thanks to the increase of the number of telescopes / baselines plus the addition of shorter baselines .
this will enable us to constrain the jet kinematics and radio / vhe connection more precisely .
the vera is operated by mizusawa vlbi observatory , a branch of national astronomical observatory of japan .
e - vlbi research infrastructure in europe is supported by the european union s seventh framework programme ( fp7/2007 - 2013 ) under grant agreement no .
ri-261525 nexpres .
the european vlbi network is a joint facility of european , chinese , south african and other radio astronomy institutes funded by their national research councils .
the submillimeter array is a joint project between the smithsonian astrophysical observatory and the academia sinica institute of astronomy and astrophysics and is funded by the smithsonian institution and the academia sinica .
the fermi lat collaboration acknowledges generous ongoing support from a number of agencies and institutes that have supported both the development and the operation of the lat as well as scientific data analysis .
these include the national aeronautics and space administration and the department of energy in the united states , the commissariata lenergie atomique and the centre national de la recherche scientifique institut national de physique nuclaire et de physique des particules in france , the agenzia spaziale italiana and the istituto nazionale di fisica nucleare in italy , the ministry of education , culture , sports , science and technology ( mext ) , high energy accelerator research organization ( kek ) and japan aerospace exploration agency ( jaxa ) in japan , and the k. a.wallenberg foundation , the swedish research council and the swedish national space board in sweden .
additional support for science analysis during the operations phase is gratefully acknowledged from the istituto nazionale di astrofisica in italy and the centre national detudes spatiales in france . | the nearby radio galaxy m87 offers a unique opportunity for exploring the connection between @xmath0-ray production and jet formation at an unprecedented linear resolution .
however , the origin and location of the @xmath0-rays in this source is still elusive .
based on previous radio / tev correlation events , the unresolved jet base ( radio core ) and the peculiar knot hst-1 at @xmath1120 pc from the nucleus are proposed as candidate site(s ) of @xmath0-ray production . here
we report our intensive , high - resolution radio monitoring observations of the m87 jet with the vlbi exploration of radio astrometry ( vera ) and the european vlbi network ( evn ) from february 2011 to october 2012 , together with contemporaneous high - energy @xmath0-ray light curves obtained by the fermi large area telescope . during this period ,
an elevated level of the m87 flux is reported at tev with veritas .
we detected a remarkable flux increase in the radio core with vera at 22/43 ghz coincident with the vhe activity .
meanwhile , hst-1 remained quiescent in terms of its flux density and structure in the radio band .
these results strongly suggest that the tev @xmath0-ray activity in 2012 originates in the jet base within 0.03 pc ( projected ) from the central supermassive black hole . |
recently , exotic nuclear systems involving a @xmath9 ( @xmath10 and @xmath11 ) as a constituent have been predicted based on phenomenologically constructed @xmath12 interactions @xcite .
the predicted bound states in @xmath13 , @xmath14 and @xmath10@xmath15be with large binding energies lie below the @xmath16 emission threshold , and thus are expected to have narrow decay widths . because of the strong @xmath12 attraction they have enormously high nucleon densities , @xmath17 @xmath18 , about 3 times the normal nuclear density @xmath19 @xmath18 .
such compact nuclear systems , which can be called
@xmath9 nuclear clusters " ( _ knc _ ) , are @xmath9 bound states in non - existing nuclei .
the basic ingredient for this new family of nuclear states is @xmath20 , which is identified as the known @xmath21 ( hereafter , expressed as @xmath22 ) with a binding energy of @xmath23 = 27 mev and a width of @xmath24 = 40 mev .
the lightest system is @xmath6 ( and its isospin partner @xmath25 ) , which was predicted to have @xmath23 = 48 mev and @xmath24 = 61 mev @xcite .
this species , which can be called _ nuclear kaonic hydrogen molecule _ , results from a fusion of @xmath22 and @xmath26 , namely , @xmath22
as a bound state of @xmath27 dissolves into a @xmath9 bound state , @xmath28 , @xmath29 the predicted rms distance of the @xmath30 system is @xmath31 the structure of @xmath32 and @xmath6 with calculated rms distances of @xmath10-@xmath26 and @xmath26-@xmath26 and rms radius of the @xmath10 distribution is shown in fig .
[ fig : kp - kpp ] . predicted structure of @xmath32 and @xmath6 .
, width=264 ] whereas some evidences for kaonic nuclear clusters for @xmath33 @xmath34 ( @xmath35 mev ) @xcite and @xmath6 ( @xmath23 = 115 mev ) @xcite were reported , it is important to produce various @xmath9 clusters by different nuclear reactions and thereby to examine their strucure and properties .
a method to determine the sizes of the @xmath9 clusters via the momentum correlation of decay particles has been proposed @xcite . the conventional methods to produce kaonic bound states are to use strangeness - transfer reactions of @xmath36 , @xmath37 , @xmath38 and @xmath39 .
the formation of @xmath9 clusters can be treated by a @xmath22 doorway model @xcite , in which a @xmath40 produced in elementary processes , typically , @xmath41 merges with a surrounding nucleon ( or nucleus ) to become a @xmath9 state .
the spectral shapes for the @xmath42 and @xmath43 , calculated by this @xmath22 doorway treatment as given in ref .
@xcite , are characterized by a large quasi - free component and a small bound - state peak . typically , the bound - state formation fraction is of the order of 1 % . in the present paper
we study the possibility to make use of the elementary process , @xmath44[eq : pp ] in connection with an experiment proposal at gsi using the fopi detector @xcite .
since the momentum transfer in this associated production of @xmath22 is very large ( @xmath45 1.6 gev/@xmath3 ) , one would expect that the formation cross section of @xmath9 clusters must be very small .
this process resembles the hypernuclear production process , @xmath46(p , k^+)^{a+1}_{\lambda}[z]$ ] , on a nuclear target , the cross section of which was evaluated by shinmura _
@xcite to be @xmath47 of the elementary @xmath48 production cross section , and this can be enhanced up to @xmath49 when a short - range correlation is taken into account . on the other hand , with a naive coalescence mechanism one
obtains a sticking probability of the order of 0.1 - 1.0 % because the internal momentum of the @xmath9 clusters is very large @xcite .
still , most of primarily produced @xmath22 are expected to escape , and the quasi - free process dominates .
we have studied this proton - induced associated production process more realistically , and found a surprisingly large production cross section by a unique mechanism , as described below .
we first treat conventional @xmath9 transfer reactions , typically , @xmath42 , based on the @xmath22 doorway model @xcite .
hereafter , we use revised theoretical values for the binding energy and width for @xmath6 ( @xmath50 mev and @xmath51 mev ) . in the elementary process , eq.([eq : pik ] ) , the produced @xmath22 interacts with a proton in the target @xmath52 , proceeding to @xmath6 , as shown in fig . [
fig : pik - pp - combined ] ( upper - left ) .
the momentum transfer at a typical incident momentum of @xmath53 gev/@xmath3 is @xmath45 600 mev/@xmath3 .
the energy spectrum involving both the bound and unbound regions was calculated following the morimatsu - yazaki procedure @xcite .
it is given by @xmath54[eq : pik - cross - section ] with a spectral function @xmath55 , \label{eq : spectral}\end{aligned}\ ] ] where @xmath56 is the energy transfer to the @xmath22-@xmath26 relative motion in doorway states , and @xmath57 the energy transfer to the @xmath10-@xmath58 relative ( internal ) motion in the @xmath6 system , and @xmath59 is a kinematical factor .
the function @xmath60 is @xmath61 with @xmath62 , @xmath63 and @xmath64 $ ] . in this derivation
we have used a zero - range approximation for @xmath65 and closure approximation to doorway states .
the calculated spectral function is shown in fig .
[ fig : pik - pp - combined ] ( lower - left ) .
the dominant part is the quasi - free component , in which the produced @xmath22 escapes , and only a small fraction constitutes a bound - state peak .
the figure shows that the bound - state peak intensity , though small , depends on the size of the @xmath7 system .
now we consider the following process with a projectile proton and a target proton , @xmath67 where @xmath28 denotes a @xmath9 cluster with structure of @xmath68 and decays subsequently as @xmath69 its diagram is shown in fig .
[ fig : pik - pp - combined ] ( upper - right ) .
when the incident proton interacts with a neutron in a deuteron target , an analogous process takes place , namely , @xmath70 with an iso - doublet partner @xmath71 ( = @xmath72 ) to be formed as well as @xmath73 .
hereafter , we take the @xmath74 case without loss of generality . spectral shapes with different size parameters.,width=283 ]
the @xmath75 process , where a @xmath76 pair is assumed to be created at zero range from a proton , is of highly off - energy shell ( @xmath77 ) .
this process is realized with a large momentum transfer to the second proton , which occurs efficiently by a short - range @xmath58 interaction , expressed by a yukawa type interaction exp@xmath78 with @xmath79 being the intermediate boson mass ( @xmath79 ) .
then , the effective interaction for the elementary process is written as @xmath80 where @xmath81 with @xmath82 and @xmath83 .
the @xmath22 is treated as a quasi - bound state of @xmath27 , and the interaction matrix element for @xmath22 formation is given by @xmath84 where @xmath85 the production cross section of @xmath86 ( @xmath87 ) is given by @xmath88\end{aligned}\ ] ] where @xmath89 with @xmath90 mev , @xmath91 mev , and @xmath92 fm .
then , the spectral function of @xmath86 ( @xmath87 ) is @xmath93 \nonumber\\ & & = -\frac{8 \mu_{\lambda^ * p}}{\hbar^2 } \sum_{l=0}^{\infty } ( 2 l + 1 ) { \rm i m } \big[\frac{1}{w(u_l^{(0 ) } u_l^{(+ ) } ) } \nonumber\\ & & \times \int_{0}^{\infty } { \rm d}r ' \int_{0}^{\infty } { \rm d}r\ , { \rm exp}(-\frac{r'}{\beta } ) \
, j_l ( q r ' ) \nonumber\\ & & \times~ u_l^{(0 ) } ( r _ < ) u_l^{(+ ) } ( r _ > ) \ , j_l ( qr ) ~{\exp}(-\frac{r}{\beta } ) \big],\end{aligned}\ ] ] where @xmath94 and @xmath95 are the stationary and outgoing solutions of the schrdinger equation , and @xmath96 is the wronskian of them . essentially , the spectral function is approximately composed of the following three factors : @xmath97 where @xmath98^{1/2}.\ ] ] they are : i ) the collision range @xmath99 , ii ) the momentum transfer @xmath100 and iii ) the structure function @xmath101 depending on the rms distance @xmath102 of the @xmath30 system .
the calculated wavefunction of @xmath28 yields @xmath103 1.05 fm .
the momentum transfer in the reaction is @xmath100 = 1.6 gev/@xmath3 .
the boson mass in producing @xmath22 in @xmath58 collision is taken to be the @xmath104 meson mass ; @xmath105 mev/@xmath106 .
the calculated spectral functions ( forward cross sections ) at @xmath107 = 4 gev are presented in the scale of @xmath108 in fig .
[ fig : pik - pp - combined ] ( lower - right ) .
the cross section integrated over the quasi - free region ( @xmath109 ) corresponds to the free emission of @xmath22 , which is known to have an empirical cross section of @xmath110 20 @xmath111b from a disto experiment at @xmath107 = 2.85 gev @xcite .
so , we have adjusted our absolute cross sections so as to give this empirical cross section .
surprisingly , in great contrast to the ordinary cases , the spectral function is peaked at the bound state with a very small quasi - free component .
this dominance of @xmath7 sticking in such a large-@xmath100 reaction can be understood as originating from the matching of the small impact parameter with the small size of the bound state . for further understanding of the mechanism we examined the dependence of the spectral function by changing the essential parameters fictitiously .
[ fig : s(e)-parameters ] ( upper ) shows that the bound - state peak decreases dramatically , when we increase the rms size @xmath102 from 1.05 fm ( the predicted size of the dense @xmath112 ) to 1.46 and 2.33 fm .
it also shows that with a denser system ( @xmath102 = 0.93 and 0.80 fm ) the peak height increases .
so , the dominant sticking of @xmath7 is the result of the dense @xmath9 system to be formed .
we also showed that with a fictitiously long - range @xmath66 collision ( @xmath79 = 10 mev ) the bound - state peak diminishes and the quasi - free component dominates . under this condition
the bound - state peak is enhanced only when the momentum transfer is small ( so called recoilless condition ) .
on the other hand , the dominant sticking of @xmath7 is assisted by the large momentum transfer ( @xmath113 1.6 gev/@xmath3 ) . as a more analytical way to show the physics behind we plot the radial dependence of each factor of the transition intensity in fig .
[ fig : t - amplitude ] .
the first row shows the @xmath114 interaction range of the yukawa type with different boson mass @xmath79 .
the second row shows the spherical bessel function @xmath115 corresponding to the momentum transfer of @xmath116 1.6 gev/@xmath3 .
these two functions have values mostly for @xmath117 0.3 fm .
the third row shows the structure dependent function @xmath101 , where the solid and broken curves represent the bound ( @xmath118 -62 mev ) and the quasi - free ( @xmath119 = 100 mev ) cases , respectively .
finally , the bottom row shows the radial dependences of the spectral strengths at the bound - state ( b ) and the quasi - free ( qf ) regions . in the realistic case ,
column ( a ) , the structure function is dense below @xmath102 = 1.05 fm so that it overlaps with the short range interaction assisted by the large momentum transfer .
the bound / qf ratio is 4.6 .
if we artificially increase the @xmath86 distance to 2.33 fm ( a deuteron like distance ) , as shown in column ( b ) the overlap drops down so that the ratio becomes 0.7 .
as far as the momentum transfer is large , a softened interaction ( @xmath120 140 mev ) , case ( c ) , does not change the bound / qf ratio from the realistic case ( a ) .
we have thus demonstrated that the dominant sticking of @xmath7 occurs as a joint effect of the short - range collision , the large momentum transfer and the compact size of the @xmath9 cluster .
it is vitally important to examine our results experimentally .
an experimental observation of @xmath28 in @xmath58 collision will not only confirm the existence of @xmath28 , but also proves the compactness of the @xmath9 cluster .
once a @xmath7 doorway is formed , the @xmath22 can not escape any more as a free particle and thus it is likely to further propagate in a complex nucleus as @xmath121 and ultimately a _ kaonic proton capture _ reaction may occur , such as @xmath122 in principle , missing mass spectra , @xmath123 and @xmath124 , may reveal monoenergetic peaks .
the elementary reaction of type @xmath126 was studied experimentally by the disto group at saturne @xcite and more recently by the anke group at cosy @xcite . the disto experiment identified @xmath48 from the invariant - mass spectrum of @xmath127 and constructed a missing mass spectrum of @xmath128 from those events involving @xmath48 at an incident proton energy of @xmath107 = 2.85 gev .
the missing mass in this case is equal to the mass of @xmath129 , @xmath130 and they found peaks associated with the production of @xmath131 , @xmath132 and @xmath133 .
they obtained a cross section of 20 @xmath111b for @xmath21 , which we used as an absolute scale in our calculations .
the anke experiment measured the energies and momenta of four emitted particles , @xmath134 , @xmath26 , @xmath135 and @xmath136 , with complete kinematical constraint at @xmath107 = 2.83 gev : @xmath137 they constructed @xmath138 and additionally @xmath139 , which was equated to @xmath140 .
when @xmath141 , it indicates that this @xmath129 decays to @xmath127 , and thus , it is assigned to be @xmath48 .
thus , the anke @xmath138 spectrum is equivalent to the disto spectrum .
so far , the reaction products were studied in terms of the elementary processes .
now , we propose to examine the new situation related to the existence of @xmath28 .
we point out that the formation / decay process of this object , as given in eq.([eq : pp2klp ] ) and ( [ eq : pp2kpp - decay-1],[eq : pp2kpp - decay-2],[eq : pp2kpp - decay-3 ] ) , are hidden in the observed spectra of @xmath138 .
the most important information is contained in a spectrum of @xmath123 , which is related to the mass of @xmath28 , @xmath142 but no such spectrum has been obtained .
presumably , the limited detector acceptance may not allow to reach the mass range of the present concern ( @xmath143 mev/@xmath106 ) .
now , a new experiment of the fopi group at gsi , which is aimed at measuring the whole products in the @xmath74 reaction at @xmath107 = 3.5 gev to reconstruct both the invariant mass @xmath144 and @xmath123 , is in progress .
the fopi group is also studying the @xmath145 process , which may lead to the formation of @xmath13 cluster .
in this note we have shown that the basic @xmath9 cluster , @xmath6 , as a dissolved state of @xmath146 can be formed with extraordinarily large enhancement , since the @xmath22 produced in a short - range collision with the participating proton automatically forms a @xmath7 doorway toward a dense @xmath6 system .
this anomalous dominance of @xmath7 sticking results from the unusual matching of the short collision range ( @xmath147 0.3 fm ) and the small radius of the produced @xmath28 ( @xmath148 1.0 fm ) , assisted by a large momentum transfer .
experimental confirmation of this effect will automatically proves the dense character of @xmath9 clusters . finally , we comment on our earlier proposal to make use of hot fireball in heavy - ion collisions as sources of various @xmath9 clusters @xcite . if the formation of @xmath7 doorway toward @xmath6 is really enhanced in the @xmath66 collisions , the @xmath9 cluster formation in heavy - ion reactions
is expected to be enhanced as well , since a heavy - ion reaction involves lots of primary @xmath66 collisions and subsequent processes .
we would like to thank prof .
s. shinmura for the illuminating discussion , and prof .
p. kienle , prof .
n. herrmann , dr .
k. suzuki and dr .
l. fabietti for the daily stimulating discussion during the course of the experimental collaboration .
we acknowledge the receipt of grant - in - aid for scientific research of monbu - kagakusho of japan .
99 y. akaishi and t. yamazaki , phys . rev .
* c 65 * ( 2002 ) 044005 . t. yamazaki and y. akaishi , phys .
* b 535 * ( 2002 ) 70 .
a. dot@xmath149 , h. horiuchi , y. akaishi and t. yamazaki , phys . lett . * b 590 * ( 2004 ) 51 .
a. dot@xmath149 , h. horiuchi , y. akaishi and t. yamazaki , phys . rev . * c 70 * ( 2004 ) 044313
. t. yamazaki , a. dot@xmath149 and y. akaishi , phys .
* b 587 * ( 2004 ) 167 . y. akaishi , a. dot@xmath149 and t. yamazaki , phys . lett . * b 613 * ( 2005 ) 140 .
t. suzuki _
et al . _ ,
b 597 * ( 2004 ) 263 .
m. agnello _
et al . _ ,
phys . rev .
* 94 * ( 2005 ) 212303 .
p. kienle , y. akaishi and t. yamazaki , phys . lett .
* b 632 * ( 2006 ) 187 .
t. yamazaki , p. kienle , k. suzuki , l. fabietti and n. herrmann , unpublished note ( 2004 ) and fopi proposal at gsi ( 2004 ) .
s. shinmura , y. akaishi and h. tanaka , prog .
* 76 * ( 1986 ) 157 .
k. suzuki _
et al . _ ,
int . conf . on exotic atoms and related topics ( 2005 , austrian academy of sciences press , isbn 3 - 7001 - 3616 - 1 ) p.83 .
o. morimatsu and k. yazaki , nucl . phys . * a 435 * ( 1985 ) 727 .
m. maggiora _
et al . _ , nucl . phys .
* a691 * ( 2001 ) 329c .
i. zychor _
et al . _ ,
* 96 * ( 2006 ) 012002 . | we have found theoretically that the elementary process , @xmath0 , which occurs in a short impact parameter ( @xmath1 0.2 fm ) and with a large momentum transfer ( @xmath2 gev/@xmath3 ) , leads to unusually large self - trapping of @xmath4 by the projectile proton , when a @xmath5 system exists as a dense bound state ( size @xmath1 1.0 fm ) propagating to @xmath6 .
the seed , called @xmath7 doorway " , is expected to play an important role in the ( @xmath8 type reactions and heavy - ion collisions to produce various @xmath9 nuclear clusters . |
the ability to detect and identify materials , hidden behind barriers , has potential applications within security and non - destructive testing @xcite .
the thz range of the electromagnetic spectrum is particularly attractive for these applications , because many barrier materials , such as clothing , plastic , and paper only attenuate thz waves moderately , while other materials , such as explosives and related compounds have characteristic spectroscopic fingerprints in the thz region @xcite .
a relevant geometry for these applications is the reflection geometry in which a pulsed or cw signal is sent towards an unknown structure and the amplitude and phase of the reflected signal is detected @xcite .
the task is then to deduce the structure from the measured reflection coefficient .
the situation is simplified in the effectively one - dimensional case , where the electromagnetic properties of the structure only vary in one direction .
retrieval of the structure parameters then becomes a one - dimensional inverse - scattering problem .
however , because the relevant materials are lossy in the thz range , they are also dispersive , according to the kramers kronig relations .
thus when applying inverse - scattering algorithms to the thz range one must take into account absorption and material dispersion .
there are several formulations and algorithms for the one - dimensional inverse scattering problem @xcite .
in particular , the layer peeling algorithms have turned out to be very efficient , and used in a wide range of applications @xcite .
these algorithms are based on the following , simple fact : consider the time - domain reflection impulse response of a layered structure . by causality ,
the leading edge is only dependent on the first layer , since the wave has not had time to feel the presence of the other layers .
thus one can identify the first layer of the structure from the impulse response .
this information can be used to remove the influence from the layer , which leads to synthetic reflection data with the first layer peeled off .
this procedure can be continued until the complete structure has been identified . in this work
we generalize the layer - peeling algorithm to dispersive stratified structures . provided
the material dispersion is sufficiently small , such that the time - domain fresnel reflections have durations less than the round - trip time in each layer , we can uniquely reconstruct the refractive indices of the structure ( section ii ) .
the method is illustrated by numerical examples in section iii . finally , in section iv we prove that for larger dispersion , the inverse scattering problem does not have a unique solution in general .
this is because one can not distinguish between the non - instantaneous temporal response of the medium itself ( due to dispersion ) , and the temporal response due to the stratification ( caused by reflections at the layer boundaries ) .
thus , extra information is needed , such as an upper bound for the dispersion combined with a lower bound for the layer thicknesses .
we first describe the model of the stratified medium . consider a layered , planar structure , consisting of @xmath0 layers with refractive indices @xmath1 and thicknesses @xmath2 , see fig .
[ fig : layers ] . here
the index @xmath3 labels the layer .
the light propagation in this structure is conveniently modeled using transfer matrices .
for simplicity we limit the analysis to normal incidence .
the transfer matrix of the transition from refractive index @xmath4 to @xmath5 is @xmath6 where @xmath7 is the fresnel reflection coefficient , associated with the index step between @xmath8 and @xmath9 .
the transfer matrix of the pure propagation in layer @xmath10 is @xmath11 & 0 \\ 0 & \exp[-i\omega n_j(\omega)d_j / c ] \end{bmatrix}},\ ] ] where @xmath12 is the vacuum light velocity .
note that eqs .
( [ eq : step])-([eq : prop ] ) are also valid when the refractive index is complex .
the refractive index is necessarily complex for dispersive materials , because dispersion is accompanied by loss according to the kramers kronig relations .
the transfer matrix of the total structure is given by @xmath13 the reflection coefficient @xmath14 from the left , and the transmission coefficient @xmath15 for the electric field from left to right , are given by the ( 2,1)- and ( 2,2)-elements of @xmath16 : @xmath17 layers .
the thicknesses and refractive indices of the layers are @xmath2 and @xmath1 , respectively.,width=377 ] we will now describe a layer - peeling method that can be used in the presence of weak dispersion .
the precise condition for the dispersion will become clear below . to be able to reconstruct the structure , we assume that the refractive index @xmath18 of the zeroth layer and the reflection spectrum of the entire structure ( as seen from @xmath19 ) , are known . the goal is to calculate @xmath1 for all @xmath20 and @xmath2 for @xmath21 .
the reflection spectrum of the layered structure can be expressed as follows : @xmath22 where @xmath23 is the impulse response , i.e. , the time - domain reflected field when a dirac delta pulse is incident to the structure .
note the lower limit @xmath24 in the integral : for a non - dispersive layer 0 , the round - trip time to the first index step would be @xmath25 ; however , due to dispersion we can only be sure that the round - trip time is no less than @xmath24 .
we take the forward and backward propagating frequency - domain fields at @xmath19 to be @xmath26 and @xmath14 , respectively , defining the field vector @xmath27 for now we assume that the layer thicknesses @xmath2 are known _ a priori _ ; the case with unknown layer thicknesses is treated below .
the field vector before the beginning of the first layer ( @xmath28 ) is given by @xmath29 we now define the reflection spectrum after the zeroth layer has been `` peeled off '' : @xmath30 the index step at @xmath31 can be regarded as a frequency - dependent reflector with ( unknown ) reflectivity @xmath32 , in accordance with eq . .
we assume that the dispersions of layers 0 and 1 are sufficiently small , such that the time - domain response associated with @xmath32 has duration less than @xmath33
. then we can write @xmath34 where @xmath35 in eq
. the lower limit in the integral reflects the fact that any reflections from the later index steps are delayed by ( at least ) the round - trip time @xmath33 . having established eq .
, we can now identify @xmath32 : @xmath36 with the local reflection coefficient @xmath32 in hand , we can calculate the refractive index of layer 1 using eq . .
once @xmath37 has been found , we can calculate the reflection spectrum with the first layer removed : @xmath38 now we find ourselves in the same situation as before , so we can continue the process until all layers have been found . from eq .
one obtains the _ complex _ reflection coefficient of each layer , and therefore , by eq . , both the real and imaginary parts of the refractive index .
one may ask if the reconstructed refractive index automatically is causal , or whether the kramers kronig relations could be used in addition to ensure causality .
the answer is that the lower limit in the integral ensures that @xmath32 is causal , and therefore by eq . that @xmath37 is analytic in the upper half - plane of complex frequency .
provided @xmath39 as @xmath40 , @xmath37 is therefore guaranteed to be causal @xcite . in practical situations
the available bandwidth is finite , the layer thicknesses may not be known _ a priori _ , and the reflection data contains noise .
we will now consider these aspects . in practice
, we only have reflection data in a limited bandwidth @xmath41 .
in other words , the reflection data @xmath42 in eq . is necessarily multiplied by a window function @xmath43 , which is nonzero only in the interval @xmath41 .
physically , this means that instead of probing the structure with a dirac delta pulse , we use an input pulse @xmath44 of duration @xmath45 : @xmath46 we require the duration of the time - domain response associated with @xmath47 to be less than @xmath33 , in order to distinguish between the response due to the first and the other layers .
the time - domain response associated with @xmath32 may already have duration up to @xmath48 , so we must require @xmath49 , or equivalently , @xmath50 .
this must be true for all layers , so @xmath51 where @xmath52 is a lower bound for the ( possibly unknown ) layer thicknesses .
in addition to condition , we recall that the time - domain response associated with @xmath32 must have duration less than @xmath53 , which means that the minimum allowable layer thickness is limited by the narrowest dispersion feature in the structure .
the response @xmath54 ( where @xmath55 denotes convolution ) is no longer guaranteed to vanish for @xmath56 , so we must extend the lower integration limit in eq . to contain the pulse @xmath44 : @xmath57 here
@xmath58 is the `` start position '' of @xmath44 , i.e. , a possibly negative number such that the response from the first layer roughly is contained in the interval @xmath59 $ ] .
eq . leads to the result @xmath47 rather than @xmath32 ; i.e. , the response of the single layer has been filtered with @xmath43 .
thus we must make sure that the bandwidth @xmath60 $ ] of the window function matches that of the dispersion of @xmath37 to be reconstructed .
we will now describe how the layer peeling algorithm also can be used when the layer thicknesses are not known _ a priori_. recall that the time - domain responses associated with @xmath61 have duration less than @xmath62 for all @xmath10 .
starting at a layer boundary , one can then perform layer peeling a distance @xmath63 using eq .
, and calculate the resulting time - domain response .
this removes the effect of the layer boundary , and the first signal in the transformed time - domain response is due to reflection from the next layer boundary .
let @xmath64 denote the start position of this signal and let @xmath58 denote the start position of @xmath44 .
if @xmath65 , this indicates that @xmath63 is less than the layer thickness . define a small thickness @xmath66 , which is sufficiently small to achieve the desired , spatial resolution .
then we can transform the fields successively using eq . with the small thickness @xmath66 until @xmath67 , and we have arrived at an index step .
we can then peel off the dispersive response associated with the index step and search for the next layer boundary , and so forth .
if @xmath43 only is nonzero in the interval @xmath41 , the time - domain responses do not have well - defined fronts , and the procedure above for finding the layer thickness is ambiguous .
however , one can use an alternative definition for the start position of the time domain signals , as shown in the numerical examples . for any real system
there is a given signal - to - noise ratio , which may be frequency dependent @xcite .
the layer peeling algorithm will fail if the reflection signal at a given index step becomes less than the noise .
this can be due to either low fresnel reflection from the index step itself , or high reflection or material absorption in the preceding part of the structure .
the case with high reflection in the preceding part of the structure was analyzed in ref .
@xcite , and it was shown that the noise amplification factor during the layer peeling algorithm was of the order of @xmath68 where @xmath69 is the minimum power transmission through the structure .
a similar conclusion can be reached by considering the effect of absorption .
let @xmath70 denote the minimum detectable reflection coefficient .
the maximum probing depth , @xmath71 , into a material with a single index step can be estimated by @xmath72\frac{\delta n}{2n}\approx \varrho,\ ] ] where @xmath73 is the change in the real part of the refractive index and @xmath74 is the average real part of the refractive index at the step . solving for the the maximum probing depth , we obtain @xmath75 we observe that the maximum probing depth into the structure is inversely proportional to the material absorption . assuming @xmath76 , @xmath77 , @xmath78 , and @xmath79 gives @xmath80 , which corresponds to @xmath81 wavelengths .
this is roughly the maximum depth one can expect the layer peeling algorithm to work for a material with comparatively low loss in the thz - range .
as a first numerical example of the algorithm in sec . [ layerpeelingdisp ]
, we consider a structure consisting of two material layers . we assume vacuum for @xmath82 , a material with refractive index @xmath83 for @xmath84 , and a material with refractive index @xmath85 for @xmath86 .
both materials are assumed to be dispersive and lossy .
the task is to determine @xmath83 , @xmath85 , and @xmath87 , given knowledge of the reflection coefficient at @xmath88 in a finite frequency range .
the refractive index of the first material is assumed to be in the form @xmath89 where @xmath90 is constant , and @xmath91 represents a lorentzian absorption feature , given by @xmath92 we take @xmath93 , @xmath94 , @xmath95 , and @xmath96 for the first material .
here @xmath97 , where @xmath98 is a scaling frequency .
note that @xmath90 must approach @xmath26 as @xmath99 for the refractive index in eq . to be causal .
however , we here use the approximation that @xmath90 is constant in the frequency range @xmath41 and assume that @xmath90 has the correct behavior as @xmath99 .
the refractive index of the second material is also assumed to be in the form eq .
( [ eq : n_mat ] ) , with @xmath93 , but here the susceptibility is taken to be the sum of ten lorentzian absorption features of various amplitudes , bandwidths , and center frequencies , in the vicinity of @xmath100 .
the resulting refractive index is seen in fig .
[ fig:2nd_layer ] . in this example
, we set @xmath101 thz , which leads to the vacuum wavelength @xmath102 mm .
we take @xmath103 , and assume that the reflection coefficient is known in the frequency range 08 thz .
the resulting power reflection coefficient is shown in fig .
[ fig : ref_w ] . in the layer peeling algorithm , we take @xmath104 . in addition , we must choose an appropriate window function .
the window function should have negligible energy outside the frequency window @xmath60=[-8~\text{thz } , 8~\text{thz}]$ ] .
additionally , the corresponding time - domain pulse should have a well - defined front . as a compromise between these two conflicting requirements
, we use a gaussian window function , defined in the time - domain by @xmath105.\ ] ] the width @xmath106 of the pulse and its central frequency @xmath107 are chosen to match its spectrum to the frequency range where the reflection coefficient is known .
we use @xmath108 ps and @xmath109 , which gives the window function @xmath43 in fig .
[ fig : ref_w ] .
this window function has a small energy outside @xmath60 $ ] .
we here set @xmath110 for @xmath111 in the numerical implementation of the algorithm , in accordance with the assumption that the reflection spectrum outside @xmath60 $ ] is unknown . the lower integration limit in eq .
( [ rho1wprinciple ] ) is set to @xmath112 ps . .
the incident pulse is given by eq .
, which is the time - domain representation of the window function .
the reflected pulse is given by the convolution of the incident pulse and @xmath113 .
the amplitude of the incident pulse has been scaled a factor 1/10 in the figure.,width=377 ] figure [ fig : pulses ] shows @xmath44 , which represents the incident pulse used to probe the structure . also shown is the convolution of the incident pulse and @xmath113 , which represents the reflected signal from the structure .
the reflected signal consists of two pulses , where the first pulse is due to reflection at @xmath19 and the second pulse is due to reflection at @xmath114 .
we observe that the duration of the first reflected pulse is similar to the duration of the incident pulse , which is due to the comparatively low dispersion of the first layer . because @xmath115 , there is no problem to correctly retrieve the refractive index of the first layer , as shown in fig .
[ fig:1st_layer ] .
.,width=377 ] having found the refractive index of the first layer , the next task is to find its thickness , which is done according to the procedure in sec .
[ layerpeelingdisp ] .
however , care must be taken to define the start position of the time domain signals , because they do not have a well - defined front , as discussed in sec .
[ sec : layer_thickness ] . in the numerical implementation of the algorithm
we define the start of the pulse as the first peak in the amplitude , given the amplitude is larger than a certain noise limit . as noted in sec .
[ sec : noise ] , if the amplitude of the second reflected pulse becomes too small , the layer peeling algorithm fails to find the thickness of the first layer .
one can also show that the layer thickness @xmath87 should be determined to within an accuracy of @xmath116 .
an incorrect layer thickness leads to oscillations in the retrieved refractive index @xcite . for the present parameters , the retrieved layer thickness is @xmath117 .
once @xmath87 is found we can " peel off the effect of the first layer using eq .
( [ eq : lp ] ) , which means that we transform the reflection coefficient to the position @xmath118 . .
the amplitude of the incident pulse has been scaled a factor 1/100 in the figure.,width=377 ] fig .
[ fig : pulses2 ] shows the incident pulse , @xmath44 , and the reflected pulse from the second layer .
the reflected pulse is calculated using the transformed reflection coefficient at the position @xmath118 .
we observe that the duration of the reflected pulse is significantly longer than that of the input pulse , which is due to the narrow dispersion feature of the second layer .
the refractive index of the second layer is determined in the same manner as for the first layer , and the retrieved refractive index is shown in fig .
[ fig:2nd_layer ] .
the error in the retrieved refractive index of the second layer is mainly due to the small inaccuracy in the retrieved @xmath87 . .
( a ) real part and ( b ) imaginary part of refractive index.,width=377 ] as noted in sec .
[ layerpeelingdisp ] , the duration of the time - domain response associated with reflection at an index step must be much less than @xmath62 for the layer peeling algorithm to work . if material dispersion is weak , the duration of the time - domain response is mainly given by the duration of the window function @xmath44 .
assuming @xmath108 ps , as in the first example , we must chose @xmath119 for successful retrieval of the refractive index in the first layer . as an example where @xmath63 is marginally too small , we set @xmath120 , with otherwise the same parameters as in the first example .
the retrieved refractive index of the first layer is shown in fig .
[ fig:1st_layer_example1 ] .
we observe that the retrieved refractive index deviates significantly from the exact refractive index . in addition
, the layer peeling algorithm fails to find the correct layer thickness in this case due to the residual time - domain response in @xmath113 caused by incomplete removal of the first layer . in the presence of noise
the incident pulse ( including the noise ) has been scaled a factor 1/100 in the figure.,width=377 ] as an example of the influence of noise on the layer peeling algorithm , we consider the same structure as in the first example , but with noise added to the input pulse @xmath44 .
the noise is assumed to be white and gaussian , with mean value 0 , standard deviation @xmath121 , and time @xmath122 between the samples .
the signal to noise ratio is thus proportional to @xmath43 in the frequency domain .
the retrieved refractive index of the first layer is shown in fig .
[ fig:1st_layer_example2 ] .
as expected , the influence of the noise is most severe at high frequencies , where the signal - to noise ratio is small .
we can estimate the maximum probing depth using eq .
( [ eq : noise ] ) .
because the loss is not constant over the frequency range where the reflection coefficient is known , we use the approximate values at @xmath123 thz , where @xmath124 , @xmath125 , and @xmath126 , giving @xmath127 for the maximum layer thickness of the first layer where we can expect the layer peeling algorithm to work .
figure @xmath128 shows the reflected pulse from the second layer , as determined from the layer peeling algorithm , and fig .
@xmath129 shows the retrieved refractive index of the second layer , in the presence of noise .
we observe that the retrieved refractive index is erroneous above 4 thz , which is due to the low signal to noise ratio at increasing frequency .
when the thickness of the first layer becomes much larger than 3@xmath130 , the reconstructed refractive index also becomes inaccurate at lower frequencies .
for the structure with three layers .
the amplitude of the incident pulse has been scaled a factor 1/5 in the figure.,width=377 ] as a final example , we will consider a structure with three layers , with vacuum for @xmath82 .
the refractive index of the first and third layer is the same as in the first example , and the refractive index of the second layer is 1 .
the thickness of the first and second layer is @xmath131 , with @xmath132 .
we take @xmath133 in the layer peeling algorithm and use the same window function as in the first example .
figure [ fig : ref_w_example_3 ] shows the power reflection coefficient of the structure .
the corresponding time - domain pulses reflected from the structure are shown in fig .
[ fig : pulses_example_3 ] . using the layer peeling algorithm
we retrieve the refractive indices and layer thicknesses of the structure .
the retrieved refractive index of layer 13 are shown in figs .
[ fig:1st_layer_example3]-[fig:3rd_layer_example3 ] , respectively .
we observe that there are errors in the retrieved refractive index at high frequencies .
the errors increase and approach shorter frequencies for the successive layers .
this is due to the unknown reflection coefficient above 8 thz .
because the window function has a small energy above this frequency , the unknown part of the reflection coefficient causes an unphysical precursor to each reflected pulse in the time - domain response .
the errors in retrieved refractive index are here due to the overlap between the main pulse from one layer and the precursor from the pulse reflected from the next layer . for the present parameters ,
this type of error in the retrieved refractive index is strongly reduced if the reflection coefficient is assumed known up to 12 thz instead of 8 thz .
we also find that an incorrect retrieved thickness of the first layer will lead to increasingly large errors in the retrieved refractive indices of the following layers . for the present parameters ,
an error of @xmath134 in @xmath87 leads to large errors in the retrieved @xmath85 and especially @xmath135 .
we will now prove that for dispersive structures , the inverse scattering problem does not have a unique solution in general . to this end
, we consider the fresnel reflection coefficient @xmath32 associated with a _
single _ index step , from @xmath18 to @xmath37 , and prove that this reflection coefficient can also be realized as a lossless and dispersionless structure ; a stack of discrete reflectors in vacuum , or equivalently , a layered structure .
let @xmath136 , @xmath137 , @xmath138 be the reflection coefficients of the discrete reflectors , and @xmath66 the distance between the reflectors .
we choose @xmath139 in accordance with the considered bandwidth , from @xmath140 to @xmath141 .
initially we set @xmath142 . by causality @xmath143 for a real time - domain response @xmath144 .
we now discretize the continuous response @xmath144 , leading to @xmath145 $ ] : @xmath146\exp(i\omega j2\delta / c ) \label{rho1principle3}\ ] ] in the bandwidth @xmath147 .
if @xmath148 were zero outside this bandwidth , the nyquist sampling theorem would immediately give the connection between @xmath145 $ ] and @xmath149 .
in general , however , the exact relation is found by extending @xmath148 to a periodic function with period @xmath150 , and setting @xmath145 $ ] equal to the associated fourier coefficients . with this procedure ,
would be identical in the relevant bandwidth , however with lower limit @xmath151 in the sum . setting the lower limit to 0 amounts to finding the optimal causal and discrete approximation to expression . in the limit where @xmath148 vanishes for @xmath152 , the error in the approximation tends to zero .
assuming that the discrete response @xmath145 $ ] is the reflection from a stack of discrete , lossless reflectors , we can perform layer peeling identifying the reflectors . by causality the first reflector is given by @xmath153 .
\label{firstrefl}\ ] ] note that the reflector @xmath136 is real , since the time - domain response @xmath144 , and therefore @xmath145 $ ] , is real : @xmath144 is a physical time - domain response as resulting from a real impulse . peeling off this reflector , and removing the subsequent pure propagation in the layer with thickness @xmath66 , can be done with the associated transfer matrices , or equivalently , by applying the schur recursion formula @xcite @xmath154 the layer peeling process can be continued until all reflectors have been found . in other words ,
two different structures give the same reflection response @xmath32 ; an index step between @xmath18 and @xmath37 , and several layers with non - dispersive , real refractive indices . to be able to solve the inverse scattering problem of a dispersive structure , it is therefore apparent that extra information ( in addition to the reflection spectrum ) must be known . in sec .
[ layerpeelingdisp ] we used the extra information that the dispersion is sufficiently small , such that the time - domain response associated with each fresnel reflection has duration less than the round - trip time to the next index step .
in addition we assumed @xmath155 for all @xmath10 .
an inverse scattering algorithm is applied to retrieve the material parameters of stratified structures . even though this problem does not have a unique solution in general , there exist cases where the algorithm can be applied , given additional information about the structure .
specifically , for a given , lower bound of the layer thicknesses , the dispersion must be sufficiently small , and the frequency range where the reflection coefficient is known must be sufficiently large .
the retrieval of material parameters of hidden layers is challenging due to absorption , noise , and unknown layer thicknesses . despite these challenges , there exist cases where the algorithm is successful , as illustrated by the numerical examples .
a. boutet de monvel and v. marchenko , `` new inverse spectral problem and its application , '' in `` inverse and algebraic quantum scattering theory ( lake balaton , 1996 ) , '' , vol .
488 of _ lecture notes in phys . _
( springer , berlin , 1997 ) , pp . 112 . | we consider the inverse scattering problem of retrieving the structural parameters of a stratified medium consisting of dispersive materials , given knowledge of the complex reflection coefficient in a finite frequency range .
it is shown that the inverse scattering problem does not have a unique solution in general .
when the dispersion is sufficiently small , such that the time - domain fresnel reflections have durations less than the round - trip time in the layers , the solution is unique and can be found by layer peeling .
numerical examples with dispersive and lossy media are given , demonstrating the usefulness of the method for e.g. thz technology . |
internet resources today are assigned by five regional internet registrars ( rirs ) .
these non - profit organisations are responsible for resources such as blocks of ip addresses or numbers for autonomous systems ( ases ) .
information about the status of such resources is maintained in publicly accessible rir databases , which are frequently used by upstream providers to verify ownership for customer networks . in general , networks are vulnerable to be hijacked by attackers due to the inherent lack of security mechanisms in the inter - domain routing protocol bgp .
real attacks have been observed in the past that led to the development of a variety of detection techniques and eventually of security extensions to bgp @xcite .
common to these attacks is a malicious claim of ip resources at the routing layer .
however , claims of network ownership can be also made at rir level , a fact that has received little attention so far . in a history of more than three decades ,
a vast number of internet resources have been handed out to numerous users under varying assignment policies .
some ases or prefixes have never been actively used in the inter - domain routing , others changed or lost their original purpose when companies merged or vanished .
it is not surprising that some internet resources became abandoned , i.e. resource holders ceased to use and maintain their resources . in this paper , we focus on threats that emerge from abandoned internet resources .
currently , there is no mechanism that provides resource ownership validation of registered stakeholders . instead
, the control over email addresses that are stored with rir database objects is often considered a proof of ownership for the corresponding resources .
our contribution is a generalized attacker model that takes into account these shortcomings .
we thoroughly evaluate the risk potential introduced by this attack by drawing on several data sources , and show that the threat is real .
since this kind of attack enables an attacker to fully hide his identity , it makes hijacking more attractive , and significantly harder to disclose .
consequently , we show that state - of - the - art detection techniques based on network measurements are ill - suited to deal with such attacks .
even so , these attacks have been evidenced in practice , and should thus be taken into account by future research .
we continue the discussion by establishing our attacker model in sec : model . in sec : resources , we estimate the risk potential of abandoned resources , and show that there is a real threat . as a result
, we outline an approach to mitigate this threat , and discuss limitations of related work in sec : research .
in particular , we outline the need for a system that provides resource ownership validation .
we conclude our discussion in sec : conclusion .
conventional attacks on bgp are based on its lack of origin validation , which allows an attacker to originate arbitrary prefixes or specific subnets from his own as .
we propose a new attacker model that accounts for attackers to take ownership of abandoned resources .
in such a scenario , an attacker is able to act on behalf of his victim , in particular to arrange upstream connectivity .
misled upstream providers unknowingly connect one or several ases including prefixes of the victims as instructed by an attacker who successfully hides his true identity . following this model ,
the anonymous attacker can participate in the cooperative internet exchange at arbitrary places without any formal incorrectness . in the following ,
we generalize a real incident to derive preconditions that enable this kind of attack . in previous work
@xcite , a corresponding attack has been observed in practice , which is known as the _ linktel incident_. the authors studied this attack and showed that a victim s prefixes originated from his own as , while the victim itself abandoned his business .
the authors reconstructed the attacker s course of action to claim ownership of the abandoned resources .
the linktel incident thereby revealed a major flaw in the internet eco - system : validation of resource ownership is most often based on manual inspection of rir databases . in this context , it was shown that the attacker was able to gain control over the victim s dns domain , and thus over corresponding email addresses .
the involved upstream provider presumably validated that the attacker s email address was referenced by the hijacked resources rir database objects .
given this proof of ownership , the upstream provider was convinced by the attacker s claim to be the legitimate holder of the resources .
surprisingly , the attacker captured the victim s dns domain by simply re - registering it after expiration . for several months ,
the attacker s abuse of the hijacked resources remained unnoticed . by combining several data sources
, the authors showed that the hijacked networks were utilized to send spam , to host web sites that advertised disputable products , and to engage in irc communication .
after the victim recovered his business , he learned that his networks were listed on spamming blacklists .
however , the attacker s upstream provider refused to take action at first , since the victim was unable to refute the attacker s ownership claims .
based on the insights gained from the linktel incident , we show that the attacker s approach can be generalized . to enable hijacking of internet resources , the following preconditions have to be met : ( a ) internet resources are evidentially abandoned and ( b ) the original resource holder can be impersonated .
if an organisation goes out of business in an unsorted manner , these conditions are eventually met . as a first consequence ,
the organisation ceases to use and maintain its resources .
if this situation lasts over a longer period of time , the organisation s domain name(s ) expire .
since day - to - day business lies idle , re - registration and thus impersonation becomes practicable for an attacker . at that moment
, upstream connectivity can be arranged on behalf of the victim , since face - to - face communication is not required in general .
routers can be sent via postal service , or even be rented on a virtualized basis .
details on bgp and network configuration are usually exchanged via email , irc , or cellular phone , and payment can be arranged anonymously by bank deposits or other suitable payment instruments . without revealing any evidence about his real identity ,
the attacker is able to stealthily hijack and deploy the abandoned resources .
the implications of this attacker model are manifold .
first , an attacker may act on behalf of a victim , thereby effectively hiding his own identity and impeding disclosure .
this makes hijacking more attractive as it enables riskless network abuse .
it hinders criminal prosecution , and could be used to deliberately create tensions between organisations or even countries . due to the lack of a system for resource ownership validation
, these attacks only depend on idle organisations or missing care by legal successors of terminated businesses . even after the discovery of such an attack ,
it is difficult for the victim to mitigate since reclaiming ownership is the word of one person against another at first .
the linktel incident @xcite proves that this is not only a realistic scenario : such attacks are actually carried out in practice . the benefit of attacks based on abandoned resources can even be higher than in the case of conventional attacks . hijacking productive networks
rarely lasts for more than a few hours , since the victim can receive great support in mitigating the attack .
moreover , for most cases , the benefit is reduced to blackholing a victim s network with the youtube - pakistan incident being a prominent example . in addition , monitoring systems for network operators exist that raise alarms for unexpected announcements of their prefixes . however , due to the very nature of abandoned resources , virtually no one is going to take notice of an attack .
our attacker model thus accounts for stealthily operating attackers who aim at persistently maintaining malicious services .
we identify readily hijackable internet resources by searching rir databases for unmaintained resource objects .
subsequently , we distinguish between resources that are still in use , with potential for network disruption , and resources that are fully abandoned and ready to be abused stealthily .
such resources are especially attractive for attackers for two reasons .
first , the resource is assigned to an organisation for operational use and thus represents a valid resource in the internet routing system .
second , an attacker can easily claim ownership by taking control of the contact address referenced by corresponding rir database objects , i . by re - registering a domain name .
consequently , we look for rir database objects that reference email addresses with _ expired dns names_. since the inference of invalid domain names can also be the result of poorly maintained resource objects or typing errors , it is important to take into account recent database activities for individual resource owners , and to correlate this information with bgp activity .
the following analysis is based on archived ripe database snapshots over 2.5 years ( 23 february , 2012 till 9 july , 2014 ) .
our results are representative for the european service region only , but similar analyses can be done with little effort for other service regions , too .
ripe , like all other rirs , provides publicly available database snapshots on a daily basis .
most of the personally related information is removed due to privacy concerns .
some attributes , however , remain unanonymized , which we utilize to extract dns names .
the ripe database holds more than 5.2 million objects .
these objects can be updated from the web or via email .
most of these objects optionally hold an email address in the ` notify ` field , to which corresponding update notifications are sent . despite anonymization , we found that these ` notify ` fields are preserved in the publicly available database snapshots , which is also the case for ` abuse - mailbox ` attributes . to extract dns names , we parse these email addresses where applicable . table : ripe_obj shows the distribution of stored objects by type along with the number of dns names we were able to extract .
although we found more than 1.5 million references to dns names , the total number of _ distinct _ names is only 21,061 .
this implies that , on average , more than 72 objects reference the same dns name .
the overall fraction of objects that reference a domain name is 29.24% , which is surprisingly high since the database snapshots are considered to be anonymized . 0.75xrrr * object type * & * frequency * & + ` inetnum ` & 3,876,883 & 1,350,537 & ( 34.84% ) + ` domain ` & 658,689 & 97,557 & ( 14.81% ) + ` route ` & 237,370 & 50,300 & ( 21.19% ) + ` inet6num ` & 231,355 & 8,717 & ( 3.77% ) + ` organisation ` & 82,512 & 0 & ( 0.00% ) + ` mntner ` & 48,802 & 0 & ( 0.00% ) + ` aut - num ` & 27,683 & 6,838 & ( 24.70% ) + ` role ` & 20,684 & 14,430 & ( 69.76% ) + ` as - set ` & 13,655 & 2,500 & ( 18.31% ) + ` route6 ` & 9,660 & 723 & ( 7.48% ) + ` irt ` & 321 & 162 & ( 50.47% ) + * total * & * 5,239,201 * & * 1,531,764 * & * ( 29.24% ) * + hijackable internet resources are given by ` inetnum ` and ` aut - num ` objects , which represent blocks of ip addresses and unique numbers for autonomous systems respectively .
exemplary database objects are provided in fig : ripe_entry , further details on the ripe database model and update procedures are available at @xcite .
it is worth noting that the attacker neither needs authenticated access to the database nor does the attacker need to change the database objects .
the attacker only needs to derive a valid contact point .
we assume that the ( publicly available ) notification address usually belongs to the same dns domain as the technical contact point .
detailed analysis is subject to future work ; in our study , we disregard groups of objects that reference more than a single dns domain as a precaution .
the ripe database is mostly maintained by resource holders themselves .
its security model is based on references to ` mntner ` ( maintainer ) objects , which grant update and delete privileges to the person holding a ` mntner ` object s password .
this security model allows us to infer objects under control of the same authority by grouping objects with references to a common ` mntner ` object .
we use these _ maintainer groups _ to estimate the impact of an attack for individual authorities : on average , we observed nearly 110 such references per ` mntner ` object , with a maximum of up to 436,558 references .
the distribution of the number of objects per maintainer group is presented in fig : mntner_group_sizes . ....
inetnum : 194.28.196.0 - 194.28.199.255 netname : ua - veles descr : llc " unlimited telecom " descr : kyiv notify : [email protected] mnt - by : veles - mnt .... .... aut - num : as51016 as - name : vales descr : llc " unlimited telecom " notify : [email protected] mnt - by : veles - mnt .... for each of the maintainer groups , we obtain the set of all dns names referenced by a group s objects .
to unambiguously identify maintainer groups with expired domains , we merge disjoint groups that reference the same dns domain , and discard groups with references to more than one dns name . from an initial amount of 48,802 maintainer groups , we discard ( a ) 937 groups of zero size , i.e. unreferenced ` mntner ` objects , ( b ) 31,586 groups without domain name references , and ( c ) 4,990 groups with multiple references .
the remaining 11,289 groups can be merged to 8,441 groups by identical dns names .
we further discard groups that do not include any hijackable resources , i.e. ` inetnum ` and ` aut - num ` objects , which finally leads us to 7,907 object groups .
note that the number of these groups is a lower bound : an attacker could identify even more with access to unanonymized ripe data .
as discussed above , each of these groups is maintained by a single entity .
if a group s dns name expires , we consider the entity s resources to be a valuable target for an attacker . to confirm that a set of resources is abandoned
, our approach is based on complementary data sources .
we start with domain names that expire , which is a strong yet inconclusive indication for a fading resource holder .
we gain further evidence by considering only resources that are neither changed in the ripe database nor announced in bgp . including both administrative ( dns , ripe ) and an operational ( bgp ) measures
gives a comprehensive picture on the utilization of the resources .
we used the _ whois system _ to query expiry dates for all extracted dns names ( cf . , section [ sec : grouping ] ) .
fig : domain_expiry_dist shows the distribution of these dates . at the time of writing , 214 domain names have been expired .
another 121 names expire within the week , given that the owners miss to renew their contracts .
the most frequent top level domains are ` .com ` ( 27.9% ) , ` .ru ` ( 21.5% ) , and ` .net ` ( 13.0% ) , while the most frequent _ expired _ tlds are ` .ru ` ( 20.1% ) , ` .it ` ( 16.4% ) , and ` .com ` ( 9.81% ) .
the longest valid domains are registered until 2108 and mostly represent governmental institutions .
the longest expired domain has been unregistered for nearly 14 years .
with respect to the maintainer groups derived above , a total of 65 groups that reference expired dns names remain .
these groups hold 773 ` /24 `
networks and 54 ases , and are subject to closer investigation . for each of the 7,907 maintainer groups
divided into 7,842 valid groups and 65 with expired dns names we extracted the minimum time since the last change for any of its database objects . note that we filtered out automated bulk updates that affected _ all _ objects of a certain type .
fig : lastchange_ripedb shows the distribution of database updates for groups with valid and for groups with expired domain names . while about 10% of the valid groups show changes within two months , dns - expired groups differ strikingly : the 10%-quantile is at nearly 5 months . hence , given these long times without updates , we consider resource groups that exhibit an object update within 6 months to be still maintained and not abandoned .
note that we do not assume inactivity in absence of such changes . to confirm inactivity , we correlate the ripe database updates with activities in the global routing system . for that , we analyze all available bgp update messages from the routeviews oregon s feed for the same time frame .
this data set comprises 83,255 files with 18.4 billion announcements and 1.04 billion withdraw messages for resources assigned by ripe . given this data , we are able to extract two indicators : ( 1 ) the time since an _ ip prefix _ was last visible from the routeviews monitor , and ( 2 ) the time since the last deployment of a ripe - registered _ as number _ by looking at as path attributes .
fig : lastchange_bgp shows the distribution of last activity in bgp for any internet resource in our maintainer groups .
nearly 90% of resources in valid groups are visible in bgp at the moment .
surprisingly , most of the remaining groups did not show any activity at all during the last 2.5 years .
about 75% of the dns - expired resources are present in today s routing table and are thus still actively used .
the remaining resources did show some activity in the past ( 10% ) or were never observed in bgp during our analysis period ( 15% ) .
these findings confirm our assumption that inactivity in the ripe database does not necessarily imply operational shutdown . while up to 85% of the expired resources were seen in bgp within the last 2.5 years , fig
: lastchange_ripedb indicates that not more than 55% of the expired resources received an update in the ripe database .
we further learn that some expired resources did show bgp activity in the past , and do not show any activity today .
note that we disregard resources with recent bgp activity .
these resources could potentially be hijacked already ; however , attacks that started before our analysis are beyond the scope of our approach .
so far , we learned that 65 maintainer groups with a total of 773 ` /24 ` networks and 54 ases reference expired dns names .
our activity measures further indicate that valid groups yield higher activity than expired groups . by combining these measures , we are able to infer resources that are inactive from both an administrative and an operational point of view .
fig : lastchange_all shows the time since the latest change by any of these measures , i.e. , the minimum value of both measures .
this combined activity measure clearly splits the 65 expired maintainer groups into two disjoint sets : 52 cases were active within the last 3 months , while 13 cases did not show any activity for more than 6 months .
we consider these remaining 13 cases to be effectively abandoned .
these resource groups represent a total number of 15 ` inetnum ` objects ( with an equivalent of 73 ` /24 ` networks ) and 7 ` aut - num ` ( i.e. , as number ) objects .
now that we have identified vulnerable resources , we feel obliged to protect these resources .
since any attacker could repeat our analysis , we are going to contact endangered resource holders before publishing our findings .
although communication via e - mail is futile due to expired domains , we can fall back on telephone numbers provided in the ripe database to reach out for the operators .
for the problem of abandoned internet resources , one might argue that the threat is not caused by a technical but a social problem because operators agree to their peering relations based on a weak authentication scheme .
this scheme can be replaced by stronger verification
the required data already exists .
rirs have contracts with the owners of delegated resources and thus are aware of more reliable contact points ( e.g. , telephone numbers ) .
however , the current situation shows that we need mechanisms , tools , and procedures which are not tedious for operators but allow for easy resource verification . our approach to identify abandoned resources can be easily extended to continuously monitor resources of all rirs .
this would allow us to warn network operators about potential risks .
finding scalable approaches to implement early warning and prevention in real - time , though , is an open research issue .
current research is particularly focused on the detection of bgp hijacking attacks .
proposed mitigation techniques look on the control plane , the data plane , or both .
_ control plane _
monitoring is used to identify anomalies in bgp routing tables to infer attacks @xcite .
such approaches are prone to false positives due to legitimate causes for anomalies .
techniques based on _ data plane _ measurements account for changes of the router topology @xcite , or of hosts in supposedly hijacked networks @xcite .
these approaches rely on measurements carried out before and during an attack . beyond that
, studies on the malicious intent behind hijacking attacks exist @xcite .
all detection approaches require the observation of suspicious routing changes .
attacks based on our attacker model take place outside the routing system , and thus do not lead to noticeable routing changes apart from a supposedly legitimized organisation starting to reuse its internet resources . hence , current detection systems are incapable to deal with this kind of attack . the dns has been widely studied in the context of malicious network activities , mainly concerning spammers or fraud websites .
proactive blacklisting of domain names @xcite does not help in our scenario as the threat is effective on the routing layer .
identifying orphaned dns servers @xcite is also out of scope of this paper as the attacker does not leverage the dns server but the expiring domain . despite its effectiveness
, we consider our approach to detect and monitor abandoned resources as outlined above an intermediate solution only .
in fact , we argue that there is a need for resource ownership validation .
there is ongoing effort to increase the deployment of a _ resource public key infrastructure ( rpki ) _ @xcite . in its present state
, the rpki allows for validation of route origins by using cryptographically secured bindings between as numbers and ip prefixes .
this mechanism prevents common hijacking attacks . in terms of hijacking abandoned resources , however , this system is ineffective in its current form since the abandoned origin as is taken over as well , and origin validation performed by bgp routers @xcite will indicate a valid bgp update . even though the rpki itself can be misused @xcite , at the moment it represents the only mechanism for proofing securely ownership of internet resources .
we merely lack a clear procedure in the context of abandoned internet resources .
one approach could be the following operational rule : a peering request is only established when resource objects of the requesting peer exist in the rpki .
recent time stamps for these objects indicate that the requesting peer has control over the resources as only authorized users can create such objects .
such a scheme seems feasible from the operational perspective and might even increase the incentives to deploy rpki .
rpki is part of _ bgpsec _ , an even larger effort to secure bgp .
this extension to the protocol remedies the risk of hijacking abandoned resources due to its path validation capabilities : in our model , an attacker can not provide valid cryptographic keys to sign update messages as specified by bgpsec @xcite .
however , the development of bgpsec is at an early stage , and the benefit compared to pure origin validation is questionable in particular in sparse deployment scenarios @xcite
. future research should be carried out on enabling internet service providers to validate resource ownership of customers .
we see the potential of such a system not only in preventing attackers from hijacking abandoned internet resources .
it would also establish trust in customer - provider and peer - to - peer relationships , as well as in resource transfers issued by rirs or lirs .
motivated by a real - world case study , we introduced a generalized attacker model that is aimed on the hijacking of abandoned internet resources .
we showed that such an attack is feasible with little effort , and effectively hides the attacker s identity by acting on behalf of a victim . by studying orthogonal data sources over a period of more than 30 months
, we could give evidence of a high risk potential of such attacks . only in the european rir database
, we found 214 expired domain names that control a total of 773 ` /24 ` networks and 54 ases , all of which can be easily hijacked .
about 90% of these resources are still in use , which enables an attacker to disrupt operational networks .
the remaining 10% of the resources are fully abandoned , and ready to be stealthily abused .
our findings led us to the conclusion that state - of - the - art systems are limited to deal with this kind of attack .
more importantly , we argued that there is a need for _ resource origin validation_. such a framework would not only prevent attacks , but could also strengthen today s internet eco - system by establishing trust in resource ownership . in this paper , we sketched a new attack vector .
up until now , it is unclear how common such attacks are ; our findings thus might trigger new malicious activities .
however , we also showed that this attack is already known to attackers , and we sketched countermeasures to mitigate this concern .
in addition , we contact the holders of vulnerable resources before publication of our findings .
this work has been supported by the german federal ministry of education and research ( bmbf ) under support code 01by1203c , project _ peeroskop _ , and by the european commission under the fp7 project _
eins _ , grant number 288021 . | the vulnerability of the internet has been demonstrated by prominent ip prefix hijacking events .
major outages such as the china telecom incident in 2010 stimulate speculations about malicious intentions behind such anomalies .
surprisingly , almost all discussions in the current literature assume that hijacking incidents are enabled by the lack of security mechanisms in the inter - domain routing protocol bgp . in this paper
, we discuss an attacker model that accounts for the hijacking of network ownership information stored in regional internet registry ( rir ) databases .
we show that such threats emerge from abandoned internet resources ( e.g. , ip address blocks , as numbers ) .
when dns names expire , attackers gain the opportunity to take resource ownership by re - registering domain names that are referenced by corresponding rir database objects .
we argue that this kind of attack is more attractive than conventional hijacking , since the attacker can act in full anonymity on behalf of a victim . despite corresponding incidents
have been observed in the past , current detection techniques are not qualified to deal with these attacks .
we show that they are feasible with very little effort , and analyze the risk potential of abandoned internet resources for the european service region : our findings reveal that currently 73 ` /24 ` ip prefixes and 7 ases are vulnerable to be stealthily abused .
we discuss countermeasures and outline research directions towards preventive solutions . |
one of the recent strategies to deal with interference is interference alignment . in a multiuser channel
, the interference alignment method puts aside a fraction of the available dimension at each receiver for the interference and then adjusts the signaling scheme such that all interfering signals are squeezed in the interference subspace .
the remaining dimensions are dedicated for communicating the desired signal , keeping it free from interference .
cadambe and jafar @xcite , proposed the linear interference alignment ( lia ) scheme for ic and proved that this method is capable of reaching optimal degrees of freedom of this network .
the optimal degrees of freedom for a @xmath0 user ic is obtained in the same paper to be @xmath1 .
the proposed scheme in @xcite is applied over many parallel channels and achieves the optimal degrees of freedom as the signal - to - noise ratio ( snr ) goes to infinity .
@xcite , proposed the so called ergodic interference alignment scheme to achieve 1/2 interference - free ergodic capacity of ic at any signal - to - noise ratio .
this scheme is based on a particular pairing of the channel matrices .
the scheme needs roughly the same order of channel extension compared with @xcite , to achieve optimum performance .
the similar idea of opportunistically pairing two channel instances to cancel interference has been proposed independently by @xcite as well .
however , ergodic interference alignment scheme considers only an special pairing of the channel matrices and does not discuss the general structure of the paired channels suitable for interference cancelation .
this paper addresses the general relationship between the paired channel matrices suitable for canceling interference , assuming linear combining of paired channel output signals .
using this general pairing scheme , to align interference at receiver , proposed scheme significantly lowers the required delay for interference to be canceled . from a different standpoint , this paper obtains the necessary and sufficient feasibility conditions on channel structure to achieve total dof of the ic using limited number of channel extension .
so far , interference alignment feasibility literature have mainly focused on network configuration , see @xcite and references therein . to ease some of interference alignment criteria by using channel structure ,
@xcite investigates degrees of freedom for the partially connected ics where some arbitrary interfering links are assumed disconnected . in this channel model , @xcite examines how these disconnected links are considered on designing the beamforming vectors for interference alignment and closed - form solutions are obtained for some specific configurations .
in contrast , our work evaluates the necessary and sufficient conditions on channel structure of an ic to make perfect interference alignment possible with limited number of channel extensions .
consider the @xmath0 user ic consisting of @xmath0 transmitters and @xmath0 receivers each equipped with a single antenna , as shown in fig .
[ figure : kuser ] .
each transmitter wishes to communicate with its corresponding receiver .
all transmitters share a common bandwidth and want to achieve the maximum possible sum rate along with a reliable communication .
channel output at the @xmath2 receiver and over the @xmath3 time slot is characterized by the following input - output relationship : @xmath4}(t)={\bf h}^{[k1]}(t){\bf x}_p^{[1]}(t)+{\bf h}^{[k2]}(t){\bf x}_p^{[2]}(t ) \cdots \nonumber \\+{\bf h}^{[kk]}(t){\bf x}_p^{[k]}(t)+{\bf z}^{[k]}(t ) , \end{aligned}\ ] ] where @xmath5 is the user index , @xmath6 is the time slot index , @xmath7}(t)$ ] is the output signal vector of the @xmath2 receiver , @xmath8 } ( t)$ ] is the transmitted precoded signal vector of the @xmath2 transmitter which will shortly be defined , @xmath9 } ( t)$ ] , @xmath10 is the fading factor of the channel from the @xmath11 transmitter to the @xmath2 receiver over @xmath3 time slot , and @xmath12}(t)$ ] is the additive white gaussian noise at the @xmath2 receiver .
the noise terms are all assumed to be drawn from a gaussian independent and identically distributed random process with zero mean and unit variance .
it is assumed that all transmitters are subject to a power constraint @xmath13 .
the channel gains are bounded between a positive minimum value and a finite maximum value to avoid degenerate channel conditions ( e.g. the case of all channel coefficients being equal or a channel coefficient being zero or infinite ) .
assume that the channel knowledge is causal and is available globally , i.e. over the time slot @xmath14 , every node knows all channel coefficients @xmath9 } ( \tau ) , \forall j , k \in \{1 , \ldots , k\ } , \tau \in \{1 , \ldots , t\}$ ] .
hereafter , time index is omitted for the sake of simplicity .
degrees - of - freedom region for a @xmath0 user ic , with the system model discussed in section [ sysmod ] , has been derived in @xcite as follows , @xmath15 and the number of dof achieved by @xmath0 user ic is obtained to be @xmath1 .
following corollay describes the only dof vector , @xmath16 , that achieves total number of dof .
[ cor1 ] the only dof vector that achieves total number of dof of an ic is @xmath17 dof point mentioned in ( [ optdof ] ) is on the vertex of the dof region . since
other vertices @xmath18 has dof less than @xmath1 . because dof region is convex , it is straightforward to see that @xmath19 is the only point that produces the largest total dof . consider a three user ic .
assuming channel coefficients to be generic , i.e. the channel coefficients are time varying and are drawn from continuous independent distributions , @xcite has shown that optimal degrees of freedom for a three user ic can not be achieved over limited number of channel usage . to maintain continuity of presentation ,
a short review is presented here .
consider using @xmath20 time slots of the channel , according to [ cor1 ] , achieving optimal degrees of freedom implies that @xmath21 degrees of freedom should be achieved for each of the transmitters .
the signal vector at the @xmath22th receiver can be stated as @xmath4}&{}={}&{\bf h}^{[k1]}{\bf x}_p^{[1]}+{\bf h}^{[k2]}{\bf x}_p^{[2]}\nonumber \\ & & { + } \ : { \bf h}^{[k3]}{\bf x}_p^{[3]}+{\bf z}^{[k]},\end{aligned}\ ] ] where @xmath23}$ ] is a @xmath24 column vector which is obtained by coding the transmitted symbols over @xmath20 time slots of the channel , as will be explained below .
@xmath7}$ ] and @xmath12}$ ] represent the @xmath20 symbol extension of @xmath25}$ ] and @xmath26}$ ] , respectively .
@xmath9}$ ] is a diagonal @xmath27 matrix which represents the @xmath20 symbol extension of the channel as shown in ( [ topeq ] ) at the top of the next page .
@xmath28}\equiv \left [ \begin{array}{c c c c } h^{[kj]}(2n(t-1)+1 ) & 0 & \cdots & 0\\ 0 & h^{[kj]}(2n(t-1)+2 ) & \cdots & 0 \\ \vdots & \cdots & \ddots & \vdots \\ 0 & 0 & \cdots & h^{[kj]}(2nt ) \end { array}\right ] \label{topeq}\ ] ] in the extended channel , message @xmath29 at transmitter @xmath30 is encoded to @xmath21 independent streams @xmath31 } , m=1 , \ldots , n $ ] and sent along the vector @xmath32 } $ ] so that @xmath33}$ ] can be written as @xmath34 } = { \bf v}^{[1 ] } { \bf x}^{[1 ] } , \end{aligned}\ ] ] where @xmath35}$ ] is a @xmath36 column vector comprised of transmitted symbols @xmath31 } , m=1 , \ldots , n $ ] , @xmath37 } $ ] is a @xmath38 dimensional precoding matrix comprised of the vectors @xmath32 } , m=1 , \ldots , n $ ] as its columns . in a similar way , @xmath39 and @xmath40 are encoded by transmitters @xmath41 and @xmath42 , respectively and sent to the channel as : @xmath43 } = { \bf v}^{[2 ] } { \bf x}^{[2 ] } , \end{aligned}\ ] ] @xmath44 } = { \bf v}^{[3 ] } { \bf x}^{[3 ] } .\end{aligned}\ ] ] the received signal at the @xmath45th receiver can be evaluated to be @xmath4}&{}={}&{\bf h}^{[k1]}{\bf v}^{[1 ] } { \bf x}^{[1 ] } + { \bf h}^{[k2]}{\bf v}^{[2 ] } { \bf x}^{[2 ] } \nonumber \\ & & { + } \:{\bf h}^{[k3]}{\bf v}^{[3 ] } { \bf x}^{[3 ] } + { \bf z}^{[k]}.\end{aligned}\ ] ] receiver @xmath46 cancels the interference by zero forcing all @xmath47 } , j\not = i$ ] to decode @xmath48 . at receiver @xmath30 ,
@xmath21 desired streams are decoded after zero forcing the interference from transmitters @xmath49 and @xmath42 . to achieve @xmath50 dimensions free of interference from the @xmath20 dimensional
received signal vector @xmath51 } $ ] , the dimension of the interference signal should not be more than @xmath21 .
this can be realized by perfectly aligning the received interference from transmitters @xmath41 and @xmath42 at the receiver @xmath30 , i.e. @xmath52 } { \bf v}^{[2 ] } \right ) = \textrm{span } \left ( { \bf h}^{[13 ] } { \bf v}^{[3 ] } \right ) , \label{se1 } \end{aligned}\ ] ] where @xmath53 denotes the column space of matrix @xmath54 .
at the same time , receiver @xmath41 zero forces the interference from @xmath35}$ ] and @xmath55}$ ] . to achieve @xmath21 dimensions free of interference
, we will have : @xmath56 } { \bf v}^{[1 ] } \right ) = \textrm{span } \left ( { \bf h}^{[23 ] } { \bf v}^{[3 ] } \right ) .
\label{se2 } \end{aligned}\ ] ] in a similar way , @xmath37}$ ] and @xmath57}$ ] should be designed in a way to satisfy the following condition : @xmath58 } { \bf v}^{[1 ] } \right ) = \textrm{span } \left ( { \bf h}^{[32 ] } { \bf v}^{[2 ] } \right ) .
\label{se3 } \end{aligned}\ ] ] hence , @xmath37}$ ] , @xmath57}$ ] and @xmath59}$ ] should be chosen to satisfy ( [ se1 ] ) , ( [ se2 ] ) and ( [ se3 ] ) .
note that the channel matrices @xmath60}$ ] are full rank almost surely .
using this fact , ( [ se1 ] ) and ( [ se2 ] ) imply that @xmath61 } \right ) = \textrm{span } \left ( { \bf t } { \bf v}^{[1 ] } \right ) , \label{cae1 } \end{aligned}\ ] ] where @xmath62 } ) ^{-1 } { \bf h}^{[23 ] } ( { \bf h}^{[21]})^{-1 } { \bf h}^{[12 ] } ( { \bf h}^{[32 ] } ) ^{-1 } { \bf h}^{[31]}. \label{tm}\end{aligned}\ ] ] if @xmath37}$ ] could be designed to satisfy this criteria , according to ( [ se2 ] ) and ( [ se3 ] ) , we can obtain @xmath57}$ ] and @xmath59}$ ] using @xmath63 } = \left ( { \bf h}^{[32 ] } \right ) ^{-1 } { \bf h}^{[31 ] } { \bf v}^{[1 ] } , \label{cae2 } \end{aligned}\ ] ] @xmath64 } = \left ( { \bf h}^{[23 ] } \right ) ^{-1 } { \bf h}^{[21 ] } { \bf v}^{[1]}. \label{cae3 } \end{aligned}\ ] ] ( [ cae1 ] ) implies that there is at least one eigenvector of @xmath65 in @xmath66 } \right ) $ ] . since all channel matrices are diagonal , the set of eigenvectors for all channel matrices , their inverse and product are all identical to the set of column vectors of the identity matrix , i.e. the vectors of the from @xmath67^t$ ] .
since @xmath68 exists in @xmath66 } \right ) $ ] , ( [ se1])-([se3 ] ) imply that @xmath69 } { \bf v}^{[j ] } \right ) , \quad \forall i , j \in \{1 , 2 , 3\}. \end{aligned}\ ] ] thus , at receiver @xmath30 , the desired signal @xmath70 } { \bf v}^{[1 ] } $ ] is not linearly independent of the interference signal , @xmath71 } { \bf v}^{[2]}$ ] , and hence , receiver @xmath30 can not fully decode @xmath72 solely by zero forcing the interference signal . therefore ,
if the channel coefficients are completely random and generic , we can not obtain @xmath73 degrees of freedom for the three user single antenna ic through lia schemes .
in previous section , if the objective was to align interference at two of the receivers , receivers @xmath30 and @xmath41 for instance , it could be easily attained using ( [ se1 ] ) and ( [ se2 ] ) .
though , as discussed above , ( [ se3 ] ) which refers to interference alignment criteria at receiver @xmath42 , can not be satisfied simultaneously with ( [ se1 ] ) and ( [ se2 ] ) .
instead , assume channel matrices which contribute to interference at receiver @xmath42 would be of a form that already satisfies ( [ se3 ] ) , interference alignment would then be accomplished .
we can wait for the specific form of the channel to happen .
the question we intend to answer in the following is that what is the necessary and sufficient condition on channel structure to make perfect interference alignment feasible in finite channel extension .
the following theorem summarizes the main result of this paper .
[ 3usertheo ] in a three user ic , the necessary and sufficient condition for the perfect interference alignment to be feasible in finite channel extension is to have the following structure on the channel matrices : @xmath74 } ) ^{-1 } { \bf h}^{[23 ] } ( { \bf h}^{[21]})^{-1 } { \bf h}^{[12 ] } ( { \bf h}^{[32 ] } ) ^{-1 } { \bf h}^{[31]}\nonumber\\&&{=}\:{\bf p } \left [ \begin{array}{c c c } \tilde{{\bf t } } & 0 & 0 \\ 0 & \tilde{{\bf t } } & 0 \\ 0 & 0 & f(\tilde{{\bf t } } ) \end{array } \right ] { \bf p}^t , \label{caem}\end{aligned}\ ] ] where @xmath75 is a @xmath76 permutation matrix , @xmath77 is an arbitrary @xmath78 diagonal matrix with nonzero diagonal elements , and with @xmath79 in the range @xmath80 , and @xmath81 is a mapping whose domain is an arbirary @xmath78 diagonal matrix and range is a @xmath82 diagonal matrix @xmath83 whose set of diagonal elements is a subset of diagonal elements of @xmath84 .
theorem [ 3usertheo ] simply states that matrix @xmath65 has no unique diagonal element .
[ lemma1 ] assuming that @xmath85}$ ] is of rank @xmath21 , ( [ cae1 ] ) implies that @xmath21 eigenvectors of @xmath65 lie in @xmath66 } \right ) $ ] . from ( [ cae1 ] )
we conclude that there exists a @xmath86 dimensional matrix @xmath87 such that @xmath88}={\bf v}^{[1]}{\bf z}.\end{aligned}\ ] ] assume that @xmath89 is an eigenvector of @xmath90 i.e. , @xmath91 where @xmath92 is its corresponding eigenvalue , then @xmath37 } { \bf u } \not = 0 $ ] and we can write : @xmath88}{\bf u}={\bf v}^{[1]}{\bf z}{\bf u}=\gamma { \bf v}^{[1]}{\bf u}.\end{aligned}\ ] ] then @xmath37}{\bf u}$ ] , which is in @xmath93 } \right ) $ ] , is an eigenvector of @xmath65 .
since @xmath87 has @xmath21 orthogonal eigenvectors , then @xmath21 orthogonal eigenvectors of @xmath65 lie within @xmath93 } \right ) $ ] .
@xmath93 } \right ) $ ] should not contain any vector of the form @xmath18 , and since @xmath93 } \right ) $ ] has dimension @xmath21 , it should have @xmath21 basis vectors of the form @xmath94 , where at least @xmath41 of @xmath95 s are nonzero .
let s call vectors with this form as non @xmath18 vectors . since @xmath21 of @xmath65 s eigenvectors lie in @xmath93 } \right ) $ ] , the matrix @xmath65 should have at least @xmath21 non @xmath96 eigenvectors .
note that this requirement is necessary not sufficient .
assuming that @xmath97 $ ] is a matrix consisted of non @xmath96 eigenvectors of @xmath65 as its columns , it is concluded that @xmath93 } \right ) \in \textrm{span } \left ( { \bf s } \right ) $ ] .
[ lemma2 ] @xmath65 has no unique diagonal element , i.e. , if @xmath98 is the @xmath99th diagonal element of @xmath65 , there is at least one @xmath100 for which @xmath101 .
it is easy to see that if @xmath102 is an eigenvector of @xmath65 , then @xmath103 .
if @xmath98 is unique , this implies that non @xmath96 eigenvectors of @xmath65 do not contain @xmath104 , and hence , @xmath105 , where @xmath106 denotes the null space of columns of matrix @xmath107 .
thus , @xmath108 } \right ) $ ] because @xmath93 } \right ) \in \textrm{span } \left ( { \bf s } \right ) $ ] .
since all channel matrices are diagonal , using ( [ cae1])-([cae3 ] ) , @xmath109})$ ] implies that @xmath110 } { \bf v}^{[j ] } \right ) , \quad \forall i , j \in
\{1 , 2 , 3\}.\end{aligned}\ ] ] thus , at receiver @xmath30 , the total dimension of the desired signal @xmath70 } { \bf v}^{[1]}$ ] plus interference from undesired transmitters , @xmath111 } { \bf v}^{[j ] } , j \neq 1 $ ] , is less than @xmath20 , and desired signals are not linearly independent from the interference signals , and hence , receiver @xmath30 can not fully decode @xmath72 solely by zeroforcing the interference signal .
lemma [ lemma2 ] conlcludes the proof of necessary part of theorem [ 3usertheo ] .
the sufficient part is easily proved by noting the fact that the matrix @xmath112 with the form given in ( [ caem ] ) has @xmath113 non @xmath96 eigenvectors @xmath114 with the property that @xmath115 and @xmath116 where @xmath117 is defined as a @xmath118 matrix consisted of @xmath119 s as its columns .
every @xmath21 subset of these eigenvectors can be considered as the columns of user @xmath30 transmit precoding matrix @xmath37}$ ] .
@xmath57}$ ] and @xmath59}$ ] can be designed using ( [ cae2 ] ) and ( [ cae3 ] ) . as an example , assume that , using @xmath42 extension of the channel , @xmath120 diagonal matrix @xmath65 has the following form , @xmath121 which has the form given in ( [ caem ] ) with @xmath122 , @xmath123 , where @xmath124 is @xmath120 identity matrix , @xmath125 , and @xmath126 .
the matrix @xmath127 for this example case can be obtained as @xmath128.\end{aligned}\ ] ] note that this choice for the set of non @xmath18 eigenvectors of @xmath65 defined in ( [ texpl ] ) , which satisfies ( [ spnprp ] ) and ( [ krnlprp ] ) , is not unique .
every @xmath129 matrix @xmath37 } \in \textrm{span}({\bf r})$ ] can be considered as the user @xmath30 transmit precoding matrix .
@xmath57}$ ] and @xmath59}$ ] can be obtained using ( [ cae2 ] ) and ( [ cae3 ] ) .
for the rest of the paper , every matrix @xmath130 which can be written in the form of ( [ caem ] ) , with the same permutation matrix @xmath75 and mapping function @xmath81 , would be stated as @xmath131 .
it can easily be seen that if @xmath131 and @xmath132 , so is @xmath133 and @xmath134 .
if the condition ( [ caem ] ) is true with the following form @xmath135 { \bf p}^t , \label{caems}\end{aligned}\ ] ] where @xmath77 is an an arbitrary @xmath86 diagonal matrix , @xmath37}$ ] can be designed as @xmath136}={\bf p}^t \left [ \begin{array}{c } { \bf i}_n \\ { \bf i}_n \end{array } \right ] , \label{vdesig}\end{aligned}\ ] ] where @xmath75 is the same permutation matrix used in ( [ caems ] ) and @xmath137 is the @xmath86 identity matrix .
@xmath37}$ ] can also be designed as any other @xmath38 matrix having the same column vector subspace with ( [ vdesig ] ) .
@xmath57}$ ] and @xmath59}$ ] are determined accordingly using ( [ cae2 ] ) and ( [ cae3 ] ) , respectively .
assuming channel aiding condition with the form given in ( [ caems ] ) , consider the special case of @xmath138 } = { \bf t}_p , \quad \forall i , j , \quad i \not = j$ ] , then @xmath139 and the channel aiding condition is already satisfied .
the case of @xmath138 } = { \bf t}_p , i \not = j$ ] is the condition to satisfy the requirement of ergodic interference alignment in @xcite , therefore , ergodic interference alignment is the special case of the scheme presented in this paper .
channel aiding conditions obtained in this paper can be considered as the perfect interference alignment feasibility conditions on channel structure . stated
conditions on channel structure are not exactly feasible , assuming generic channel coefficients .
approximation can be used and its effect on residual interference can be analyzed .
overall , this paper aims at reducing the required dimensionality and signal to noise ratio for exploiting degrees of freedom benefits of interference alignment schemes .
n. lee , d. park , and y. kimi , `` degrees of freedom on the k - user mimo interference channel with constant channel coefficients for downlink communications , '' in _ proc .
ieee global commun .
_ , honolulu , hawaii , dec .
2009 , pp . | interference alignment(ia ) is mostly achieved by coding interference over multiple dimensions . intuitively , the more interfering signals that need to be aligned , the larger the number of dimensions needed to align them .
this dimensionality requirement poses a major challenge for ia in practical systems .
this work evaluates the necessary and sufficient conditions on channel structure of a 3 user interference channel(ic ) to make perfect ia feasible within limited number of channel extensions .
it is shown that if only one of interfering channel coefficients can be designed to a specific value , interference would be aligned perfectly at all receivers .
interference channels , interference alignment , degrees of freedom , generic channel coefficients , vector space . |
observations of the cosmic microwave background ( cmb ) with high precisions are of great importance in modern cosmology , since it is the most powerful way to understand the origin and evolution of our universe .
it is commonly believed that currently observed cmb anisotropies and large scale structure are originated from the quantum fluctuations at very early times preceding the hot expansion .
theoretical hypothesis about this very early universe includes inflation @xcite , ekpyrotic / cyclic universe @xcite and so on . according to these theories
the universe was smoothed and flattened at the early epoch , and at the same time the quantum fluctuations from vacuum extended outside the horizon and formed the primordial perturbations , i.e. , the initial conditions of cosmic structure formation .
the primordial perturbations can be decomposed into three types : scalar , vector and tensor perturbations .
the scalar perturbations cause the anisotropies of cmb and the structure formation of large scale structure , while the vector perturbations , correspond to vortices decline rapidly during the expanding universe and can be neglected at later time ( see e.g. @xcite ) .
more interestingly , the tensor perturbations , i.e. , the primordial gravitational waves @xcite leave the unique imprint on the large scales power spectrum of cmb b - modes polarization @xcite .
therefore , probes of cmb primordial b - modes polarization are always treated as crucial ways to test the early universe models .
searching for the primordial gravitational waves is one of the key scientific goals of the cmb polarization experiments , such as the bicep3 @xcite , the ebex @xcite , the spt polarization @xcite and the act polarization @xcite . in order to describe the primordial tensor perturbations
, we usually use the amplitude parameter , the tensor - to - scalar ratio , @xmath5 , where @xmath6 and @xmath7 denote the power spectra of tensor and scalar perturbations . due to the precision limitations of current cmb experiments ,
the constraints on @xmath0 are still weak and consistent with zero very well @xcite . in 2014 , the bicep2 collaboration published their final three - year cmb observational data and announced the detection of cmb b - modes polarization at scales @xmath8 @xcite , which corresponds to @xmath9 , and the null detection ( @xmath10 ) is disfavored at about @xmath11 . however , few months later , the planck collaboration measured the angular power spectra of the polarized thermal emission from diffuse galactic dust , which is the main foreground present in measurements of the cmb polarization at frequencies above 100 ghz @xcite , and pointed out that the contribution of thermal dust on the cmb bb power spectrum has been significantly underestimated in the bicep2 foreground analyses @xcite .
they found that the magnitude of the bb power spectrum , contributed by the polarized thermal dust emission , at 150 ghz in the sky field observed by the bicep2 experiment is at the same level as the one predicted by the early universe model with the primordial gravitational waves @xmath9 . when the polarized dust contribution is properly taken into account ,
the joint analysis of the bicep2/keck and planck collaborations preferred the null detection of the primordial gravitational waves and obtained the upper limit of the tensor - to - scalar ratio @xmath12 at 95% confidence level @xcite .
besides the polarized thermal dust emission , the cosmic birefringence can also generate the cmb b - modes polarization at large scales .
this phenomenon can be introduced by the interaction between cmb photons and the external scalar field @xmath13 through the chern - simons ( cs ) term @xmath14 . here
@xmath15 is the dual of the electromagnetic tensor .
this coupling is gauge invariant if @xmath16 .
this is possible if @xmath13 is a constant field over the spacetime or arises from the derivative of a cosmic scalar field @xmath17 .
the scalar field @xmath17 can be identified as the dark energy in the quintessential baryo-/leptogenesis @xcite and as the ricci scalar @xmath18 in the gravitational baryo-/leptogenesis @xcite .
the cs term violates the lorentz and cpt symmetries spontaneously if @xmath13 has nonzero background values .
this induces rotations of polarization directions of the propagating photons @xcite and productions of cmb tb and eb cross - correlations , even though they are absent before the recombination in the traditional cmb theory @xcite .
the crucial point for this paper is that this phenomenon provides a mechanism to produce the cmb b - modes , alternative to the primordial gravitational waves and weak lensing . because the rotation converts part of e - modes polarization to b - modes polarization , even the primordial b - modes polarization is absent , sizable cmb b - modes power spectrum
can still be obtained from the e - modes power spectrum through the rotation .
as shown in @xcite , the cosmological birefringence degenerates with primordial tensor mode perturbations at large scales and with the gravitational lensing at small scales , respectively . and due to this degeneracy , considering the anisotropic polarization rotation could slightly lower the best - fit value of @xmath0 and relax the tension on the constraints of @xmath0 between from bicep2 and from planck 2013 data @xcite , even without the help of the isotropic part of the rotation angle , which is removed by the `` self - calibration '' method in the bicep2 data analysis @xcite . therefore , it is important and necessary to study the influences of the polarization rotation angle on the measurements of cmb primordial tensor perturbations using the current and future cmb experiments
. recently , @xcite found that constraints on the tensor - to - scalar ratio could be significantly biased , when the isotropic rotation angle can not be taken into account properly . in this paper , our main task is to focus on the influences of the anisotropic polarization rotation angle on the detection of primordial gravitational waves by performing the global fitting with the latest cmb datasets and simulated mock data of future cmb experiments included .
this paper is organized as follows : in section [ theory ] we briefly introduce the cosmic birefringence mechanism and the effects on cmb polarization power spectra from the isotropic and anisotropic polarization rotation angle .
section [ result ] contains our main results from the current observations and future measurements , while section [ summary ] is dedicated to the conclusions and discussions .
in this section we briefly introduce the cosmic birefringence which has been widely studied in the literatures @xcite .
we consider the cs coupling between the cosmic scalar field @xmath17 and the electromagnetic field : @xmath19 where @xmath20 is an arbitrary function of @xmath17 . as mentioned before
, the cs term violates lorentz and cpt symmetries but not the gauge symmetry of the electromagnetic field , so will not change the number of dynamical components of photons .
however , the states with opposite helicity propagates with different velocities , so that the polarization vectors of photons will not be parallel transported along the light rays .
consequently the observed polarization directions are changed by rotations when compared with the original directions at the source . in terms of the stokes parameters @xmath21 and @xmath22 , the rotated linear polarization state ( denoted by a prime )
is related to the unrotated one through @xmath23 the rotation angle @xmath24 corresponds to the cpt violation and is frequency independent .
it can be calculated by the integration along the light rays @xcite @xmath25 where @xmath26 is the affine parameter of the light ray , the subscripts @xmath27 and @xmath28 represent the source and the observer , respectively .
naturally , we observe cmb photons at a single point , hence @xmath29 should be a constant .
however the dynamical field @xmath17 is expected to fluctuate on the last scattering surface ( lss ) .
the observer at the fixed position receives cmb photons from all directions and will find that the rotation angle varies across the sky , as pointed out in @xcite .
the anisotropies are proportional to the distribution of @xmath17 on the lss , which means we can separate the rotation angle @xmath24 into the isotropic part and the small anisotropic one as follows @xmath30 with @xmath31-f(\phi_0)~,\nonumber\\ \delta\alpha(\hat{\bf n})&=&\frac{df}{d\phi}\delta\phi(\vec{x}_{\rm lss } , \eta_{\rm lss})~,\end{aligned}\ ] ] where @xmath32 and @xmath33 denote the position and the conformal time on the last scattering surface . for the isotropic part @xmath34 ,
the first evidence in terms of the full cmb datasets was done in @xcite and stimulated many interests in this field ( see @xcite , and references within ) .
assuming the anisotropic rotation angle @xmath4 is a zero mean field defined on two dimensional sphere , it can be decomposed in terms of spherical harmonics : @xmath35 furthermore , if we assume the perturbation field @xmath4 is a gaussian random field and satisfies the statistical isotropy , so we can define the angular power spectrum of the anisotropic rotation angle : @xmath36 in this paper we set @xmath37 for simplicity , in which @xmath38 represents the cut - off mass in the effective field theory and @xmath39 stands for the dimensionless coupling coefficient .
we find the relationship between the angular power spectrum of anisotropic rotation angle @xmath40 and the power spectrum of the scalar field @xmath41 at the lss @xmath42 where @xmath43 is the fourier transform of @xmath44 and @xmath45 is the spherical bessel function .
@xcite has considered several specific dark energy models , such as the massless scalar field and the quintessence field , and then calculated their power spectra explicitly . in this paper
we study the constraint on the perturbations of this scalar field through the polarization rotation angle in a model - independent way .
it is convenient to expand its power spectrum on the super hubble scale , just as what we usually do in the study of slow - roll inflation , @xmath46 where @xmath47 , @xmath48 and @xmath49 represent the amplitude and spectral index of the power spectrum and the pivot scale , respectively . due to the precision limitation of current cmb experiments
, we set @xmath48 to be zero and neglect the higher order running parameters of the spectral index [ @xmath50 in the following calculations . in order to investigate the influence on the cmb polarization power spectra due to the cosmic polarization rotation angle
, we firstly define two - point correlation function using the following relation @xmath51 where @xmath52 corresponds to the angle between these two directions , @xmath53 . combining eqs .
( [ rotate ] ) and ( [ sphericalharmonics ] ) , one can express the rotated cmb polarization power spectra @xmath54 in terms of the unrotated ones @xmath55 ( see @xcite for details ) : @xmath56}\sum_{\ell'}\frac{2\ell'+1}{2}(c_{\ell'}^{ee}+c_{\ell'}^{bb } ) \int^{1}_{-1}d^{\ell'}_{22}(\beta)d^{\ell}_{22}(\beta)e^{4c^{\alpha\alpha}(\beta)}d\cos(\beta)\nonumber\\ & & { c'_{\ell}}^{ee}-{c'_{\ell}}^{bb}=\cos(4\bar{\alpha})\exp{[-4c^{\alpha\alpha}(0)]}\sum_{\ell'}\frac{2\ell'+1}{2}(c_{\ell'}^{ee } -c_{\ell'}^{bb})\int^{1}_{-1}d^{\ell'}_{-22}(\beta)d^{\ell}_{-22}(\beta)e^{-4c^{\alpha\alpha}(\beta)}d\cos(\beta)\nonumber\\ & & { c'_{\ell}}^{eb}=\sin(4\bar{\alpha})\exp{[-4c^{\alpha\alpha}(0)]}\sum_{\ell'}\frac{2\ell'+1}{4}(c_{\ell'}^{ee}-c_{\ell'}^{bb } ) \int^{1}_{-1}d^{\ell'}_{-22}(\beta)d^{\ell}_{-22}(\beta)e^{-4c^{\alpha\alpha}(\beta)}d\cos(\beta)\nonumber\\ & & { c'_{\ell}}^{te}=\cos(2\bar{\alpha})\exp{[-2c^{\alpha\alpha}(0)]}\sum_{\ell'}\frac{2\ell'+1}{2}c_{\ell'}^{te}\int^{1}_{-1 } d^{\ell'}_{02}(\beta)d^{\ell}_{20}(\beta)d\cos(\beta)\nonumber = c_{\ell}^{te}\cos(2\bar{\alpha})e^{-2c^{\alpha\alpha}(0)}\nonumber\\ & & { c'_{\ell}}^{tb}=\sin(2\bar{\alpha})\exp{[-2c^{\alpha\alpha}(0)]}\sum_{\ell'}\frac{2\ell'+1}{2}c_{\ell'}^{te}\int^{1}_{-1 } d^{\ell'}_{02}(\beta)d^{\ell}_{20}(\beta)d\cos(\beta)=c_{\ell}^{te}\sin(2\bar{\alpha})e^{-2c^{\alpha\alpha}(0)}~,\end{aligned}\ ] ] where @xmath57 is the auto - correlation of the anisotropic polarization rotation angle .
note that , in the calculations we have assumed the original cmb fields and anisotropic rotation angle field are statistically isotropic gaussian random fields .
we also show the theoretical cmb bb power spectrum with the tensor - to - scalar ratio @xmath58 ( magenta thin line ) for comparison . ] in the eqs .
( [ newcmb ] ) we have separated the influences of isotropic and anisotropic polarization rotation angle .
it is apparently that they both get involved in all scales of cmb polarization power spectra .
the isotropic rotation angle copies the original cmb ee power spectrum , zooms it out and then prints it onto the rotated bb power spectrum .
it will inevitably create additional cmb tb and eb power spectra which should be vanished in normal cmb theories .
we can see the terms @xmath59 or @xmath60 appear as coefficients in almost all equations .
this suggests us that the isotropic rotation angle should perform a global modulation on rotated cmb power spectra if it exists .
however , the situation for the anisotropic one is somehow different .
non - zero tb and eb spectra are not longer necessary if the isotropic part [ @xmath61 vanishes , which is consistent with some current cmb experiments .
the bicep2/keck array collaboration has used the `` self - calibration '' method for the detector polarization orientations , and as a result , any signal of the isotropic polarization rotation angle has been eliminated from data .
fortunately , the anisotropic part is not affected by this calibration method , and will still contaminate the observations of the primordial gravitational waves . in figure [ plotbb ]
, we plot the cmb bb power spectra for different values of @xmath62 $ ] with @xmath63 . for comparison
, we also show the expected cmb bb power spectrum generated by the primordial tensor perturbations @xmath58 .
interestingly , we find that , even the isotropic rotation angle is zero , the anisotropic polarization rotation angle with @xmath62 \sim 2.5 $ ] could still generate the cmb bb power spectrum to mimic cmb primordial b - modes signal with @xmath58 at large scales . at scales
@xmath64 , the contribution on bb power spectrum from the anisotropic polarization rotational angle can be negligible . therefore , probing the reionization bump of cmb bb power spectrum at very large scale may provide a smoking gun for distinguishing the polarization rotational angle and primordial gravitational waves .
in this paper we perform a global analysis to all the power spectra of the cmb data with the public available markov chain monte carlo package cosmomc @xcite , which has been modified to compute the rotated cmb polarization power spectra discussed above .
we assume the purely adiabatic initial conditions and impose the flatness condition motivated by inflation .
our basic parameter space is : @xmath65 , where @xmath66 and @xmath67 are the physical baryon and cold dark matter densities relative to the critical density , @xmath68 is the dark energy density relative to the critical density , @xmath69 is the optical depth to re - ionization , @xmath70 and @xmath71 characterize the primordial scalar power spectrum , @xmath0 is the tensor to scalar ratio of the primordial spectrum . for the pivot of the primordial spectrum we set @xmath72mpc@xmath73 . for the polarization rotation angle
, we set the isotropic rotation angle @xmath74 and then have one free parameter , @xmath75 , which represent the amplitude of power spectrum of the anisotropic rotation angle .
furthermore , in our analysis we include the cmb lensing effect , which also produces b - modes from e - modes @xcite , when we calculate the theoretical cmb power spectra .
in our calculations we mainly use the polarization power spectra from bicep2 experiments and the joint analysis data of bicep2/keck array and planck experiments : * the _ background imaging of cosmic extragalactic polarization _
( bicep2 ) @xcite , locating at the south pole , specializes in searching polarized cmb signal at large scales with low angular resolution , concentrating 150 ghz on a roughly 1@xmath76 patch of sky at high galactic latitudes .
recently the bicep2 collaboration released their three - year data accumulated from 2010 to 2012 and announced the detection of cmb b - modes polarization at scales @xmath8 .
it is noteworthy that they claimed that the cmb eb power spectrum is only used for the `` self - calibration '' of the detector polarization orientations @xcite .
any polarization rotation has been removed from the results . *
the products of the joint analysis of bicep2/keck array and planck data ( bkp2015 ) @xcite . very recently , bicep2/keck array and planck collaborations have re - analysed their data and published the cmb bb power spectrum at scales @xmath8 , in order to investigate the influence on the cmb b - modes measurements from the polarized thermal dust emission .
they convert the planck maps into the format usable for bicep2/keck array experiments , and make cross - spectra between the combined bicep2/keck maps and planck maps of the all polarized bands . similar with the bicep2 data ,
the `` self - calibration '' method was also applied in the data analysis . from the bicep2 data with @xmath77 ( red dashed line ) . ] due to the `` self - calibration '' method used in the bicep2/keck array experiments , the constraints on the isotropic rotation angle [ @xmath61 are quite consistent with zero @xcite .
therefore , we only use these data to study the anisotropies of the polarization rotation angle . furthermore , in order to take the contribution of polarized thermal dust emission on the cmb b - modes power spectrum into account properly , we add one more nuisance parameters @xmath78 in our calculations and consider the polarized dust bb power spectrum satisfies @xmath79 @xcite . in figure [ r_acpt_tri ] we show the constraints on the tensor - to - scalar ratio and the anisotropic polarization rotation angle , and their 2-dimensional contours from the bicep2 data .
firstly , we neglect the contribution on the cmb b - modes power spectrum from the polarization rotation angle , which means @xmath80 . due to the strong degeneracy between @xmath0 and @xmath81 , the huge uncertainty of the polarized dust b - modes power spectrum
significantly reduces the constraining power of bicep2 data on the tensor - to - scalar ratio @xmath0 .
the obtained @xmath82 upper limit is @xmath83 and there is a peak around @xmath84 , shown as the red dashed line in figure [ r_acpt_tri ] .
there is no signal for the primordial gravitational waves , which is consistent with the results from the bkp2015 data @xcite .
when we include the effect of anisotropic polarization rotation angle in the calculations , the constraint on the tensor - to - scalar ratio is slightly different .
as we mentioned in @xcite , the anisotropic polarization rotation angle would also contribute to the cmb bb power spectrum and partly explain the cmb b - modes data of bicep2 . in this case
, the 1-dimensional distribution of @xmath0 slightly moves towards to the smaller value , due to this degeneracy , see the black solid line in figure [ r_acpt_tri ] . and
the minimal value of @xmath85 becomes slightly lower , @xmath86 .
we also check this result using the latest bkp2015 data , which is shown in figure [ r_bkp2015 ] .
when we include the contribution on cmb b - modes power spectrum from the anisotropic polarization rotation angle , the upper limit of tensor - to - scalar ratio is @xmath87 ( 95% c.l . ) and the peak structure totally disappears .
these results imply that even considering the effect of polarized thermal dust emission , the anisotropic polarization rotation angle can still affect the constraint on the tensor - to - scalar ratio slightly . in figure [ r_acpt_tri ]
we also show the one - dimensional constraint on the amplitude parameter @xmath88 .
the current bicep2 data still do not have enough constraining power to limit @xmath88 accurately , so the obtained constraint on @xmath88 is consistent with zero , namely the 95% upper limit is @xmath62 < 3.8 $ ] .
however , we interestingly find that there is a peak around @xmath62 \sim 2.5 $ ] . as we show in figure [ plotbb ] , the model with @xmath62 \sim 2.5 $ ] could mimic the contribution of cmb b - modes power spectrum of primordial gravitational waves with @xmath58 at large scales .
and this degeneracy moves the distribution peak of @xmath0 towards to the lower value , which is similar with the results in @xcite .
finally , we plot the two - dimensional contour on the panel ( @xmath0,@xmath62 $ ] ) in figure [ r_acpt_tri ] . apparently , these two parameters are slightly correlated . the model with a larger value of @xmath62
$ ] provides more power of cmb b - modes , and consequently the contribution from the primordial gravitational waves is suppressed , namely the model with @xmath10 is more favored by the bicep2 data .
conversely , if we limit the parameter @xmath62 $ ] to be a very small value , the large value of @xmath0 will be needed , in order to match the bicep2 data . since the current cmb data are not accurate enough , the degeneracy between @xmath0 and @xmath62 $ ] is not very clearly . but
this could be important in the future cmb b - modes experiments .
therefore , it is worthwhile discussing the effect of non - zero anistropic polarization rotation angle on the measurement of cmb primordial b - modes signal for the future experiments . following @xcite , we simulate the future mock cmb power spectra with bicep3-like sensitivity : @xmath89 sky converge , the isotropic noise with variance @xmath90 , a symmetric gaussian beam of @xmath91 arcminutes full - width half - maximum ( fwhm ) @xcite , and the maximum multipoles @xmath92 . here
, we neglect the systematic errors of future cmb measurements and the contaminations from the cmb various foregrounds , such as the thermal dust emission and the synchrotron emission .
the basic fiducial model we choose is the best - fit planck model @xcite : @xmath93 , @xmath94 , @xmath95 , @xmath96 , @xmath97 , @xmath98}=3.08919 $ ] at @xmath99mpc@xmath73 .
firstly , we simulate the mock data using the basic fiducial model and the tensor - to - scalar ratio @xmath100 . when using the basic seven cosmological parameters ( @xmath80 ) to fit the mock data , we obtain a @xmath101 detection of non - zero value of tensor - to - scalar ratio : @xmath102 ( 68% c.l .
however , when we set the parameter @xmath62 $ ] to be free , the constraint on @xmath0 is obviously weakened , @xmath103 ( 68% c.l . ) and @xmath104 ( 95% c.l . ) , as shown in figure [ simulate1 ] .
note that , the standard deviations of @xmath0 we obtain are apparently underestimated , since in the calculations we do not consider any contamination from the cmb foregrounds .
this result implies that in the future cmb experiments , if we want to measure the signal of cmb primordial b - modes accurately , we need to know very well about the anisotropic polarization rotation angle .
otherwise , the degeneracy among them will contaminate the signal of primordial gravitational waves significantly .
derived from the simulated mock cmb data without including the anisotropic rotation angle .
the anisotropic rotation angles considered in the fiducial model are : @xmath62=0 $ ] ( blue dash - dotted line ) , @xmath62=1 $ ] ( red dashed line ) , and @xmath62=2 $ ] ( black solid line ) . ]
we also simulate another mock data using the basic fiducial model and the non - zero anisotropic polarization rotation angle . here
, we consider three fiducial values @xmath62=0,1,2 $ ] .
when we include the parameter of anisotropic polarization rotation angle in the calculations , the fiducial values of input parameters are always recovered .
however , if we neglect the effect of anisotropic rotation angle at all @xmath80 , the contribution of non - zero anisotropic rotation angle on the cmb bb power spectrum will be wrongly considered as the signal of the primordial cmb tensor perturbations . since the fiducial model @xmath10 , the bb power spectrum due to the primordial tensor b - modes should vanish .
but in the presence of non - zero fiducial rotation angle ( @xmath62=0,1,2 $ ] ) , the bb power spectrum should be non - vanishing .
if we force the anistropic rotation angle to be zero in the analysis , the value of @xmath0 will be enlarged to match the mock non - zero bb power spectrum . in figure
[ simulate2 ] , we find that the larger the fiducial value of @xmath62 $ ] is , the larger the obtained central value of @xmath0 becomes .
for example , in the calculation with the fiducial model @xmath62 = 2 $ ] , if we fix @xmath77 , the obtained constraint on the primordial tensor perturbations @xmath105 ( 68% c.l . ) .
the significance of this fake signal is about @xmath101 confidence level .
therefore , if we do not take the effect of polarization rotation angle into account properly , the constraints on the tensor - to - scalar ratio @xmath0 from the future cmb data with the high precision will be significantly biased .
one of the most important scientific aims in the cmb cosmology is searching for the primordial gravitational waves on the cmb b - modes power spectrum .
however , the signal is very small and could be contaminated by other sources easily , such as the cmb foregrounds , polarized thermal dust emission and synchrotron emission . in this paper , we discuss another mechanism , the cosmic birefringence , which can be introduced by the cpt - violating interaction between cmb photons and the external scalar field .
this mechanism can also produce the cmb b - modes power spectrum and mimic the signal of primordial gravitational waves at large scales .
here we summarize our main conclusions in more detail : * we use the bicep2 data and the joint analysis data of bicep2/keck array and planck experiments to constrain the tensor - to - scalar ratio .
when we include the contribution of polarized thermal dust emission on the cmb b - modes power spectrum , the constraints on @xmath0 are consistent with zero at @xmath106 confidence level .
furthermore , when we consider the effect of the anisotropic polarization rotation angle into the analyses , the degeneracy between @xmath0 and @xmath88 slightly changes the constraint on @xmath0 .
the one - dimensional distribution of @xmath0 slightly moves towards to the smaller value , since the anisotropic polarization rotation angle would also contribute to the cmb bb power spectrum and partly explain the cmb b - modes data .
* the current bicep2 data still do not have enough constraining power to study the polarization anisotropic rotation angle accurately , so the obtained constraint on @xmath88 is consistent with zero .
however , there is a clear peak around @xmath62\sim2.5 $ ] , which could mimic the contribution of cmb b - modes power spectrum of primordial gravitational waves at large scales , as shown in figure [ plotbb ] .
we also find the slight correlation between @xmath0 and @xmath62 $ ] in the two - dimensional contours .
* we simulate the mock cmb data using the bicep3-like sensitivity and study the degeneracy between the tensor - to - scalar ratio and the anisotropic rotation angle .
when we neglect the effect of nonzero @xmath88 , the mock data can constrain @xmath0 tightly . however , when including this effect into analysis , the obtained standard deviation of @xmath0 is significantly enlarged by a factor of three .
this degeneracy will contaminate the signal of primordial gravitational waves obviously .
* finally , we simulate the mock data with the different fiducial values .
if we neglect the effect of anisotropic rotation angle in the calculations , the contribution of non - zero anisotropic rotation angle on the cmb bb power spectrum will be wrongly considered as the signal of the primordial tensor perturbations . therefore , we need to take the effect of polarization rotation angle into account properly .
otherwise , the constraints on the tensor - to - scalar ratio @xmath0 could be biased from the future cmb experiments with high precisions .
s. l. and x. z. are supported in part by the national science foundation of china under grants no . 11121092 , no.11375202 and no .
x. is supported by the national youth thousand talents program and the nsfc under grant no .
m. l. is supported in part by nsfc under grant no .
11422543 , the fundamental research funds for the central universities , and the program for new century excellent talents in university .
h. l. is supported in part by the nsfc under grant no .
11033005 and the youth innovation promotion association project and the outstanding young scientists project of the chinese academy of sciences .
the research is also supported by the strategic priority research program `` the emergence of cosmological structures '' of the chinese academy of sciences , grant no . xdb09000000 . | searching for the signal of primordial gravitational waves in the b - modes ( bb ) power spectrum is one of the key scientific aims of the cosmic microwave background ( cmb ) polarization experiments .
however , this could be easily contaminated by several foreground issues , such as the interstellar dust grains and the galactic cyclotron electrons . in this paper
we study another mechanism , the cosmic birefringence , which can be introduced by a cpt - violating interaction between cmb photons and an external scalar field .
such kind of interaction could give rise to the rotation of the linear polarization state of cmb photons , and consequently induce the cmb bb power spectrum , which could mimic the signal of primordial gravitational waves at large scales .
with the recently released polarization data of bicep2 and the joint analysis data of bicep2/keck array and planck , we perform a global fitting analysis on constraining the tensor - to - scalar ratio @xmath0 by considering the polarization rotation angle [ @xmath1 which can be separated into a background isotropic part [ @xmath2 and a small anisotropic part [ @xmath3 . since the data of bicep2 and keck array experiments have already been corrected by using the `` self - calibration '' method , here we mainly focus on the effects from the anisotropies of cmb polarization rotation angle .
we find that including @xmath4 in the analysis could slightly weaken the constraints on the tensor - to - scalar ratio @xmath0 , when using current cmb polarization measurements .
we also simulate the mock cmb data with the bicep3-like sensitivity . very interestingly , we find that if the effects of the anisotropic polarization rotation angle can not be taken into account properly in the analysis , the constraints on @xmath0 will be dramatically biased .
this implies that we need to break the degeneracy between the anisotropies of the cmb polarization rotation angle and the cmb primordial tensor perturbations , in order to measure the signal of primordial gravitational waves accurately . |
natural sciences in general are based primarily on experiment , and , what is more , on quantitative experiment .
however , no series experiments can constitute a theory until a rigorous logical relationship is established between them .
theory not only allows us to systematize the available experimental results , but also makes it possible to predict new facts which can be experimentally verified @xcite .
statistical mechanics is very prominent part of theoretical physics .
it forms a separate field which includes the theory of phase transitions and is closely related with the more mathematical ergodic theory and some parts of probability theory . in the annals of statistical mechanics ,
the last century and recent decades mark a very vibrant epoch @xcite .
statistical mechanics is the art of predicting the behavior of a system with a large number of degrees of freedom , given the laws governing its microscopic behavior .
the statistical description of liquids , in comparison with the solid and gas phases , was incomplete for a long time @xcite . due to the simultaneous presence of strong interactions and large atomic displacements , common models and approximations used for
gases and solids , do not apply to liquids . for this reason
, liquids do not generally fall into any simple classication , and have as a result been mostly treated as general many - body systems .
therefore , the problem of formulating a rigorous mathematical description of all basic states of matter such as solids , liquids and gases based on a unified approach has always been regarded as illusive . in this work ,
we introduce the _ phonon hamiltonian _ with an interaction part reflecting the fact that interacting phonons can be created or annihilated making materials anharmonic over the heat production .
we derive an effective hamiltonian with two naturally emerged transverse phononic band gaps as a result of symmetry breaking in phonon interactions .
the emergence of two transverse phononic gaps in a liquid spectrum rigorously proves the frenkel s phenomenological microscopic picture about phonon excitations @xcite .
further , we construct the statistical mechanics of states of matter existing on the pressure - temperature ( p - t ) phase diagram ( see fig .
( [ fig1 ] ) ) by employing the debye approximation .
the introduced formalism covers the continuous theory of solids by debye @xcite , the phonon theory of liquids and supercritical fluids , the well - known thermodynamic limits such as the delong - petit law ( @xmath0 ) and the ideal gas limit ( @xmath1 ) , and we derive the new thermodynamic limit ( @xmath2 ) , dubbed here the frenkel line thermodynamic limit . in the framework of the phonon theory of liquids
we discuss the phonon propagation and localisation effects in disordered phases above and below the frenkel line , and explain the `` fast sound '' phenomenon . as a test for our theory
we calculate velocity - velocity autocorrelation ( vacf ) and pair distribution functions @xmath3 in the framework of the linear response green - kubo phenomenological formalism .
we show the consistency between dynamics ( phonon excitations and phononic gaps ) , structure ( pair correlations ) and thermodynamics ( heat capacity @xmath4 ) on the p - t phase diagram ( see fig .
( [ fig1 ] ) ) within the introduced formalism .
this point is of prime significance , since the manipulation of phononic band gaps in the so - called phononic crystals @xcite is customarily realized through structural engineering .
this is of crucial importance to implement heat management , since heat transfer in insulators uses phonons as carriers . in this work
, it is shown that the phononic gap manipulation can be achieved even in simplest materials through a suitable tuning of thermodynamic conditions .
this reflects into a tuning of the atomic dynamics .
tuning of the atomic dynamics and thermally triggered localization of low- and high - frequency propagating phonon modes are the key factors for the advancement of technologies based on the sound control and manipulation at the terahertz regime .
finally , we outline new directions towards future advancements in the scope of sound control technologies and thz phononic band gaps engineering .
for most cases of interest , dynamics of a complex condensed matter system can be effectively described by collective excitations of quantised lattice vibrations , the so - called phonons . in the most general form
the corresponding hamiltonian can be written as @xmath5 , where @xmath6 defines a free theory @xmath7\ ] ] and the interaction potential @xmath8 is of higher order in fields .
the coefficient @xmath9 here gives the dispersion relation of a phonon and the additional dimensionless prefactor @xmath10 will be needed further to identify the states of matter ( see table i ) .
the hat emphasises that the resulting frequency for longitudinal and transverse modes will be different after symmetry breaking . following the construction @xcite to describe the major states of matter by a single model we introduce the following higgs - like potential with a metastable vacuum which leads to spontaneous symmetry breaking @xmath11 \label{hint}\ ] ] here
, @xmath12 is a multiindex @xmath13 , @xmath14 is the debye frequency , the parameter @xmath15 that takes values 1 or 0 , @xmath16 are some real non - negative couplings @xcite . the collective canonical coordinates
@xmath17 and @xmath18 are introduced as follows @xmath19 here , @xmath20 is space coordinate of an atom of a lattice sitting in a vertex labelled by the multiindex @xmath21 whose components run @xmath22 .
hence the total number of atoms is @xmath23 .
the row @xmath24 gives the maximal size of the lattice in each direction .
in addition , @xmath25 is the size of each lattice cell , @xmath26 is the imaginary unit ( @xmath27 ) and @xmath28 is the mass of an atom in the lattice . to keep the coordinates @xmath29 real
, one imposes the condition @xmath30 , where star denotes the complex conjugation .
the above construction is the discrete version of the inverse fourier transform @xmath31 however , we avoid using the vector notation for multiindex labelling a lattice vertex so as to not to confuse it with a space vector labelled by small greek indices . note , that the model is defined for the canonical collective coordinates @xmath32 and @xmath33 that describe dynamics of phonons in the reciprocal space . hence , although we have started with a lattice of finite spacing @xmath34 there is still an @xmath35 symmetry associated with the collective excitations acting on the small greek indices @xmath36 the hamiltonian is apparently invariant under these transformations for any @xmath37 .
obviously , this transformation acts only on the phonon fields and does not transform the lattice itself .
the kinetic energy is minimal at the static configurations @xmath38 and minima of the potential @xmath39=\sum_{0\leq\hat{\w}_q\leq\omega_{\rm d}}\left[\frac{\mu\hat{\w}_q^2}{2 } |q_q^\a|^2-\frac{\sigma}{2 } |q_{q}^\a|^{4}+\frac{\theta}{6}|q_{q}^\a|^6\right]\ ] ] can be found in the standard way by varying along the coordinate @xmath40}{\delta q^\beta_q}=0 $ ] .
in general , there are six solutions to a six order polynomial .
however in this case two roots coincides , and two are negative and hence should be dropped . in total , keeping only non - negative solutions we have @xmath41 where @xmath42 plays the role of the frenkel frequency ( see fig .
( [ fig2 ] ) ) .
the square root in the non - trivial solutions @xmath43 implies that @xmath44 cuts the spectrum of modes and is related to the viscosity and infinite - frequency shear modulus of a liquid .
indeed , using the maxwell relation @xmath45 the frenkel frequency can be represented as @xmath46 where @xmath47 is viscosity , g@xmath48 is the infinite - frequency shear modulus and @xmath49 is the average time between two consecutive atomic jumps at one point in space .
this equation is noteworthy since it relates parameters of the potential to measurable variables .
oscillation frequency of excitations @xmath50 in a solid body is bound from above by the debye frequency @xmath51 .
hence , in the most general case one has @xmath52 .
this implies very different behaviour of modes with the oscillation frequency @xmath50 above and below the frenkel frequency @xmath44 .
indeed , for modes with @xmath53 one has to keep the solution @xmath54 , and hence phonons propagate over the trivial vacuum with unbroken @xmath35 symmetry .
as we will see , this corresponds to keeping all three modes , longitudinal and two transverse , in the spectrum .
in contrast , the trivial vacuum appears to be not the true vacuum for modes with @xmath55 .
these are allowed to propagate over the vacuum @xmath43 with broken symmetry .
this corresponds to removing transverse modes from the spectrum .
hence , we see that symmetry breaking occurs when the frequency @xmath50 crosses the bound @xmath44 .
propagate over the trivial vacuum @xmath56 , where @xmath57 .
( b ) modes with @xmath58 propagate over the vacuum @xmath59 with broken symmetry , where @xmath60 . ]
hamiltonian over the vacuum @xmath61 can be obtained in a trivial way and basically has the same form as the initial one .
consider now excitations of the phonon field around the ground state @xmath62 @xmath63 for a particularly chosen vacuum @xmath64 we obtain the following hamiltonian @xmath65 & = \frac{1}{2}\sum_{0\leq\hat{\w}_q^{l , t , t}\leq\omega_{\rm d}}\pi_q^\a\pi_{-q}^\a+\sum_{0\leq\hat{\w}_q^l\leq\omega_{\rm d}}\left[\frac{\m \w_q^2}{2}\varphi^1_q\varphi^1_{-q}\right]+\\ & + \sum_{\omega_{\rm f}\leq\hat{\w}_q^{t , t}\leq\omega_{\rm d}}\left[\frac{\mu\w_q^2}{2}(\varphi^2_q\varphi^2_{-q}+\varphi^3_q\varphi^3_{-q})\right]\\ & + v_{int}[\f_q^\a]+v_0 , \end{aligned}\ ] ] where @xmath66 we call transverse modes and the mode @xmath67 we call longitudinal since it becomes naturally separated
. the corresponding canonical momenta are denoted by @xmath68 .
all terms of the order higher than two in fields are collected into the interaction potential @xmath69 .
a constant term @xmath70 remains that does not affect the dynamics and can be dropped .
in addition we define oscillation frequency @xmath71 as follows @xmath72 it is important to mention that @xmath73 is continuous at the point @xmath74 and hence the longitudinal mode does not fill the bound between low- and high - frequency regimes : @xmath75 and @xmath76 , respectively .
for the transverse modes @xmath77 the situation is drastically different . due to the goldstone theorem
there are no terms quadratic in @xmath77 in the region @xmath78 where the symmetry is broken . in quantum field theory ( qft )
this situation corresponds to having massless excitations in the spectrum . from the point of view of condensed matter physics
this implies a dynamical cut - off of transverse modes . in other words ,
there are no transverse modes with @xmath79 .
such behaviour allows reproducing energy spectrum of a gas , a liquid and a solid in a single framework . referring the reader to @xcite for details we repeat here the table with results for completeness .
[ tab ] p2cmp3.5cmp1.5 cm + phase & coupling constants & normal modes + + .states of matter . *
ideal gas * : no collective excitations ; * fluid below frenkel line * : only longitudinal excitations @xmath80 ; * liquid above frenkel line * : both longitudinal ( @xmath80 ) and transverse ( @xmath81 ) modes ; and * solid * : all modes are supported ( @xmath82 ) . [ cols= " < , < " , ] & no phonon modes / no interactions & no order / disorder + & 1-l+2-t ( @xmath83 ) phonon modes & medium - range order + & 1-l phonon mode & short - range order + [ 1ex ] [ table1 ] the emergence of local - order heterogeneities ( non - uniformities ) and the existence of topologically confined transverse phonon modes ( standing plane waves ) on an intermediate length scale are closely interrelated . from equation ( [ ham ] ) we know that liquids support high - frequency propagating transverse phonon modes but lose this ability progressively as temperature increases extending well into the supercritical region .
oxygen atoms in the second coordination shell define a boundary between propagating and localized ( topologically confined sound waves ) phonon modes in the supercritical co@xmath84 @xcite , @xmath4 : @xmath85 @xcite , where @xmath1 is the ideal gas thermodynamic limit .
0.8 nm value is estimated as the length scale for highly localized transverse phonon modes which corresponds to the size of clusters where the phonon packets are localized .
similarly , the propagating lengths of the transverse acoustic modes were determined to be 0.4 nm1 nm , corresponding to the size of cages formed instantaneously in liquid ga @xcite and liquid sn @xcite .
thus , the supercritical state is also amenable to supporting non - propagating ( localised ) transverse phonon excitations on an intermediate length scale .
a rigorous mathematical description of liquid thermodynamics has always been regarded as much more difficult than that of the kinetic theory of gases or collective displacements of crystalline solids @xcite .
it is now known that liquids do not have a simple interpolated status between a gas and a solid phases , nevertheless , similarities to the properties of both adjacent phases can certainly be observed @xcite .
the coefficients of self - diffusion of liquids ( 10@xmath86 @xmath87s@xmath88 ) and solids ( 10@xmath89 @xmath87s@xmath88 ) are orders of magnitude below those of gases .
viscosities of fluids and gases are some 13 orders of magnitude lower than those of solids , which we may easily understand in terms of momentum exchange . in terms of vibrational states , liquids differ from solids because they can not support static shear stress ( @xmath90 ) .
however , liquids support shear stress at high frequencies ( @xmath91 ) .
flow in a solid arises primarily from rupturing of bonds and the propagation of dislocations and imperfections . in a liquid
, flow is characterized by both configurational and kinetic processes , whilst in a gas the flow is understood purely in terms of kinetic transport . in this very limited sense , liquids may have a minor partial interpolated status between gases and solids @xcite . here , we recall the analytical expression for the liquid energy derived earlier ( see eq .
( [ e ] ) ) @xmath92 analogously , we represent the formula for liquid energy below the frenkel line on the pressure - temperature diagram recalling the analytical expression ( see eq .
( [ fen ] ) ) @xmath93 the comparison of both dynamic ( eqs .
( [ e1])-([e2 ] ) ) and structural ( eq . ( [ e3 ] ) ) approaches to liquid energy we discuss below . alternatively , internal energy @xmath94 of a system can be calculated by computing the pair distribution function @xmath3 . here , we consider the spherical shell of volume @xmath95 which contains @xmath96 particles , where @xmath97 is the number density .
if the pair potential at a distance @xmath98 has a value @xmath99 , the internal energy of interaction between the particles in the shell and the central particle is @xmath100 .
the total energy of @xmath101 interacting particle is given as @xmath102 and the pair distribution function @xmath3 can be expressed through the static structure factor @xmath103 @xmath104\frac{\sin{qr}}{qr}q^{2}dq\ ] ] which can be measured in diffraction experiments . in general , both dynamic ( based on phonon contributions , see eqs .
( [ e1])-([e2 ] ) ) and structural ( based on contributions from pair correlations , see eq .
( [ e3 ] ) ) approaches to internal energy and heat capacity can be considered as equivalent . however , these approaches operate differently considering specific thermodynamic limits . regarding solids , approaching
the quantum limit at low temperatures the dynamic approach easily accounts for quantized portions of energy and predicts low-@xmath105 thermodynamic regime for solids ( @xmath106 ) ( for detail see subsection _ low - t quantum limit _ ) .
in contrast , the structural approach has a limitation describing solids at low temperatures .
even , in case of weak interactions the second term in eq .
( [ e3 ] ) occurs negligible and the minimum achievable heat capacity becomes @xmath107 reaching the ideal gas thermodynamic limit only , far beyond the quantum limit though .
nevertheless , both @xmath108 and @xmath109 approaches are interconnected over a wide range of pressure and temperature on the phase diagram ( see table iii ) and it can also be observed through calculation the green - kubo transport coefficients .
green and kubo showed @xcite that phenomenological coefficients like self - diffusion coefficient , thermal conductivity and shear viscosity could be written as integrals over a time - correlation function in thermal equilibrium .
lars onsager stated in his regression hypothesis @xcite that fluctuations are present in every equilibrated system .
dissipation of fluctuations has the same origin as the relaxation toward equilibrium once the system is out of equilibrium due to an external force .
both the dissipation and the relaxation time scales are determined by the same transport coefficients . in the standard linear response theory for a system subject to an external field ,
the green - kubo formula for a linear response coefficient can be expressed in terms of time- and/or distance dependant correlation function .
let us consider time - dependant @xmath110 and distant - dependant @xmath111 correlation functions , where @xmath112 represents the average over initial conditions .
function eq .
( [ ct ] ) is not a function of time but the function of the shift in time or correlation time .
the same is applicable also to correlations in space ( see eq .
( [ cd ] ) ) , instead of time .
vibrational spectrum for a system can be calculated by taking the fourier transform of the correlation function , @xmath113 that transfers the information on the correlations along the atomic trajectories from time to frequency frame of reference @xmath114 where @xmath113 is the velocity - velocity autocorrelation function for an ensemble of n particles @xmath115 therefore , the vibrational spectrum of a system @xmath116 on the other hand , the structural spectrum can be measured in neutron and x - ray scattering experiments @xmath117 \label{i2}\end{aligned}\ ] ] because of eq .
( [ gor ] ) and @xmath118 is the form factor . in general , combining results from eqs .
( [ i1 ] ) and ( [ i2 ] ) one may write @xmath119 mathematically repeating the reconstruction of experimental measurement procedure . hence
, both vibrational and structural correlations can be studied independently in temporal and spatial domains , respectively . as dynamic and structural approaches
are interlinked and complementary , change in structure with temperature variations leads to change in dynamics and @xmath120 @xmath121 , which is consequence of the equation below @xmath122 therefore , dynamic ( phonon approach ) and structural ( pair correlations approach ) properties in condensed phases change consistently with temperature variations .
below we test this general statement studying evolution of correlation functions within the green - kubo phenomenological transport coefficients formalism . to calculate these correlation functions , we have used lammps simulation code to run a lennard - jones ( lj , @xmath123=119.8 k , @xmath124=3.405 ) fluid fitted to ar properties @xcite with 32678 atoms in the isothermal - isobaric ( npt ) ensemble .
an explicit calculation of the velocity - velocity autocorrelation function ( vacf ) and of its frequency spectrum for liquid argon in particular has been well studied @xcite .
if the velocity vector for a system of atoms is @xmath125 , then the velocity autocorrelation can be written as @xmath126 and @xmath127 , where @xmath128 is the diffusion constant . in a liquid
the vacf is not a simple exponential but has a negative part as its characteristic feature .
the decay of the correlation function , in a liquid in particular , reflects the decay in the correlations in atomic motion along the trajectories of the atoms , not in the amplitude .
@xmath129 has the long time tail which can be explained as a collective dynamic effect : part of the momentum of a particle is stored in a microscopic vortex that dies off very slowly @xcite .
if the motion of the atoms tends to an oscillatory pattern , the vacf will characterize the ossilations because the velocity of the atoms will self - correlate in a periodic fashion .
similarly , if the velocity of atoms tends to one direction , then the vacf will gradually grow in magnitude characterizing the diffusion of the atoms . in fig .
( [ vacf ] ) we plot the evolution of vacf with temperature variations at fixed pressure for liquid ar and evidence the dynamic crossover : the continuous transition from oscillatory to ballistic - collisional motion respectively @xcite .
distant - dependant correlation function @xmath130 , see eq .
( [ cd ] ) , can be expressed as ( also see @xcite ) @xmath131 for isotropic system @xmath3 can be averaged over angles by calculating an average number of particles at distances @xmath98 and @xmath132 @xmath133 in fig .
( [ pdf ] ) we show the evolution of @xmath3 with temperature variations at fixed pressure for liquid ar and evidence of the structural crossover : the continuous transition from medium - range order pair correlation to short - range order pair correlation respectively @xcite .
the structural crossover in the supercritical state is the new effect ( see fig .
[ pdf ] ) , in view of the currently perceived physical homogeneity of the supercritical state . here
, we show that a liquid ar undergoes a structural crossover which corresponds to the qualitative change of atomic structure , the transition of the substance from the `` rigid '' liquid structure ( pair correlations presented at an intermediate length scale ) to the `` non - rigid '' fluid structure ( ( pair correlations preserved at a short length scale only ) .
it is noteworthy , that any approach what bears a `` small parameter '' and makes it possible to expand the internal energy of the system into the series such as virial expansion , cluster expansion , meyer expansion , percus - yevick approach and hard - sphere model is only capable of explaining weakly interacting systems @xcite .
however , all these approaches fail in describing the systems with strong interactions like @xmath134/compressed liquids .
hence , these approaches are unable to predict and describe the structural crossover in the supercritical state @xcite . at fixed pressure p=0.8 gpa .
the @xmath3 functions of simulated one - component lennard - jones ( lj ) fluid at different temperatures showing the disappearance of the medium - range order upon crossing the structural boundary in the supercritical state between t=400 k and t=500 k. ] we now address the origin of the structural crossover ( see fig .
[ pdf ] ) , and relate this origin to the changes of both dynamics ( see fig .
[ vacf ] ) and thermodynamics ( see tables [ table]-[table1 ] ) of the liquid matter . indeed , @xmath3 peaks ( their heights and positions ) on both sides of the structural crossover change differently with temperature increase . in the _ rigid liquid _ regime ,
@xmath3 peaks decrease rapidly due to the exponential decrease of the relaxation time ( also see @xcite ) . in the _ non - rigid fluid _ regime where the oscillatory component of motion ( see fig . [ vacf ] ) and the medium - range order pair correlations are no longer present ( see fig . [ pdf ] ) ,
pair correlations are less sensitive to temperature increase because the dynamics is already randomized by ballistic motions as in a gas meaning that the system reached the frenkel line thermodynamic limit @xmath2 .
this picture supports the relationship between the structure , dynamics and thermodynamics and its consistent temperature evolution over a wide range of temperature and pressure on the phase diagram ( also see tables [ table]-[table1 ] ) .
the interatomic interactions in a liquid are strong , and therefore firmly affect calculations of its thermodynamic properties . at the same time ,
liquids , occupy an interesting intermediate state with a combination of strong interactions and cohesive state as in solids and large flow - enabling particle displacements as in gases , the calculation of liquid energy in particular requires the explicit knowledge of the interactions . for this reason
, it is argued that no general expressions for liquid energy can be obtained , in contrast to gases and solids @xcite .
an apt summary of this state of affairs , attributed to landau , is that liquids `` have no small parameter . ''
perhaps for this reason , liquid energy and heat capacity have not , or barely , been mentioned in statistical physics textbooks as well as books dedicated to liquids @xcite .
this is believed have precluded the calculation of thermodynamic properties of liquids in general form . nevertheless , in the last few decades a great amount of activity has taken place in the field of statistical mechanics of disordered materials and liquids in particular .
this activity was mainly originated by the development of various theories and novel experiments with which the predictions of these theories could be tested .
one of the most outstanding examples of these theories are the mode - coupling theory ( mct ) @xcite and the fundamental - measure theory by rosenfeld and tarazona which gives a unified analytical description of classical bulk solids and fluids @xcite , and describe scaling in fluids @xcite . in this work
, we propose an unified approach to states of aggregation existing on the p - t phase diagram : solid , liquid and gas phases .
we derive the effective hamiltonian with low - energy cutoff in two transverse phononic polarisations as a result of symmetry breaking in phonon interactions .
this result rigorously proves the frenkel s intuitive microscopic picture about phonon excitations in condensed bodies @xcite on the fundamental basis .
further , we construct the statistical mechanics of states of aggregation existing on the p - t phase diagram by employing the debye approximation .
the introduced formalism also covers the well - known thermodynamic limits such as the delong - petit thermodynamic limit ( @xmath0 ) and the ideal gas limit ( @xmath1 ) . in the framework of the proposed formalism
we derive the new thermodynamic limit @xmath2 , dubbed here the frenkel line thermodynamic limit . in the framework of the phonon theory of liquids @xcite , which we derive from the effective hamiltonian ( [ eham ] )
, we discuss the phonon propagation and localisation effects in liquids above and below the frenkel line , and explain the `` fast sound '' phenomenon . as a test for our theory
we calculate velocity - velocity autocorrelation and pair distribution functions using the linear response phenomenological formalism ( introduced by green and kubo ) . as a result
, we show the consistency between dynamics , structure and thermodynamics on the p - t phase diagram in the framework of the introduced unified approach . in particular , we prove that dynamic , structure and thermodynamics evolve consistently upon crossing the frenkel line .
the existence of the frenkel line ( can be predicted from eq .
( [ eham ] ) ) is important for studies fluids under extreme conditions of pressure and temperature .
recently , it was shown that in jupiter and saturn , supercritical molecular hydrogen undergoes a smooth dynamic transition around 10 gpa and 3000 k from the _ rigid liquid _ regime to the _ non - rigid fluid _ regime upon crossing the frenkel line @xcite .
the continuous transition is accompanied by qualitative changes of all major physical properties .
the implication of that work provides a physically justified way to demarcate the interior and the atmosphere in gas giants @xcite .
a similar work was done on supercritical carbon dioxide providing a compelling evidence for a structural crossover @xcite in the supercritical state @xcite .
an intriguing question whether venus may have had co@xmath135 oceans was raised urging experimental groups for the verification of the proposed hypothesis @xcite . from eq .
( [ eham ] ) we derive the phonon theory of liquid thermodynamics @xcite ( in the _ rigid liquid _ regime section ) which has successfully been applied to calculating various thermodynamic properties in nonofluidics @xcite and at solid / liquid interfaces @xcite by other research groups .
phononic band gaps manipulation ( see the last term in eq .
( [ eham ] ) ) is perhaps the most obvious application in a scope of sound control technologies .
sound is immensely valuable in our daily lives for communication and information transfer .
the existence of liquid structures with complete phononic band gaps has obvious applications .
for example , a phononic crystals immersed in a liquid or aqueous solution will reflect incoming sound waves with frequencies within the gap and can therefore be used as an acoustic insulator .
moreover , the introduction of defects within hybrid structures ( dissolved nano - clusters in liquids ) allows sound waves with frequencies in the band gap to be trapped near a point - like defects and/or guided along linear defects .
therefore , we expect that the reported results in this work will advance technologies based on the thz phononic band gaps control .
this work was partly supported by the u.s .
department of energy , office of science , office of basic energy sciences , under contract no .
de - sc0012704 .
we are indebted to stefano ruffo , yang zhang , salvatore torquato , oleg kogan , ivar martin , yugang zhang and oleg gang for stimulating discussions . | we introduce an unified approach to states of matter ( solid , liquid and gas ) and describe the thermodynamics of the pressure - temperature phase diagram in terms of phonon excitations .
we derive the effective hamiltonian with low - energy cutoff in two transverse phonon polarisations ( phononic band gaps ) by breaking the symmetry in phonon interactions .
further , we construct the statistical mechanics of states of aggregation employing the debye approximation .
the introduced formalism covers the debye theory of solids , the phonon theory of liquids , and thermodynamic limits such as the delong - petit thermodynamic limit ( @xmath0 ) , the ideal gas limit ( @xmath1 ) and the new thermodynamic limit ( @xmath2 ) , dubbed here the frenkel line thermodynamic limit .
we discuss the phonon propagation and localisation effects in liquids above and below the frenkel line , and explain the `` fast sound '' phenomenon . as a test for our theory
we calculate velocity - velocity autocorrelation and pair distribution functions within the green - kubo formalism .
we show the consistency between dynamics of phonons and pair correlations in the framework of the unified approach .
new directions towards advancements in phononic band gaps engineering , hypersound manipulation technologies and exploration of exotic behaviour of fluids relevant to geo- and planetary sciences are discussed .
the presented results are equally important both for practical implications and for fundamental research . + * email : [email protected] |
the high - temperature expansion can be used in any dimension and has provided significant information on variety of models.@xcite it is based on the taylor expansion of the boltzmann factor @xmath1 in @xmath2 , around the high - temperature limit . although the concept of the high - temperature expansion is quite simple , its high - order calculation requires sophisticated strategies . a standard method of the high - temperature expansion
is called the finite cluster method@xcite ( fcm ) .
it has been used for a long time,@xcite and progress of the high - temperature series has been due mainly to improvement of computers . in the fcm , the series expansion is reduced to calculation in connected finite - size clusters
. however , the fcm is inefficient for calculating a correlation function between rather distant sites : ( i ) the clusters have to include those two sites and calculation in such large clusters consumes a lot of time .
( ii ) the fcm does not use a valuable piece of information .
that is , the high - temperature series of a long - range correlation starts at a certain finite order , namely , low - order coefficients are equal to zero physically , it means that a long - range correlation develops at low temperature . in this paper , we formulate a new method oriented to the long - range correlation , namely , it overcome ( i ) and ( ii ) above .
the new method is valid for the spin-1/2 heisenberg model , or , more generally , for su(@xmath0 ) heisenberg model with arbitrary @xmath0 .
high - temperature series coefficients are written in terms of cumulants .
we start from the lowest - order nonzero cumulant in the correlation - function series .
next , we consider a deviation from it
. then , the deviation can be regarded as a sort of moment , which we call a quasi - moment .
the corresponding quasi - cumulant enables us to calculate a number of terms higher than the lowest - order nonzero cumulant . in ref . , the fourier transform of the correlation function is calculated using the new method and the fcm complementarily . that is , we have used the new method for the long - range correlations and the fcm for the short - range correlations .
in addition , also for the specific heat , we have adopted a similar strategy in ref . .
that is , contributions from the required largest cluster are calculated by a new technique , and the fcm is used only for the smaller clusters .
we utilize cumulants relevant to the ordering of quantum variables . by choosing nonzero cumulants from them , some of the series coefficients
are calculated simply .
we explain also this new technique in this paper .
the structure of this paper is as follows . in sec .
[ sec : model ] , we present the model .
we show a representation of su(@xmath0 ) heisenberg model in terms of exchange operators .
the relation of a correlation function and an expectation value of the exchange operator is mentioned . in sec .
[ sec : htse ] , we review the high - temperature expansion .
although we use this model for illustration , what is written here can generally be applied to other models . to clarify the purpose of our new method , we review the fcm in detail using a one - dimensional system as an explicit example . in sec . [ sec : newcor ]
, we describe the new method for the correlation function . here
, we derive some properties of the cumulants . owing to those properties
, we can define quasi - moments / cumulants . how to calculate series coefficients using those quasi - cumulants is explained . in sec .
[ sec : newsh ] , we show the new technique for the specific heat .
finally , we give a summary in sec .
[ sec : con ] .
let us consider the spin-1/2 heisenberg model , @xmath3 where @xmath4 is the spin operator at site @xmath5 , and @xmath6 is the coupling constant . here
, @xmath7 represents that the summation is performed only once for each @xmath7 pair .
the spin operators are rewritten as @xmath8 where @xmath9 is an exchange operator defined by @xmath10 for arbitrary @xmath11 , and @xmath12 .
then , we can rewrite eq .
( [ hamilsu2 ] ) as @xmath13 this hamiltonian has su(2 ) symmetry . since our method is valid more generally for an su(@xmath0 ) symmetric case with arbitrary positive integer @xmath0 , we define the su(@xmath0 ) heisenberg model below .
let each site take one of the @xmath0 colors , and denote them as @xmath14 with @xmath15 .
an exchange operator is defined using @xmath16 by @xmath17 for @xmath18 .
colors at sites @xmath5 and @xmath19 are exchanged by @xmath9 , namely , eq .
( [ eq : exchangeop ] ) is satisfied for arbitrary colors @xmath20 and @xmath2 .
furthermore , for a later convenience , let us define @xmath21 in the case of su(2 ) , @xmath22 is satisfied .
although we mostly use expressions for general @xmath0 in this paper , we expect that eq .
( [ eq : su2spexp ] ) helps instant understanding of explanations . in some cases
, we explicitly use the su(2 ) notation for a simple explanation .
the hamiltonian of the su(@xmath0 ) heisenberg model is given by @xmath23 the constant term just yields an energy shift .
hence , for calculating thermal averages , we can use @xmath24 we define the partition function as @xmath25 and the free energy as @xmath26 the average in this system is denoted by @xmath27 where @xmath28 is an arbitrary operator . in the case of spin-1/2 system , there are three independent interacting components , @xmath29 . in general , when there are @xmath0 states per site , the maximum number of independent interacting components is @xmath30 .
the model , isotropic with respect to these components , is the su(@xmath0 ) heisenberg model as explicitly shown in appendix [ app : derive ] .
what we present in this paper is the high temperature expansion of @xmath31 .
a correlation function @xmath32 is calculated using a relation , @xmath33 for arbitrary @xmath34 , @xmath18 .
in addition , when @xmath35 , @xmath36 for arbitrary @xmath34 .
these relations are shown also in appendix [ app : derive ] . in the case of su(2 ) , @xmath37 .
the uniform susceptibility is equal to inverse temperature times the fourier transform of the correlation function with wave - number zero .
furthermore , the specific heat is calculated from the internal energy @xmath38 .
in this section , we review high - temperature expansion . although the heisenberg model is used for explanation , this review is more general .
that is , one can easily replace the heisenberg model with a different model .
fundamental properties of moments and cumulants are frequently used in this paper . in order to conveniently refer to those properties ,
let us review moments and cumulants of classical variables @xmath39 .
we put most of the fundamental details of the review in appendix [ appn : fdclvbl ] , and let us here just show one important formula used frequently in this paper . here , a moment is denoted by @xmath40 , and a cumulant is denoted by @xmath41_x$ ] .
moments can be expanded using cumulants , and vice versa . however , in those relations , the number of terms in the expansion drastically increases as the order of moments or cumulants becomes higher . to avoid it , another relation , @xmath42_{x } & = & \langle x_i \cdots x_\ell \rangle_{x } \nonumber \\ & & - \sum_{{\cal p}(\xi,2 ) } [ x_i \cdots x_j ] _ { x } \langle x_k \cdots x_m \rangle_{x } , \label{eq : mixexpand}\end{aligned}\ ] ] can be used.@xcite here , @xmath43 is the number of @xmath44-variables in the bracket in the l.h.s .
then , the summation denoted by @xmath45 is taken over every partition of @xmath43 elements into two groups on the condition that one of the variables , for example @xmath46 , must always be included in @xmath47_{x}$ ] .
in addition , each bracket includes at least one @xmath44-variable . in this paper , we call this formula `` the mixed expansion pivoting on @xmath46 '' . for example , the mixed expansion of @xmath48_x$ ] pivoting on @xmath49 is written as @xmath50_x & = & \langle x_1 x_2 x_3 \rangle_{x } - [ x_1 x_2 ] _ { x } \langle x_3 \rangle_{x } - [ x_1 x_3 ] _ { x } \langle x_2 \rangle_{x } \\ & & - [ x_1]_{x } \langle x_2 x_3\rangle_{x } .\end{aligned}\ ] ] as a reference system , we take a non - interacting system such that @xmath51 for every @xmath7 pair . its partition function is written as @xmath52 , and the average in the non - interacting system is denoted by @xmath53 where @xmath28 is an arbitrary operator .
this average @xmath54 plays a role of @xmath55 in the previous subsection .
there are a number of ways to define moments and cumulants of quantum variables because of their noncommutativity.@xcite one of the ways is defining a moment by a symmetrized product , @xmath56 where @xmath57 represents a permutation of the indeces .
note that the ordering of exchange operators @xmath9 is unimportant in @xmath58 .
hereafter , we use @xmath59 rather than @xmath58 if they are obviously equivalent to each other , _ e.g. _ , @xmath60 ; further examples are commented in appendix [ appn : sym ] . the partition function , eq .
( [ eq : parfunc ] ) , is rewritten as @xmath61 this averaged boltzmann factor works as the generating function of the symmetrized moments mentioned above , namely , with @xmath62 , @xmath63 the corresponding cumulants are denoted by @xmath47_{\rm s}$ ] , and given by @xmath64_{\rm s}= \left .
\frac{\partial}{\partial \lambda_{i , j } } \cdots \frac{\partial}{\partial \lambda_{k , l } }
\ln \langle e^{- \beta { \cal h } } \rangle_{0 } \right|_{\lambda=0}. \label{eq : symcumdef}\ ] ] these moments and cumulants have the properties explained in the previous subsection and appendix [ appn : fdclvbl ] .
the free energy is rewritten as @xmath65_{\rm s } , \label{eq : hmcs0}\ ] ] @xmath66_{\rm s } & = & \sum_{(i_1,j_1 ) } \cdots \sum_{(i_{m},j_{m } ) } j_{i_1,j_1}\cdots j_{i_{m},j_{m } } \nonumber \\ & & \qquad\qquad\qquad \times \left [ p_{i_1,j_1 } \cdots p_{i_{m},j_{m } } \right]_{\rm s}. \label{eq : hmcs}\end{aligned}\ ] ] many terms in the summation of eq .
( [ eq : hmcs ] ) are equivalent because the ordering of operators in @xmath67_{\rm s}$ ] is unimportant .
that is , in the case of the cumulant @xmath68_{\rm s}$ ] , then , @xmath69 terms are equivalent . from eqs .
( [ eq : hmcs0 ] ) and ( [ eq : hmcs ] ) , we can derive @xmath70_{\rm s } \nonumber \\ & & \quad= \sum_{m=0}^\infty \frac{(-\beta)^m}{m ! } \left [ p_{i , j}{{\cal h}}^m \right]_{\rm s}. \label{eq : wheikin}\end{aligned}\ ] ] even if @xmath51 in the original hamiltonian ( _ e.g. _ , for non - nearest neighbors in the case of nearest - neighbor interaction ) , one can start from the general formulation written above .
after deriving all the formulae , one can put @xmath71 or 0 according to the original hamiltonian .
let us consider cumulants for @xmath72 instead of @xmath9 .
see the definition of cumulants , eq .
( [ eq : symcumdef ] ) .
if one replaces @xmath9 by @xmath72 , the generating function of cumulants changes only by @xmath73 , and thus only the first - order cumulants are affected by this replacement . in other words , @xmath74_{\rm s } = \langle p_{i , j}\rangle_0 = 1/n \label{eq:1stcump}\ ] ] differs from @xmath75_{\rm s}= \langle { p}_{i , j}\rangle_0 = 0 . \label{eq:1stcumxx}\ ] ] on the other hand , the higher order cumulants are equal .
namely , @xmath76_{\rm s}= [ p_{i_1,j_1 } p_{i_2,j_2}\cdots p_{i_\ell , j_\ell}]_{\rm s } \label{eq : cumequal}\ ] ] when @xmath67_{\rm s}$ ] has two or more @xmath72 operators . for practical calculation
, we mainly use @xmath9 for convenience . in some cases
, however , the property @xmath77 helps simplification .
hereafter , we use @xmath9 and @xmath72 interchangeably with keeping in mind eqs .
( [ eq:1stcump])-([eq : cumequal ] ) .
let us introduce a diagrammatic representation of the moments and the cumulants .
although we mainly use those diagrams for later calculation , let us define them here in advance because they are convenient also for simple explanation in this review .
figure [ fig : bonddia1a ] shows examples of the diagrams for @xmath78 and @xmath79_{\rm s}$ ] .
we call them bond - diagrams .
each dot represents a site .
a segment connecting sites @xmath5 and @xmath19 represents @xmath9 , and we call it a bond .
note that a product of two equivalent operators appearing in @xmath58 [ _ e.g. _ , @xmath80 in the r.h.s@xmath81 of fig .
[ fig : bonddia1a](a ) ] can not be reduced to the identity because these are _ not _ always next to each other in the symmetrization .
figure [ fig : bonddia1a](a ) is a _ moment _ bond - diagram , and ( b ) is a _ cumulant _ bond - diagram .
a cumulant is equal to zero if the variables in @xmath67_{\rm s}$ ] can be partitioned into two groups which are independent of each other on averaging @xmath54 .
for example , @xmath82_{\rm s}=0 $ ] because the trace of sites 1,2 is taken independently of that of sites 3,4 in the non - interacting system .
in other words , a cumulant is equal to zero if the bond - diagram is not _ connected _ as shown in fig .
[ fig : bonddia1b](a ) . here , a solid rectangle in fig .
[ fig : bonddia1b ] represents arbitrary bonds .
therefore , as for cumulants , we consider only connected diagrams hereafter .
figure [ fig : bonddia1b](b ) is a first - order cumulant . in a higher - order cumulant bond - diagram ,
more than one bonds must meet at each dot . if not , it is equal to zero as shown in fig .
[ fig : bonddia1b](c ) and ( d ) . furthermore , a cumulant bond - diagram is equal to zero if two parts in it are connected by only one bond as shown in fig .
[ fig : bonddia1b](e ) .
we prove these properties in appendix [ app : addprf ] .
the finite cluster method ( fcm ) is a standard method for the high - temperature expansion.@xcite it is also used for perturbative expansions starting from exactly solvable models.@xcite in order to explain it in detail , we give an explicit example using the one - dimensional nearest - neighbor interacting system with a periodic boundary condition , namely , @xmath83 , @xmath84 , where the site @xmath85 is equivalent to site 1 . in this case , the procedure of the fcm is reduced to a very simple form . in the fcm ,
series coefficients in the thermodynamic limit are exactly obtained by summing contributions from finite - size clusters .
the clusters are subsets of the infinite lattice , and we need only connected clusters because cumulant bond - diagrams must be `` connected '' .
namely , the one dimensional system requires only open chains .
the cluster of size @xmath86 is defined as @xmath87 where we have chosen to use @xmath88 rather than @xmath9 in order to make use of @xmath89_{\rm s}=0 $ ] . the expectation value in this cluster is given by @xmath90_{\rm s}.\end{aligned}\ ] ] we assume @xmath91 hereafter .
note that @xmath92 when @xmath93 or @xmath94 because of a property shown in fig .
[ fig : bonddia1b](c ) and ( d ) .
since a finite cluster is used instead of the original system , @xmath95 contains only a subset of terms appearing in @xmath96 .
the important point is that @xmath97 includes contributions from clusters smaller than @xmath86 , and thus simple summation of @xmath97 over @xmath86 yields multiple counting of those terms . to avoid this
, the fcm requires subtraction of contributions from the smaller clusters .
we define a _ net _ contribution from the @xmath86-site cluster , @xmath98 where the summation @xmath99 excludes @xmath100 . here , @xmath101
( @xmath102 ) represents the number of removed sites from the side of 1 ( @xmath86 ) . when @xmath103 and @xmath104 , this expression is reduced to @xmath105
let @xmath106 denote the total net contribution to @xmath107 from clusters smaller than or equal to @xmath86 , @xmath108 here , @xmath106 does not depend on @xmath5 because of the translational symmetry .
note that @xmath109 .
when @xmath86 is finite , the series for @xmath106 is correct up to a finite order . to think about the `` finite order '' , let us take a cumulant bond - diagram appearing in the @xmath110-th order .
the diagram has a bond @xmath111 and @xmath110 bonds from the hamiltonian . in order to give a nonzero contribution
, there must be at least one bond per nearest - neighbor pair between @xmath5 and @xmath19 .
furthermore , nearest - neighbor pairs not between @xmath5 and @xmath112 less than @xmath5 or greater than @xmath112 must have at least two bonds per pair .
therefore , there is a contribution from @xmath113-site cluster when @xmath114 .
that is , @xmath106 is correct up to @xmath115 $ ] .
the above is the usual method of the fcm .
however , in one dimension , these equations can be reduced to simpler forms as shown in appendix [ app : freeene ] .
let us think about calculating the fourier transform of the correlation function , @xmath116 series of @xmath117 up to @xmath118 $ ] requires series of each @xmath119 up to @xmath118 $ ] . here , the important point is that the series coefficients of @xmath119 up to @xmath120 $ ] are equal to zero as explained in fig .
[ fig : bonddia1b](c),(d ) and in sec .
[ sec : exfcm ] .
hence , we need @xmath119 only for @xmath121 . in addition , that property provides us a weak point of the fcm : when @xmath122 is fixed , large @xmath44 requires a larger cluster than small @xmath44 does .
thus , the most time - consuming part is calculation in the largest cluster , @xmath123 , namely , calculation of tr@xmath124 .
however , what we need for @xmath123 is only the @xmath122-th order coefficient because we already know that the lower orders are equal to zero .
therefore , if only the fcm is used , the most time - consuming calculation yields very little information .
our goal in this paper to formulate a new method is as follows .
if the contributions from large clusters can be calculated by another method , the fcm can be used only for smaller clusters.@xcite accordingly , one can calculate up to high orders . in ref .
, we have calculated @xmath117 up to @xmath125 $ ] .
however , we have used the fcm only for @xmath126 .
the corrections to the results have been calculated by the method explained in this section .
we calculate traces of products of exchange operators by decomposing every permutation into a product of independent cyclic permutations@xcite as explained in the following , ( an explicit example is given in appendix [ sec : xtrace ] ) .
let us consider a trace of @xmath127 , with @xmath128 denoting the trace in the @xmath86-site system , @xmath129 where @xmath130 refers to the color at position @xmath5 after the permutation @xmath131 .
the summation of @xmath132 makes a contribution only when @xmath133 for every @xmath5 .
consider using this relation successively starting from @xmath5 .
that is , @xmath132 is equal to @xmath130 , and then @xmath130 is equal to @xmath134 , , one can repeat this procedure until coming back to @xmath132 at a certain power of @xmath131 , namely , @xmath135 . in other words ,
all the variables whose subscript belong to one cyclic permutation in @xmath131 have to be equal .
since any permutation can be decomposed into a product of independent cyclic permutations , the number of independent variables of the summation is the number @xmath136 of cyclic permutations of @xmath131 .
therefore the trace is given by , @xmath137 , and accordingly , @xmath138 in fact , this method allows us to calculate tr@xmath139 at the same time as tr@xmath140 for an arbitrary @xmath7 pair.@xcite let us remember the permutation @xmath131 is decomposed into cycles in calculating @xmath141 . when the cycle which @xmath5 belongs to is different from what @xmath19 belongs to , the next operation @xmath9 unites the two cycles into one
. then , the number of cycles decreases by one , namely , @xmath142 . on the contrary , if @xmath5 and @xmath19 belong to one cycle of @xmath131 , then @xmath9 breaks the cycle into two .
then , the number of cycles increases by one , namely , @xmath143 . to represent @xmath144
, we introduce an unsymmetrized version of a bond - diagram .
figure [ fig : defdia ] shows its examples . since this diagram is equivalent to an `` amida lottery '' , used in japan to peacefully decide how to distribute a fixed number of objects among an equal number of people , here we call it an amida - diagram .
each vertical line represents a site , and a horizontal line between site @xmath5 and @xmath19 represents @xmath9 .
the order of horizontal lines from bottom to top should coincide with the order of @xmath9 in @xmath54 from right to left .
we bracket an amida - diagram by @xmath145 in order to distinguish it from a cumulant version of the amida - diagram introduced later in sec .
[ sec : newsh ] .
since expectation values are not changed by transpositions of neighboring commutative operators , the ordering of @xmath146 and @xmath147 is irrelevant to the results . on the other hand ,
two exchange operators sharing one of the site - indeces , such as @xmath146 and @xmath148 , are not commutative , and in general the ordering of those bonds affects the result . for example , the value of fig . [ fig : defdia](a ) is different from that of ( b ) .
however , in some cases , even the order of such `` site - sharing '' bonds does not affect the result .
let us think about the value of an amida - diagram in fig .
[ fig : property](a ) by counting cycles .
the `` initial '' state is the identity permutation , which has @xmath86 cycles .
then , every time @xmath9 is applied , it unites two cycles into one .
after all the @xmath9 operators are applied , there is only one cycle , which includes all the @xmath86 sites .
hence the expectation value is @xmath149 , and it does not depend on the ordering of those @xmath9 operators . therefore , after the symmetrization , the moment bond - diagram fig . [
fig : property](b ) also gives @xmath149 .
let us think about adding @xmath150 to fig .
[ fig : property](a ) on the top .
since sites 1 and @xmath86 belong to the same cycle , the operation of @xmath150 cuts the cycle .
namely , the permutation has two cycles and the diagram gives @xmath151 .
again , after the symmetrization , the corresponding moment bond - diagram also gives @xmath151 . in fig .
[ fig : bonddia2](a ) , we calculate a cumulant using the values derived above .
this is the lowest - order nonzero cumulant in the @xmath152 series .
we use the mixed expansion , eq .
( [ eq : mixexpand ] ) , pivoting on @xmath150 . the expression of the expansion is very simple because @xmath153_{\rm s}$ ] is the one and only nonzero cumulant which includes @xmath150 .
next , let us consider adding one more bond , @xmath146 . the mixed expansion pivoting on @xmath150 is shown in fig .
[ fig : bonddia2](b ) .
the coefficient 2 of the last term is due to the two ways of choosing which one of the two @xmath146 is put in @xmath67_{\rm s}$ ] .
the first term of the r.h.s@xmath81 is calculated as shown in fig .
[ fig : extrabond ] .
the symmetrization is equivalent to averaging over all the possible configurations of bonds .
in fact , however , the value of each amida - diagram is determined only by relative configurations of two @xmath146 and one @xmath148 .
there are only two patterns , namely , fig .
[ fig : extrabond](b ) and ( c ) .
their r.h.s@xmath81 are obtained by simplification @xmath154 and @xmath155 . here
, an `` @xmath156-shape '' bond in fig . [ fig : extrabond](c ) represents that the bond is not connected to the site at `` @xmath156 '' .
we already know how to calculate the r.h.s@xmath81 of fig .
[ fig : extrabond](b ) and ( c ) .
we also know that configuration of @xmath157 ( @xmath158 ) does not affect the value of the amida - diagram .
that is , the value of the amida - diagram is determined only by the position of @xmath148 . in order to obtain the l.h.s@xmath81 of fig .
[ fig : extrabond](a ) , we have to count how often each of the configurations fig .
[ fig : extrabond](b ) and ( c ) appears in the symmetrization .
for example , the prefactor 1/3 in the last term of fig .
[ fig : extrabond](a ) is obtained by @xmath159 it can be considered like this : the vertical line of site 2 is partitioned into three regions by the two @xmath146 .
the probability of finding @xmath148 in the intermediate region is 1/3 .
more explicitly , it can be given also by integrals , @xmath160 where @xmath161 , @xmath162 , @xmath163 represent positions of lower @xmath146 , upper @xmath146 , @xmath148 , respectively .
here , @xmath164 is the bottom , @xmath165 is the top . on the other hand , the `` probability '' of the other configuration fig .
[ fig : extrabond](b ) is 2/3 .
consequently , the symmetrized value is calculated as shown in fig .
[ fig : extrabond](a ) .
this formula is true only when @xmath166 .
if @xmath167 , we can not take an average over position of @xmath148 , and fig .
[ fig : extrabond](c ) does not appear .
examples of higher order results are shown in fig .
[ fig : bondcums ] .
we can find simple properties in these cumulants . 1 .
a cumulant is reduced by a factor @xmath168 as @xmath86 increases by one .
2 . a double bond in fig .
[ fig : bondcums](a ) yields a factor of @xmath169 , two distant double bonds in fig .
[ fig : bondcums](d ) yield square of this factor .
in fact , we can make a more general statement : when multiple bonds are distant , we can simply make a product of contributions from each of multiple bonds .
we more rigorously derive property ( i ) in the next subsection and property ( ii ) in the subsection after the next .
after that , we formulate the new method using those properties . first , we consider a contraction of an amida - diagram . in an amida - diagram ,
let us focus on a part that includes two sites as shown in fig .
[ fig : amidacontract ] .
because of a property of a trace , the ordering of operators can be cyclically rotated . in other words ,
amida - diagrams have a periodic boundary condition .
hence we can put a bond @xmath170 at the bottom of the amida - diagram without loss of generality .
let us compare the l.h.s@xmath81 and the r.h.s@xmath81 at each position of bonds . in the l.h.s .
, the bottommost bond @xmath170 unites two cycles at sites @xmath19 and @xmath171 of the initial state , and makes a cycle of the two sites . at this stage
, nothing happens to the r.h.s . , and there is a cycle at @xmath19 .
then , the rest of the operations of the l.h.s@xmath81 and the r.h.s@xmath81 are the same . that is , @xmath9 unites this cycle and another cycle ; when @xmath172 in the l.h.s@xmath81 unites cycles or breaks a cycle , @xmath173 in the r.h.s@xmath81 does the same thing .
therefore , the number of cycles of the l.h.s@xmath81 is equal to that of the r.h.s . as for the denominator of the trace
, the l.h.s@xmath81 has one more site than the r.h.s .
hence , in total , the r.h.s@xmath81 needs a factor @xmath168 .
the logic above is true even if site @xmath5 is the same as site @xmath174 .
the symmetrization of all the operators except @xmath170 is equivalent to the symmetrization of all the operators as explained in appendix [ appn : sym ] .
hence , in all the terms in the symmetrization , one can put @xmath170 at the bottom of amida - diagrams and apply the contraction rule above .
therefore , we can obtain a contraction property also for a moment bond - diagram as shown in fig .
[ fig : contractm ] . finally , we prove that a cumulant bond - diagram has the same contraction property as the corresponding moment bond - diagram does .
we use a mathematical induction for the proof . 1 . the simplest nonzero cumulant , the l.h.s@xmath81 of fig . [ fig : bonddia2](a ) , has this contraction property .
2 . let us think about a certain cumulant bond - diagram .
we assume that `` the lower - order cumulant bond - diagrams have this contraction property . ''
then , we use the mixed expansion , eq .
( [ eq : mixexpand ] ) , pivoting on one of the bonds as shown fig .
[ fig : contract ] .
we use this assumption for the third term .
the property of moments can be applied to the first and second terms .
because of ( i ) and ( ii ) , the contraction property is true for any cumulant bond diagram that satisfies the condition .
we call this relation the theorem i. some of the cumulants can be decomposed into a `` background '' and local quantities .
we give an example in fig .
[ fig : bonddia3x ] . as already appeared several times , an intersite that has only one bond plays a special role in this paper .
we call it a single bond here . on the other hand , an intersite that has more than one bond
is called a multiple bond . in this subsection , we prove theorem ii shown in fig . [
fig : bonddia3 ] . here , a solid rectangle represents arbitrary bonds . a dotted line stands for a sequence of single bonds .
the bond - diagram in the r.h.s@xmath81 of ( a ) has the same length as that of the l.h.s . ; the bond - diagram in a denominator of ( b ) has the same length as that of the numerator . this decomposition is possible even if the bond - diagram has more than two blocks of multiple bonds , on condition that the number of single bonds should be greater than that of blocks .
in other words , the theorem ii is described as follows : a bond - diagram can be rewritten by the product of a background and local quantities .
the background is obtained by replacing each multiple bond with a single bond .
one of the local quantities is obtained by replacing each multiple bond with a single bond except those in that block , and dividing it by the background . for the proof , we introduce some notation . it is schematically shown in fig .
[ fig : bonddia4 ] . a diagram is denoted by @xmath175 . here , solid rectangle @xmath176 ( @xmath177 ) represents `` rectangle i ( ii ) and a sequence of single bonds '' in fig .
[ fig : bonddia3 ] unitedly .
we define @xmath178 ( @xmath179 ) as a diagram such that each multiple bond in @xmath177 ( @xmath176 ) of @xmath175 is replaced with a single bond .
we define the `` background '' @xmath180 as a cumulant bond - diagram such that each multiple bond in @xmath176 and @xmath177 of @xmath175 is replaced with a single bond .
hence , there is a relation , @xmath181 .
let us consider the cumulant of @xmath182 . as is also shown in fig .
[ fig : bonddia4b ] , the mixed expansion pivoting on @xmath9 is written as , @xmath183_{\rm s } = c_1(g ) + c_2(g ) , \label{eq : c1plusc2}\ ] ] @xmath184_{\rm s } \big\langle \prod_{k , l } p_{k , l}\big\rangle_{\rm s},\ ] ] @xmath185_{\rm s } \big\langle \prod_{k'',l '' } p_{k'',l''}\big\rangle_{\rm s}.\ ] ] that is , the second term of @xmath186 is decomposition into @xmath9 and the rest .
the other decompositions are included in @xmath187 . in @xmath187 ,
each cumulant includes more than one exchange operators , and thus , to give a nonzero cumulant , each site index has to appear at least twice as explained in fig .
[ fig : bonddia1b](c ) and ( d ) . in other words ,
each cumulant bond - diagram in @xmath187 has to have a loop . in order to prove the theorem ii
, we first prove a lemma , @xmath188 in the following .
first , we give an explicit example of @xmath186 using the cumulant in fig . [
fig : bonddia2](b ) . in sec .
[ sec : amida - diagrams ] , we have symmetrized each of the terms in the mixed expansion individually . here
, however , we postpone the symmetrization procedure .
that is , as shown in fig .
[ fig : upsilon0 ] , we first sum amida - diagrams of the terms in @xmath186 with a certain configuration of bonds .
after that , we symmetrize by averaging over all possible configuration of bonds .
we carry out the symmetrization of all the operators except @xmath150 this is equivalent to the symmetrization of all the operators as shown in appendix [ appn : sym ] .
let us consider fig .
[ fig : upsilon0](a ) . in the amida - diagram in the second term
, sites 1 and @xmath86 belong to different cycles .
hence , when one more bond @xmath150 is added ( to make the amida - diagram of the first term ) , the two cycles are united into one .
thus , the ratio of these amida - diagrams is @xmath168 , which is equal to @xmath153_{\rm s}$ ] in the second term .
accordingly , these terms cancel out each other .
therefore , there is a nonzero contribution to @xmath186 only when two sites 1 and @xmath86 belong to one cycle in the `` amida - diagram without @xmath150 '' in the second term of @xmath186 .
hereafter , such a configuration is called a @xmath186-finite configuration .
figure [ fig : upsilon0](b ) is a @xmath186-finite configuration .
its contribution to @xmath189 is determined by `` deviation from the background '' .
there is an `` extra '' bond compared to the background , and here it increases the number of cycles by one to make a factor @xmath0 . on the other hand , the `` probability '' for this configuration to occur
is 1/3 as shown in eqs .
( [ eq : onethird0 ] ) and ( [ eq : onethird ] )
. therefore we obtain @xmath190 .
the example above is a simple case . in general , `` the probability of the configuration '' times `` the deviation from the background '' is summed over all the @xmath186-finite configurations , namely , @xmath191 let us consider a diagram @xmath175 shown in fig .
[ fig : bonddia4 ] .
then , in order to put sites 1 and @xmath86 in one cycle in the `` amida - diagram without @xmath150 '' of @xmath175 , each of @xmath178 and @xmath179 must also have a @xmath186-finite configuration . then
, the deviation from the background as a whole is given by ( deviation in @xmath176)@xmath192(deviation in @xmath177 ) .
in fact , `` probability '' is also a product of those of @xmath176 and @xmath177 as shown below .
for example , the probability of configuration in fig .
[ fig : amidax2](a ) is written as @xmath193 the first three integrals are independent of the last three integrals , and thus the probability is written by the product of two parts .
next , let us think about fig .
[ fig : amidax2](b ) .
if we naively use @xmath194 , the expression of the integrals is complicated .
however , if we utilize the periodic boundary condition of @xmath194 and shift the integration variables as @xmath195 for @xmath196 , we obtain , @xmath197 this expression is equivalent to eq .
( [ eq : indepint ] ) .
hence the probability is written by the product of two parts again . the important point in deriving eq .
( [ eq : indepint2 ] ) is existence of @xmath198 .
suppose the diagram does not have @xmath198 , and has @xmath199 instead of @xmath200 .
the uppermost bond @xmath199 is always at @xmath165 .
if the integration variables are shifted , @xmath201 appears in lower or upper bounds of the integrals .
that is , the integrals are not written by a product of two parts any more in that case .
the shift of the integration variables mentioned above can generally be applied to any blocks @xmath176 and @xmath177 .
hence , the probability in eq .
( [ eq : non0confc1 ] ) is also a product of those of @xmath176 and @xmath177 .
the summation over @xmath186-finite configuration is double summation over @xmath186-finite configuration of @xmath176 and over @xmath186-finite configuration of @xmath177 .
therefore , @xmath202 by using @xmath181 , we obtain the lemma , eq .
( [ eq : lemma ] ) .
note that the shift of the integration variables mentioned above can be applied even if a diagram has more than two blocks of multiple bonds .
a block and a single bond is regarded as one group whose members have a common shift of the integration variables .
this is the reason for the restriction of the theorem ii ; the number of single bonds should be greater than that of blocks . in order to focus on the `` multiple bond '' region ,
we introduce another notation for moments and cumulants .
let us think about @xmath178 .
a set of bonds , @xmath176 , is partitioned into two subsets in calculating @xmath203 .
let @xmath204 denote a subset of @xmath176 .
then , @xmath205 denotes the cumulant in which @xmath176 is replaced by @xmath204 . in choosing @xmath204
, at least one bond per intersite must be chosen to make @xmath205 nonzero .
a moment for bonds @xmath206 is denoted by @xmath207 . then , eq . ( [ eq : c1plusc2 ] ) for @xmath178 can be rewritten as @xmath208 the l.h.s@xmath81 is equal to the excluded term @xmath209 in the summation in the r.h.s@xmath81 because @xmath210 . therefore , we obtain @xmath211 the corresponding equation for graph @xmath175 is @xmath212 since @xmath176 is not connected to @xmath177 , @xmath213 on the other hand , according to the lemma and eq .
( [ eq : therefore ] ) , @xmath214 here , we use a mathematical induction again for the proof . 1 .
the theorem ii is true when there are two extra bonds on the background as explicitly calculated in fig .
[ fig : bondcums](d ) .
we assume that the theorem ii is true when the number of extra bonds on the background is less than that of @xmath175 .
according to the assumption , @xmath215 combine the r.h.s@xmath81 of both eqs .
( [ eq : upsilon1 - 1 ] ) and ( [ eq : upsilon1 - 2 ] ) , use eqs .
( [ eq : momentdecompose ] ) and ( [ eq : assumption ] ) . then , all the terms except for @xmath216 cancel out and we obtain @xmath217 according to ( i ) and ( ii ) , the theorem ii is true for any @xmath175 that satisfies the conditions .
the proof mentioned above can be applied even if block @xmath176 or @xmath177 includes single bonds .
then , the theorem ii for a diagram with two blocks is repeatedly used to prove the theorem ii for a diagram with more than two blocks .
therefore , also a diagram with more than two blocks has the `` detachable '' property as mentioned first .
now that the theorem ii is proved , the contraction can be written in a more general form .
although the theorem i is for a sequence of three single bonds , sequences of two single bonds also can be contracted as shown in fig .
[ fig : ip ] as theorem i , on condition that each block must be separated by at least one single bond and two of the blocks must be separated by a sequence of at least two single bonds . the theorem i is easily proved using fig .
[ fig : bonddia3](a ) ; contract @xmath180 in the r.h.s@xmath81 and reversely use the same relation to the contracted diagram .
apart from the restriction , the theorem ii is equivalent to a property of moments a disconnected moment is equal to a product of connected moments . in order to effectively use the theorem
ii , we define _ quasi - moments _ and corresponding _ quasi - cumulants _ below . our definition of a quasi - moment is a factor times a `` local quantity '' in the theorem ii .
although another definition can be the local quantity itself , it is less convenient for our purpose as discussed in appendix [ sec : otherdef ] .
let us think about the one - dimensional system again .
we define the generating function of quasi - moments as @xmath218_{\rm s}}. \label{eq : defgqm}\end{aligned}\ ] ] only up to a finite order , the theorem ii can be applied and @xmath219 works as a generating function .
the larger @xmath220 is , the higher order the theorem ii is valid up to . here
, we assume that @xmath220 is large enough , and forget about this restriction for the time being .
it will be commented on later .
then , the quasi - moments are defined as @xmath221 the zeroth - order quasi - moment satisfies @xmath222 . some examples of quasi - moments are shown in fig.[fig : figqua ] . more explicitly , the quasi - moments are obtained by @xmath223_{\rm s } } { ( m_1 + 1)(m_2 + 1)\cdots(m_{k-1}+1 ) \;\ ; \left [ p_{1,2 } p_{2,3 } \cdots p_{k-1,k } p_{k , k+1 } p_{1,k+1 } \right]_{\rm s } } , \label{eq : qmobtain}\end{aligned}\ ] ] where we have assumed that all the bonds are on the `` background '' .
we need a special care when @xmath172 is outside the background , namely , @xmath224 or @xmath225 in the case of @xmath226 .
such a operator is explicitly denoted by @xmath227 .
then , we can more generally write as , @xmath228_{\rm s } } { ( m_i+1 ) \cdots(m_{k-1}+1)\;\ ; \left [ p_{i , i+1 } \cdots p_{k-1,k } p_{k , k+1 } p_{i , k+1 } \right]_{\rm s}}. \label{eq : qmobtain2}\ ] ] next , using quasi - moment defined above , the generating function of quasi - cumulants is defined as , @xmath229 then , the quasi - cumulants are defined by , @xmath230_{\rm q}:= \left .
\frac{\partial}{\partial \lambda_{k , l } } \cdots \frac{\partial}{\partial \lambda_{y , z } } { \cal g}_{\rm qc}^{(i , j ) } \right|_{\lambda=0}. \label{eq : defqcum}\ ] ] according to the theorem ii , when none of @xmath171 or @xmath174 is equal to @xmath231 or @xmath232 , @xmath233 \right\rangle_{\rm q } = \nonumber \\ & & \left\langle \exp\left[\sum_{k , l}\lambda_{k , l } p_{k , l}\right ] \right\rangle_{\rm q } \left\langle \exp\left[\sum_{y , z}\lambda_{y , z } p_{y , z}\right ] \right\rangle_{\rm q } , \label{eq : prodqm}\end{aligned}\ ] ] and thus , as discussed in appendix [ appn : fdclvbl ] , @xmath234_{\rm q } = 0.\ ] ] therefore , only `` connected '' quasi - cumulants remain nonzero .
if only the quasi - moments are used , we have to classify connected and disconnected quasi - moments .
however , if the quasi - cumulants are used , we need only _ connected _ quasi - cumulants and can avoid the complicated classification .
suppose we set @xmath235 for every @xmath236 pair .
then , @xmath237 @xmath238 is a function of @xmath44 and @xmath239 .
the quasi - cumulants have a translational symmetry in the background .
for example , @xmath240_{\rm q}=[p_{i+1,i+2}^{m}]_{\rm q}=\cdots= [ p_{i+x-1,i+x}^{m}]_{\rm q}$ ] ; and thus , these `` length-1 '' quasi - cumulants can be summed up to @xmath241_{\rm q}$ ] in @xmath237 .
if we in the same way simplify @xmath237 , length-2 quasi - cumulants are multiplied by @xmath242 , length-3 quasi - cumulants are multiplied by @xmath243 , and so on .
the quasi - cumulants including @xmath227 are located at both ends , and thus , they are multiplied by two .
namely , @xmath244 @xmath245_{\rm q } \nonumber \\ & & + ( x-1 ) \sum_{\stackrel{\scriptstyle 1\le m_1,m_2}{m_1+m_2\le m } } \frac{\lambda^{m_1+m_2}}{m_1!m_2 ! } [ p_{i , i+1}^{m_1}p_{i+1,i+2}^{m_2}]_{\rm q } \nonumber \\ & & + \quad \cdots\cdots\cdots \nonumber \\ & & + ( x - m+1 ) \lambda^m [ p_{i , i+1 } \cdots p_{i+m-1,i+m}]_{\rm q},\end{aligned}\ ] ] @xmath246_{\rm q } \nonumber \\ & & + 2 \sum_{\stackrel{\scriptstyle 2\le m_1 , 1\le m_2}{m_1+m_2\le m } } \frac{\lambda^{m_1+m_2}}{m_1!m_2 ! } [ ( p_{i-1,i}^{\rm ( o)})^{m_1 } p_{i , i+1}^{m_2}]_{\rm q } \nonumber \\ & & + 2 \sum_{\stackrel{\scriptstyle 2\le m_1 , m_2}{m_1+m_2\le m } } \frac{\lambda^{m_1+m_2}}{m_1!m_2 ! } [ ( p_{i-2,i-1}^{\rm ( o)})^{m_1 } ( p_{i-1,i}^{\rm ( o)})^{m_2}]_{\rm q } \nonumber \\ & & + \quad \cdots\cdots\cdots .
\label{eq : gqcxend2}\end{aligned}\ ] ] hence , this expression for arbitrary @xmath44 can be derived if we obtain all the quasi - cumulants up to @xmath110-th order , which are calculable in small systems . therefore , we can obtain @xmath247 up to @xmath248-th order for _ arbitrary _
@xmath44 by tracing back eqs .
( [ eq : gqcxend])-([eq : gqcxend2 ] ) , ( [ eq : gqcgqm ] ) and ( [ eq : defgqm ] ) .
we call this method the quasi - cumulant method ( qcm )
. the qcm has been done by a brute - force program of mathematica .
the quasi - cumulants can be calculated by eq .
( [ eq : defqcum ] ) , that is , differentiating the series of their generating function with respect to corresponding @xmath249 , where @xmath249 are coupling constants of the hamiltonian of a rather general form , @xmath250 .
we take the system as small as possible on condition that it be large enough to correctly calculate those quasi - cumulants .
that is , quasi - cumulants of length @xmath174 can be obtained in the system of length @xmath251 .
first , we calculate the series of @xmath252 to obtain @xmath226 up to @xmath110-th order . at each order of calculating @xmath253 ,
we exclude terms irrelevant to the derivative in order to save computational time and memory .
let us rewrite @xmath226 as @xmath254 then , eq .
( [ eq : mixexpand ] ) , the mixed expansion pivoting on one of @xmath255 can be used , @xmath256_{\rm q}=\langle{\cal h}^k\rangle_{\rm q } - \sum_{l=1}^{k-1}\frac{(k-1)!}{(l-1)!(k - l ) ! } [ { \cal h}^l ] _ { \rm q } \langle { \cal h}^{k - l } \rangle_{\rm q},\ ] ] @xmath257_{\rm q } + o(\beta^{m+1}).\ ] ] now we are ready for the derivatives .
the point is that we need at most @xmath258 sites here , while we can reconstruct the generating function of larger systems from the quasi - cumulants and thus we can obtain @xmath247 up to @xmath248-th order arbitrary @xmath44 . finally , we comment on the restriction of the qcm .
the qcm is based on the theorem ii .
let us consider diagrams shown in fig .
[ fig : notvalid ] .
we can not apply the theorem ii to these diagrams , and thus eq .
( [ eq : prodqm ] ) is not valid .
however , we can apply the theorem ii if the order is lower than these .
hence , we can use eqs .
( [ eq : gqcxend])-([eq : gqcxend2 ] ) , only for @xmath259 ( @xmath44 : even ) or @xmath260 ( @xmath44 : odd ) .
note that the qcm is useful for contributions from large clusters rather than those from small clusters .
hence , by combining the qcm with the fcm , we can calculate the high - temperature series coefficients up to a high order .
figure [ fig : combi ] shows which of the fcm and the qcm was used to obtain the series coefficients in ref . .
we calculated @xmath247 up to @xmath261-th order for @xmath262 by the qcm .
note that the qcm results are valid for arbitrary @xmath262 , including the @xmath263 limit as indicated in fig .
[ fig : combi ] .
the reason why we stopped the qcm at this order was mainly due to the fcm .
if the fcm can treat larger systems , the restriction of the theorem ii is less important , and the qcm can go further .
our fcm program was for general su(@xmath0 ) ; we expect that more terms can be obtained if the interest is restricted to su(2 ) in making the fcm program . in ref .
, the largest cluster we used for the fcm was composed of @xmath264 sites . as noted in the context of eq .
( [ eq : totcont ] ) , @xmath106 is correct only up to @xmath115 $ ] , and higher orders include contributions from the larger clusters .
hence , for high - order coefficients of @xmath265 , corrections to the fcm results must be calculated .
the qcm can be used also for this purpose because these corrections are a subset of the terms obtained by the qcm .
we put a fictitious coefficient on each quasi - cumulant located at both ends in eq .
( [ eq : gqcxend2 ] ) .
these fictitious coefficients carry information about the cluster - size of a cumulant , and we can distinguish terms needed for the corrections to the fcm result .
coefficients obtained by this procedure in ref . are indicated by `` fcm+qcm '' in fig .
[ fig : combi ] .
since these series coefficients are obtained as explicit functions of @xmath0 , the data of the series coefficients are too numerous to be fully listed in this paper .
hence , we list here only the results of the qcm , @xmath247 up to @xmath261-th order , and the other data will be provided on request .
the coefficients of the series @xmath266 are given in the following : @xmath267 @xmath268 @xmath269 & & \nonumber \\
\qquad ( x \ge 4),\end{aligned}\ ] ] @xmath270 \qquad ( x \ge 6),\end{aligned}\ ] ] @xmath271 \qquad ( x \ge 8),\end{aligned}\ ] ] @xmath272 \nonumber \\ & & \qquad ( x \ge 10),\end{aligned}\ ] ] @xmath273 \qquad ( x \ge 12).\end{aligned}\ ] ] even for @xmath274 ,
the qcm is valid for a number of series coefficients , and should coincide with the results of the fcm or the `` fcm+qcm '' .
we have used this property for checking whether the computer codes are correct .
our strategy for the specific heat in ref . was similar to that for the correlation functions .
that is , the most time - consuming part of the fcm is calculated by another method instead .
if only the fcm is used , the series for @xmath275 up to @xmath118 $ ] requires systems with @xmath276 ( @xmath122 : even ) or @xmath277 ( @xmath122 : odd ) . in ref . , we calculated @xmath275 up to @xmath278 $ ] .
however , the fcm was used only for @xmath279 ; contributions from the required largest cluster , namely , the lowest- and the second - lowest - order nonzero contribution from a cluster , was directly calculated by a new technique explained below .
in fact , we can use @xmath54 as a moment and define a cumulant@xcite without the symmetrization .
the generating functions of the moments and the cumulants are defined as @xmath280 @xmath281 respectively .
then , @xmath282 works as a moment , and the corresponding cumulant is define by @xmath283_0 : = \left .
\frac{\partial}{\partial \lambda_1 } \cdots \frac{\partial}{\partial \lambda_k } { \cal g}_{0\rm c}^{i_1,j_1,\cdots , i_k , j_k } \right|_{\lambda=0}.\end{aligned}\ ] ] here , the ordering of exchange operators in @xmath67_{0}$ ] is important . in other words ,
the result can be changed by exchanging positions of uncommutative operators .
note that simplification such as @xmath154 is _ not _ possible in @xmath67_0 $ ] while it is possible in @xmath59 . in expanding the cumulants using the moments ,
the ordering of operators in @xmath59 must be the same as that in @xmath67_0 $ ] , for example , @xmath284_0 = \langle p_{1,2 } p_{2,3 } \rangle_0 - \langle p_{1,2 } \rangle_0 \langle p_{2,3 } \rangle_0 $ ] .
this cumulant is represented by the cumulant version of an amida - diagram .
hereafter , we distinguish between a moment amida - diagram and a cumulant amida - diagram by bracketing the diagram by @xmath145 and @xmath67 $ ] , respectively .
the free energy is rewritten as @xmath65_{0},\ ] ] @xmath66_{0 } & = & \sum_{(i_1,j_1 ) } \cdots \sum_{(i_{m},j_{m } ) } j_{i_1,j_1}\cdots j_{i_{m},j_{m } } \nonumber \\ & & \qquad\qquad\qquad \times \left [ p_{i_1,j_1 } \cdots p_{i_{m},j_{m } } \right]_{0}.\end{aligned}\ ] ] these are equivalent to eqs .
( [ eq : hmcs0 ] ) and ( [ eq : hmcs ] ) except for the subscript `` 0 '' replacing `` s '' . here , however , the cumulants are generally dependent on the ordering of the operators while `` s '' makes many terms equivalent .
we can use general property of moments and cumulants again . if , @xmath285 \prod_{l } \exp\left[\lambda_l p_{i_l , j_l } \right ] \right\rangle_0 = \nonumber \\ & & \left\langle \prod_{k } \exp\left[\lambda_k p_{i_k , j_k } \right ] \right\rangle_0 \left\langle \prod_{l } \exp\left[\lambda_l p_{i_l , j_l } \right ] \right\rangle_0,\end{aligned}\ ] ] then , the derivative of its logarithm gives , @xmath286_0 = 0.\ ] ] when none of @xmath287 or @xmath288 is equal to @xmath289 or @xmath290 , this relation is obviously satisfied . furthermore , the cumulant is equal to zero also when an amida - diagram is separated into two diagrams by cutting only one point on a vertical line as shown in fig .
[ fig : cumzero ] . for example , the cumulant versions of the amida - diagrams in fig .
[ fig : defdia](a ) and ( b ) are equal to zero .
let us think about the moment amida - diagram , fig .
[ fig : cumzero](a ) , where bonds are located only in the painted regions .
bonds located left of @xmath43 are denoted by @xmath291 , and those located right of @xmath43 are denoted by @xmath292 .
the diagram can be written as @xmath293 by exchanging positions of commutative @xmath291 and @xmath292 .
let us imagine calculating each of @xmath294 and @xmath295 by counting cycles , and make the product of these two at the end .
this final operation unites the two cycles relevant to @xmath43 into one cycle because @xmath294 and @xmath295 are sharing only one site @xmath43 .
hence , the number of cycles in the l.h.s@xmath81 is less than that in the r.h.s@xmath81 `` by one '' .
however , the number of sites in the l.h.s@xmath81 is also less than that in the r.h.s@xmath81 `` by one '' . as a result ,
the l.h.s@xmath81 is equal to the r.h.s .
we can make the same argument to @xmath296 .
therefore , @xmath297_0=0 $ ] . here , we must be careful about the periodic boundary condition of amida - diagrams .
if a diagram satisfies the condition above by using the periodic boundary condition , the amida - diagram is also equal to zero .
we give an example in fig .
[ fig : defdia2 ] . at a sight , @xmath298_0 $ ]
does not satisfy the condition above .
however , it is equivalent to @xmath299_0 = 0 $ ] .
suppose there are two bonds per nearest - neighbor pair as shown in fig .
[ fig : amida1st1 ] .
it is a contribution from cluster - size @xmath86 for @xmath300 $ ] of @xmath275 , or equivalently , @xmath301 $ ] of the free energy . because of the property mentioned above
, many configurations of bonds make cumulant amida - diagrams equal to zero . in order to give a nonzero contribution , at each vertical line bonds at left and right have to appear alternately .
we can make such diagrams by repeating a enlarging procedure shown in fig .
[ fig : amida1st2 ] .
that is , there are two possibilities for next bonds .
in fact , the value of such a cumulant amida - diagram does not depend on those configurations as explained in appendix [ sec : proofspht ] .
hence , we can just multiply the value of the cumulants by the probability for such nonzero configurations .
we have calculated the probability by mathematica using integrals like eqs .
( [ eq : onethird ] ) and ( [ eq : indepint ] ) .
the values of the cumulants are shown in fig .
[ fig : amida1st1 ] .
namely , the cumulant is equal to @xmath302 when @xmath86 is an even number , and @xmath303 when @xmath86 is an odd number .
we can prove this by using the mixed expansion pivoting on one of @xmath146 shown in fig .
[ fig : amida1st4 ] .
the first term is the moment .
the second term is @xmath304 $ ] times the moment of the others .
the rest of the terms are obtained by separating the diagram at sites from 2 to @xmath305 .
in fact , only the last term survives and every term else cancels out one another .
when @xmath86 is an even number , the sum of the first , second and third terms , the sum of the fourth and fifth terms , , are equal to zero .
when @xmath86 is an odd number , the sum of the first and second terms , the sum of the third and fourth terms , , are equal to zero .
the details are explained in appendix [ sec : proofspht ] .
we can calculate also the next order by the same technique as above .
now , one of the nearest - neighbor pairs has three bonds .
the value of the cumulant amida - diagram does not depend on the position of this triple bond as shown in fig .
[ fig : amida2nd ] .
that is , the cumulant is equal to @xmath306 when @xmath86 is an even number , and equal to @xmath307 when @xmath86 is an odd number .
we have formulated a new method of high - temperature series expansion using the su(@xmath0 ) heisenberg model .
it is designed for efficiently calculating contributions from large clusters , which is actually the most time - consuming part of a standard method .
the net contributions from a cluster to high - temperature series start from a certain order .
we have focused on this property , and considered deviation from the lowest - order nonzero contribution .
although we have mainly used one - dimensional systems to explain the new method , many of the techniques can be used also in higher dimensions .
in contrast to one - dimensional systems , however , the calculation requires various clusters , and some of the clusters are unsuitable for our method .
figure [ fig : suitable ] shows an example .
the main advantage to use our method is that diagrams can be reduced to smaller diagrams .
however , we can not apply it to a cluster in fig .
[ fig : suitable ] because it has many sites at which more than two bonds meet , and chains connecting those sites are too short . nevertheless , the number of embeddings of such unsuitable clusters in the infinite lattice is small , and thus contributions from them are usually small .
in fact , the number of embeddings of open chains , the most preferable clusters for our method , is the largest .
hence , the dominant contributions can be calculated by our method .
we expect that similar approaches are possible also for other models , namely , for hamiltonians of lower symmetries .
the author is grateful to m. arikawa for indicating him the equivalence of an amida - lottery and a permutation .
he also thanks n. shannon , k. penc and y. kuramoto for useful comments on the manuscript .
this work was financially supported by visitors program of the mpi pks and the japan society for the promotion of science .
for example , using @xmath30 generators such as @xmath308 @xmath309 @xmath310 the exchange operator is rewritten as @xmath311 let us prove that each component contributes equally , namely , @xmath312 1 .
the hamiltonian is invariant under re - labeling of colors .
therefore , @xmath313 does not depend on @xmath20 or @xmath2 .
2 . @xmath314 for arbitrary @xmath20 , @xmath2 , because these @xmath315 , @xmath316 operators are @xmath44 , @xmath231 components of pauli matrix @xmath317 regarding only @xmath20 and @xmath2 .
we can apply the same argument also for @xmath232-component , and thus @xmath318 .
4 . by a direct calculation using @xmath319 for arbitrary @xmath20 , and @xmath320 for arbitrary @xmath321
, we can derive @xmath322 for arbitrary @xmath20 . from ( i)-(iv )
, we obtain eq .
( [ eq : equalcontr ] ) . using eq .
( [ eq : equalcontr ] ) , we can derive @xmath323 for @xmath324 .
in addition , for @xmath35 , since the average of @xmath325 does not depend on @xmath20 , @xmath326 then , using @xmath327 we obtain @xmath328
in this section , we review fundamental properties of moments and cumulants of classical variables @xmath39 . note that some other properties are written in sec .
[ sec : mcclassic ] . here , the expectation value with respect to the distribution of @xmath329 is denoted by @xmath330 .
the generating function of moments is defined as @xmath331 that is , the moments are derived by @xmath332 where the subscript @xmath333 represents that all the @xmath239-variables are set to zero . on the other hand ,
cumulants are denoted by @xmath67_x$ ] and defined as @xmath334_x = \left . \frac{\partial}{\partial\lambda_j } \frac{\partial}{\partial\lambda_k } \cdots \frac{\partial}{\partial\lambda_l } \ln \left\langle e^{\sum_i \lambda_i x_i } \right \rangle_x \right|_{\lambda=0}.\ ] ] namely , the generating function of the cumulants is written as @xmath335_{x}.\ ] ] a cumulant is equal to zero if the variables in @xmath67_x$ ] can be partitioned into two groups which are independent of each other in averaging @xmath55 .
suppose @xmath329 is independent of @xmath336 .
let us in advance set @xmath337 to zero if @xmath338 is not included in either @xmath329 or @xmath336 .
then , @xmath339 by taking the logarithm , we obtain @xmath340 that is , the generating function of the cumulants does not have any term which has a product of @xmath341 and @xmath342 .
thus , the corresponding cumulant is equal to zero
. there are relations between moments and cumulants .
moments can be expanded using cumulants , for example , @xmath343_{x } , \\
\langle x_1 x_2\rangle_x & = & [ x_1 x_2]_{x } + [ x_1 ] _ { x } [ x_2 ] _ { x } , \\
\langle x_1 x_2 x_3 \rangle_x & = & [ x_1 x_2 x_3 ] _ { x } + [ x_1 x_2 ] _ { x } [ x_3 ] _ { x } + [ x_1 x_3]_{x } [ x_2 ] _ { x } \\ & & + [ x_1 ] _ { x } [ x_2 x_3]_{x } + [ x_1]_{x } [ x_2]_{x } [ x_3 ] _ { x}.\end{aligned}\ ] ] these relations are generally written as @xmath344_{x } \cdots [ x_k \cdots x_m ] _ { x}. \label{eq : m2c}\ ] ]
these summations include every partition of @xmath44-variables . here , @xmath43 is the number of @xmath44-variables in the bracket in the l.h.s .
then , the summation denoted by @xmath345 is taken over every partition of @xmath43 elements into @xmath346 groups ( @xmath347 ) . in other words , @xmath346 is the number of brackets @xmath67_x$ ] in the r.h.s .
here , each @xmath67_x$ ] includes at least one @xmath44 .
conversely , cumulants can be expanded using moments , for example , @xmath348_{x } & = & \langle x_1 \rangle_x , \\ { } [ x_1 x_2]_{x } & = & \langle x_1 x_2\rangle_x -\langle x_1 \rangle_{x } \langle x_2 \rangle_{x } , \\ { } [ x_1 x_2 x_3 ] _ x & = & \langle x_1 x_2 x_3 \rangle_{x } - \langle x_1 x_2 \rangle_{x } \langle x_3 \rangle_{x } - \langle x_1 x_3\rangle_{x } \langle x_2 \rangle_{x } \\ & & - \langle x_1 \rangle_{x } \langle x_2 x_3\rangle_{x } + 2 \langle x_1\rangle_{x } \langle x_2\rangle_{x } \langle x_3 \rangle_{x}.\end{aligned}\ ] ] in general , eq .
( [ eq : m2c ] ) is inverted to @xmath42_{x } & = & \sum_\zeta \sum_{{\cal p}(\xi,\zeta ) } ( -1)^{\zeta-1 } ( \zeta-1 ) ! \nonumber \\ & & \quad \times \langle x_i\cdots x_j \rangle_{x } \cdots \langle x_k \cdots x_m \rangle_{x}. \label{eq : c2m}\end{aligned}\ ] ]
let us consider expanding @xmath349 .
it includes all the orderings of @xmath9 operators . in other words
, it is already symmetrized . therefore ,
@xmath350 accordingly , @xmath351 because of a a property of trace , one can cyclically rotate variables in the average . in each term of the symmetrization ,
let us put a certain operator , for example @xmath352 , at the leftmost . then
, @xmath171 terms give the same ordering .
namely , @xmath353 where @xmath354 is a permutation of @xmath355 elements .
the factor @xmath171 cancels out @xmath171 in @xmath356 in eq .
( [ eq : symmetrization ] ) .
that is , the symmetrization of all the operators except one is equivalent to the symmetrization of all the operators .
therefore , @xmath357 is also already symmetrized , and thus @xmath358
here , we prove properties of cumulants in fig . [ fig : bonddia1b ] . for a simplicity , we use the su(2 ) notation here . however
, for general su(@xmath0 ) , one can easily replace spin operators below by @xmath359 , @xmath360 , @xmath361 in appendix [ app : derive ] .
in order to consider fig . [
fig : bonddia1b](c ) and ( d ) .
let us expand the l.h.s@xmath81 of eq .
( [ eq : cumequal ] ) using moments , namely , in terms of eq . ( [ eq : c2 m ] ) . then , let us rewrite the @xmath88 operators using spin operators in terms of eq .
( [ eq : su2spexp ] ) .
if a site index @xmath5 appears only once in @xmath67 $ ] , every term in the expansion is equal to zero because the average at site @xmath5 is evaluated as @xmath362 consequently , the cumulant is equal to zero .
note that , in contrast to the ising model , a moment bond - diagram in which an odd number of bonds meet at a site can give nonzero value because a product of three operators can give nonzero average , for example , @xmath363 .
next , we prove fig . [
fig : bonddia1b](e ) .
after sec .
[ sec : prooflemma ] is explained , this property can be easily proved .
let us think about the mixed expansion pivoting on the bond in the center .
then , @xmath186 is equal to zero because this middle bond always unites two cycles .
furthermore , @xmath187 in eq .
( [ eq : c1plusc2 ] ) is equal to zero because every cumulant in it is equal to zero .
in one dimension , the equations in sec . [ sec : exfcm ] can be reduced to simpler forms as shown below . from eq .
( [ eq : excsum ] ) , we can derive @xmath364 in other words , eq .
( [ eq : excsum ] ) is inverted into eq .
( [ eq : relation ] ) using _ full perimeter lattice constants _ explained in ref . .
combining eqs .
( [ eq : totcont ] ) and ( [ eq : relation ] ) , contributions from small clusters are canceled out and we obtain a simple formula , @xmath365 there are formulae also for the free energy and we have used it for calculating the specific heat of the su(@xmath366 ) limit in ref . .
the free energy from the interaction part for @xmath367 is @xmath368 .
the _ net _ contribution from the @xmath86-site cluster is defined as , @xmath369 the total net contribution from clusters smaller than or equal to @xmath86 is defined as @xmath370 the simplified version of these relations also exists .
equation ( [ eq : netf ] ) is rewritten as@xcite @xmath371 finally , eq . ( [ eq : netftot ] ) is reduced to @xmath372 here , @xmath373 is correct up to @xmath374 $ ] . note that @xmath375 , where @xmath376 .
in fact , eq . ( [ eq : freeenedrv ] ) is reduced to @xmath377 in the @xmath378 limit .
let us consider a five - site system .
then , @xmath379 we show the calculation for @xmath380 below . here ,
@xmath9 exchanges @xmath5-th state and @xmath19-th state .
@xmath381 let us consider a condition to give a nonzero contribution . here
, @xmath382 requires @xmath383 , then @xmath384 requires @xmath385 , then @xmath386 requires @xmath387 .
these are reduced to @xmath388 .
in fact , this procedure is equivalent to finding a cyclic permutation @xmath389 . by the same argument ,
variables in another cyclic permutation @xmath390 also should be equal to each other , namely , @xmath391 .
hence , there are only two independent variables in the multiple summation above .
accordingly , we obtain @xmath392 in other words , permutation @xmath393 is composed of cycles @xmath394 , and then , the number of independent variables is given by the number of cycles , namely , two .
let us operate one more exchange operator to @xmath131 .
when @xmath146 is operated , both 1 and 2 belong to a cycle @xmath395 , and this operation breaks this cycle . that is , @xmath396 has three cycles ( 1)(24)(35 ) . in the case of @xmath397
, however , 1 belongs to ( 142 ) , and 3 belongs to ( 35 ) , and these cycles are united , namely , @xmath397 has only one cycle ( 15342 ) .
in fact , a quasi - moment can be defined without a prefactor in eqs .
( [ eq : qmobtain ] ) and ( [ eq : qmobtain2 ] ) . in this case ,
[ fig : bonddia3](b ) is equal to a quasi - moment .
the generating function for this quasi - moments is defined by a derivative as , @xmath398_{\rm s}}.\end{aligned}\ ] ] consequently , the generating function for the corresponding quasi - cumulants is written as , @xmath399 however , we have to integrate the quasi - moment generating function @xmath400 times to come back to the correlation function . hence , it is difficult to calculate general dependence on @xmath401 , and thus this definition is not suitable for our purpose .
this is the reason why we have adopted eq .
( [ eq : defgqm ] ) .
first , let us think about moment amida - diagrams . in order to give a nonzero contribution , a configuration at a nearest - neighbor pair
is topologically equivalent to one of the two types shown in fig .
[ fig : amida1st22 ] . a pair of @xmath402 can be simplified by exchanging subscripts @xmath44 and @xmath231 of operators between them . we can simplify left and right pairs of two bonds to give the r.h.s@xmath81 of fig .
[ fig : amida1st22 ] . in the parentheses , we schematically represent which sites are included in which cycle ; sites belonging to a cycle are connected by a line . in the moment versions of amida - diagrams in fig .
[ fig : amida1st1 ] , we simplify bonds at every other intersite , namely , @xmath146 , @xmath147 , .
the simplified diagrams are made of the pieces in fig .
[ fig : amida1st22 ] and special pieces for the ends as shown in fig .
[ fig : amida1st221 ] . when the end pieces are absent , any combination of the pieces in fig .
[ fig : amida1st22 ] gives two cycles .
when @xmath86 is an even number , the end pieces are simple vertical lines , and do not do anything to cycles .
hence , the diagram has two cycles and gives @xmath403 .
when @xmath86 is an odd number , one of the end pieces is not a simple vertical line as shown in fig .
[ fig : amida1st221](b ) , and it unites two cycles into one .
therefore , the diagram has only one cycle and gives @xmath404 .
the important point is that the amida - diagrams of odd @xmath86 are equal to those of @xmath405 .
then , we can prove fig . [
fig : amida1st1 ] using a mathematical induction . in fig .
[ fig : amida1st222 ] , we explicitly calculates the lowest order .
next , let us consider the mixed expansion in fig .
[ fig : amida1st4 ] for the higher orders .
suppose @xmath86 is an odd number .
the first and second terms cancel out each other by the same argument as noted in the context of eq .
( [ eq : lemma ] ) and fig .
[ fig : upsilon0 ] . if fig .
[ fig : amida1st1 ] is true for the lower orders , the sum of the third and fourth terms , the sum of the fifth and sixth terms , , are equal to zero . as a result , only the last term survives .
since the moment amida - diagram in it is equal to unity , we obtain @xmath303 .
next , suppose @xmath86 is an even number .
the sum of the first and the second terms is @xmath406 .
the third term is @xmath407 times @xmath408 .
hence , the first , second and third terms cancel out .
furthermore , if fig .
[ fig : amida1st1 ] is true for the lower order , the sum of the fourth and fifth terms , the sum of the sixth and seventh terms , , are equal to zero .
namely , again , only the last term survives .
since the moment amida - diagram in it is equal to unity , we obtain @xmath302 . for this purpose , an extrapolation of the fcm result and a direct calculation of a few leading orders
have been done : a. bhler , n. elstner and g. s. uhrig , eur .
j. b * 16 * , 475 ( 2000 ) ; w. o. putikka , m. u. luchini and r. r. p. singh , phys .
lett . * 81 * , 2966 ( 1998 ) . | we formulate a new method of performing high - temperature series expansions for the spin - half heisenberg model or , more generally , for su(@xmath0 ) heisenberg model with arbitrary @xmath0 .
the new method is a novel extension of the well - established finite cluster method .
our method emphasizes hidden combinatorial aspects of the high - temperature series expansion , and solves the long - standing problem of how to efficiently calculate correlation functions of operators acting at widely separated sites .
series coefficients are expressed in terms of cumulants , which are shown to have the property that all deviations from the lowest - order nonzero cumulant can be expressed in terms of a particular kind of moment expansion .
these `` quasi - moments '' can be written in terms of corresponding `` quasi - cumulants '' , which enable us to calculate higher - order terms in the high - temperature series expansion .
we also present a new technique for obtaining the low - order contributions to specific heat from finite clusters .
* keywords * : high - temperature expansion , heisenberg model , correlation function , specific heat , new method |
fermions at finite density or finite chemical potential is a subject of a wide range of interest .
it is relevant to condensed matter physics , such as the hubbard model away from half - filling .
the research about nuclei and neutron stars at low and high nucleon density is actively pursued in nuclear physics and astrophysics .
the subject of quark gluon plasma is important for understanding the early universe and is being sought for in relativistic heavy - ion collisions in the laboratories .
furthermore , speculation about color superconducting phase has been proposed recently for quantum chromodynamics ( qcd ) at very high quark density @xcite . although there are models , e.g. chiral models and nambu jona - lasinio model which have been used to study qcd at finite quark density , the only way to study qcd at finite density and temperature reliably and systematically is via lattice gauge calculations .
there have been extensive lattice calculations of qcd at finite temperature @xcite . on the contrary ,
the calculation at finite density is hampered by the lack of a viable algorithm . in this talk
, i shall first review the difficulties associated with the finite density algorithm with chemical potentials in sec .
i will then outline in sec .
3 a proposal for a finite density algorithm in the canonical ensemble which projects out the nonzero baryon number sector from the fermion determinant . in sec . 4 ,
a newly developed noisy monte carlo algorithm which admits unbiased estimate of the probability is described .
its application to the fermion determinant is outlined in sec .
i will discuss an efficient way , the pad - z@xmath0 method , to estimate the @xmath1 of the fermion matrix in sec .
the recent progress on the implementation of the kentucky noisy monte carlo algorithm to dynamical fermions is presented in sec .
finally , a summary is given in sec .
the usual approach to the finite density in the euclidean path - integral formalism of lattice qcd is to consider the grand canonical ensemble with the partition function @xmath2 e^{-s_g[u]},\ ] ] where the fermion fields with fermion matrix @xmath3 has been integrated to give the determinant .
@xmath4 is the gauge link variable and @xmath5 is the gauge action .
the chemical potential is introduced to the quark action with the @xmath6 factor in the time - forward hopping term and @xmath7 in the time - backward hopping term . here
@xmath8 is the lattice spacing .
however , this causes the fermion action to be non - hermitian , i.e. @xmath9 . as a result , the fermion determinant @xmath10 $ ] is complex and this leads to the infamous sign problem .
there are several approaches to avoid the sign problem : it was proposed by the glasgow group @xcite that the sign problem can be circumvented based on the expansion of the grand canonical partition function in powers of the fugacity variable @xmath11 , @xmath12 where @xmath13 is the canonical partition function for the baryon sector with baryon number @xmath14 .
@xmath15 is calculated with reweighting of the fermion determinant @xmath16}{det m[u,0 ] } \rangle_{\mu = 0}.\ ] ] since the reweighting is based on the gauge configuration with @xmath17 , it avoids the sign problem .
however , this does not work , except perhaps at small @xmath18 or near the finite temperature phase transition .
we will dwell on this later in sec .
this is caused by the ` overlap problem ' @xcite where the important samples of configurations in the @xmath17 simulation has exponentially small overlap with those relevant for the finite density . as a result ,
the onset of baryon begins at @xmath19 instead of the expected @xmath20 which resembles the situation of the quenched approximation . in this approach
, the chemical potential is taking an imaginary value @xmath21 .
the fermion determinant is real in this case and one can avoid the sign problem @xcite .
the partition function is @xmath22 which is periodic with respect to @xmath23 with a period of @xmath24 . comparing with eq .
( [ zgc ] ) , one can in principle obtain canonical partition function @xmath13 from the fourier transform @xmath25 in this approach , one needs to integrate over the whole range of @xmath23 from @xmath26 to @xmath24 after one obtains the monte carlo configurations of @xmath27 at different @xmath23 . in practice
, it is proposed to calculate the following ratio in the two - dimensional hubbard model @xcite , @xmath28 with a reference value @xmath29 .
several patches each centered around a different reference point @xmath29 are used to cover the range of @xmath23 .
this was successful for the two - dimensional hubbard model with a @xmath30 lattice up to b = 6 where the determinant was calculated exactly .
while this works for a small lattice in the hubbard model , it would not work for reasonably large lattices in qcd .
this is because the direct calculation of the determinant is a @xmath31 ( or @xmath32 for a sparse matrix ) operation which is an impracticable task for the quark matrix which is typically of the dimension @xmath33 .
any stochastic estimation of the determinant will inevitably introduce systematic error .
furthermore , this will also suffer from the ` overlap ' problem discussed above .
any monte carlo simulation at a reference point @xmath29 will have exponentially small overlap with those configurations important to a nonzero baryon density . to alleviate the sign problem with the real chemical potential and the overlap problem due to reweighting , it is proposed @xcite to do the reweighting in the multiple parameter space .
the generic partition function @xmath15 in eq .
( [ zgc ] ) is parametrized by a set of parameters @xmath34 , such as the chemical potential @xmath18 , the gauge coupling @xmath35 , the quark mass @xmath36 , etc .
the partition function can be written to facilitate reweighting @xmath37 e^{-s_g[u,\alpha_0 ] } \{e^{-s_g[u,\alpha ] + s_g[u,\alpha_0 ] } \frac{det m[u , \alpha]}{det m[u,\alpha_0]}\},\ ] ] where the monte carlo simulation is carried out with the @xmath38 set of parameters and the terms in the curly bracket are treated as observables .
this is applied to study the end point in the t-@xmath18 phase diagram . in this case , the monte carlo simulation is carried out where the parameters in @xmath38 include @xmath17 and @xmath39 which corresponds to the phase transition at temperature @xmath40 . the parameter set @xmath34 in the reweighted measure include @xmath41 and an adjusted @xmath35 in the gauge action .
the new @xmath35 is determined from the lee - yang zeros so that one is following the transition line in the t-@xmath18 plane and the large change in the determinant ratio in the reweighting is compensated by the change in the gauge action to ensure reasonable overlap .
this is shown to work to locate the transition line from @xmath42 and @xmath43 down to the critical point on the @xmath44 and @xmath45 lattices with staggered fermions @xcite . while the multi - parameter reweighting is successful near the transition line , it is not clear how to extend it beyond this region , particularly the @xmath46 case where one wants to keep the @xmath35 and quark mass fixed while changing the @xmath18 .
one still expects to face the overlap problem in the latter case . for large volumes , calculating the determinant ratio will be subjected to the same practical difficulty as discussed in the previous section [ icp ] .
we would like to propose an algorithm to overcome the overlap problem at zero temperature which is based on the canonical ensemble approach . to avoid the overlap problem
, one needs to lock in a definite nonzero baryon sector so that the exponentially large contamination from the zero - baryon sector is excluded . to see this
, we first note that the fermion determinant is a superposition of multiple quark loops of all sizes and shapes .
this can be easily seen from the property of the determinant @xmath47 upon a hopping expansion of @xmath48 , @xmath49 represents a sum of single loops with all sizes and shapes .
the determinant is then the sum of all multiple loops .
the fermion loops can be separated into two classes .
one is those which do not go across the time boundary and represent virtual quark - antiquark pairs ; the other includes those which wraps around the time boundary which represent external quarks and antiquarks .
the configuration with a baryon number one which entails three quark loops wrapping around the time boundary will have an energy @xmath50 higher than that with zero baryon number .
thus , it is weighted with the probability @xmath51 compared with the one with no net baryons .
we see from the above discussion that the fermion determinant contains a superposition of sectors of all baryon numbers , positive , negative and zero . at zero temperature where @xmath52 , the zero baryon sector dominates and all the other baryon sectors are exponentially suppressed .
it is obvious that to avoid the overlap problem , one needs to select a definite nonzero baryon number sector and stay in it throughout the markov chain of updating configurations . to select a particular baryon sector from the determinant
can be achieved by the following procedure @xcite : first , assign an @xmath53 phase factor @xmath54 to the links between the time slices @xmath55 and @xmath56 so that the link @xmath57 is multiplied by @xmath58 ; then the particle number projection can be carried out through the fourier transformation of the fermion determinant like in the bcs theory @xmath59\ ] ] where @xmath60 is the net particle number , i.e. particle minus antiparticle .
note that all the virtual quark loops which do not reach the time boundary will have a net phase factor of unity ; only those with a net n quark loops across the time boundary will have a phase factor @xmath61 which can contribute to the integral in eq .
( [ projection ] ) . since qcd in canonical formulation
does not break @xmath62 symmetry , it is essential to take care that the ensemble is canonical with respect to triality . to this end
, we shall consider the triality projection @xcite to the zero triality sector @xmath63.\ ] ] this amounts to limiting the quark number n to a multiple of 3 .
thus the triality zero sector corresponds to baryon sectors with integral baryon numbers .
another essential ingredient to circumvent the overlap problem is to stay in the chosen nonzero baryon sector so as to avoid mixing with the zero baryon sector with exponentially large weight .
this can be achieved by preforming the baryon number projection as described above _ before _ the accept / reject step in the monte carlo updating of the gauge configuration .
if this is not done , the accepted gauge configuration will be biased toward the zero baryon sector and it is very difficult to project out the nonzero baryon sector afterwords .
this is analogous to the situation in the nuclear many - body theory where it is known @xcite that the variation after projection ( zeh - rouhaninejad - yoccoz method @xcite ) is superior than the variation before projection ( peierls - yoccoz method @xcite ) .
the former gives the correct nuclear mass in the case of translation and yields much improved wave functions in mildly deformed nuclei than the latter .
to illustrate the overlap problem , we plot in fig.1 @xmath64 $ ] for a configuration of the @xmath65 lattice with the wilson action with @xmath66 and @xmath67 which is obtained with 500 @xmath68 noises .
we see that the it is rather flat in @xmath69 indicating that the fourier transform in eq .
( [ projection ] ) will mainly favor the zero baryon sector . on the other hand , at finite temperature
, it is relatively easier for the quarks to be excited so that the zero baryon sector does not necessarily dominate other baryon sectors .
another way of seeing this is that the relative weighting factor @xmath51 can be @xmath70 at finite temperature .
thus , it should be easier to project out the nonzero baryon sector from the determinant .
we plot in fig .
2 a similarly obtained @xmath64 $ ] for a configuration of the @xmath71 lattice with @xmath72 and @xmath73 .
we see from the figure that there is quite a bit of wiggling in this case as compared to that in fig .
1 indicating that it is easier to project out a nonzero baryon sector through the fourier transform at finite temperature . we should mention that while we think we can overcome the overlap problem and the determinant @xmath74 $ ] is real in this approach , nevertheless in view of the fact that the fourier transform in eq .
( [ projection ] ) involves the quark number @xmath60 the canonical approach may still have the sign problem at the thermodynamic limit when @xmath60 and @xmath75 are very large
. however , we think it might work for small @xmath60 such as 3 or 6 for one or two baryons in a finite @xmath75 .
this should be a reasonable start for practical purposes . while it is clear what the algorithm in the canonical approach entails ,
there are additional practical requirements for the algorithm to work .
these include an unbiased estimation of the huge determinant in lattice qcd and , moreover , a monte carlo algorithm which accommodates the unbiased estimate of the probability .
we shall discuss them in the following sections .
there are problems in physics which involve extensive quantities such as the fermion determinant which require @xmath31 steps to compute exactly .
problems of this kind with large volumes are not numerically applicable with the usual monte carlo algorithm which require an exact evaluation of the probability ratios in the accept / reject step . to address this problem , kennedy and kuti @xcite proposed a monte carlo algorithm which admits stochastically estimated transition probabilities as long as they are unbiased .
but there is a drawback .
the probability could lie outside the interval between @xmath26 and @xmath76 since it is estimated stochastically .
this probability bound violation will destroy detailed balance and lead to systematic bias .
to control the probability violation with a large noise ensemble can be costly .
we propose a noisy monte carlo algorithm which avoids this difficulty with two metropolis accept / reject steps .
let us consider a model with hamiltonian @xmath77 where @xmath4 collectively denotes the dynamical variables of the system .
the major ingredient of the new approach is to transform the noise for the stochastic estimator into stochastic variables .
the partition function of the model can be written as @xmath78\ , e^{-h(u ) } \nonumber \\ & = & \int [ du][d\xi]p_\xi(\xi)\ , f(u,\xi).\end{aligned}\ ] ] where @xmath79 is an unbiased estimator of @xmath80 from the stochastic variable @xmath81 and @xmath82 is the probability distribution for @xmath81 . the next step is to address the lower probability - bound violation .
one first notes that one can write the expectation value of the observable @xmath83 as @xmath84[d\xi]\,p_\xi(\xi ) \,o(u)\,{\rm sign}(f)\,|f(u,\xi)|/z,\ ] ] where @xmath85 is the sign of the function @xmath86 . after redefining the partition function to be @xmath87[d\xi]p_\xi(\xi)\
, |f(u,\xi)|,\ ] ] which is semi - positive definite , the expectation of @xmath83 in eq .
( [ o ] ) can be rewritten as @xmath88 as we see , the sign of @xmath79 is not a part of the probability any more but a part in the observable .
notice that this reinterpretation is possible because the sign of @xmath79 is a state function which depends on the configuration of @xmath4 and @xmath81 .
it is clear then , to avoid the problem of lower probability - bound violation , the accept / reject criterion has to be factorizable into a ratio of the new and old probabilities so that the sign of the estimated @xmath79 can be absorbed into the observable .
this leads us to the metropolis accept / reject criterion which incidentally cures the problem of upper probability - bound violation at the same time .
it turns out two accept / reject steps are needed in general . the first one is to propose updating of @xmath4 via some procedure while keeping the stochastic variables @xmath81 fixed .
the acceptance probability @xmath89 is @xmath90 the second accept / reject step involves the refreshing of the stochastic variables @xmath81 according to the probability distribution @xmath91 while keeping @xmath4 fixed .
the acceptance probability is @xmath92 it is obvious that there is neither lower nor upper probability - bound violation in either of these two metropolis accept / reject steps .
furthermore , it involves the ratios of separate state functions so that the sign of the stochastically estimated probability @xmath79 can be absorbed into the observable as in eq .
( [ onew ] ) .
detailed balance can be proven to be satisfied and it is unbiased @xcite .
therefore , this is an exact algorithm .
one immediate application of nmc is lattice qcd with dynamical fermions .
the action is composed of two parts the pure gauge action @xmath93 and a fermion action @xmath94 . both are functionals of the gauge link variables @xmath4 .
to find out the explicit form of @xmath79 , we note that the fermion determinant can be calculated stochastically as a random walk process @xcite @xmath95 this can be expressed in the following integral @xmath96,\end{aligned}\ ] ] where @xmath97 is the probability distribution for the stochastic variable @xmath98
. it can be the gaussian noise or the @xmath68 noise ( @xmath99 in this case ) .
the latter is preferred since it has the minimum variance @xcite .
@xmath100 is a stochastic variable with uniform distribution between 0 and 1 .
this sequence terminates stochastically in finite time and only the seeds from the pseudo - random number generator need to be stored in practice .
the function @xmath101 ( @xmath81 in eq .
( [ z ] ) is represented by two stochastic variables @xmath102 and @xmath103 here ) is represented by the part of the integrand between the the square brackets in eq .
( [ trln ] ) .
one can then use the efficient pad - z@xmath0 algorithm @xcite to calculate the @xmath104 in eq .
( [ trln ] ) .
we shall discuss this in the next section .
finally , there is a practical concern that @xmath105 can be large so that it takes a large statistics to have a reliable estimate of @xmath106 from the series expansion in eq .
( [ trln ] ) . in general , for the taylor expansion @xmath107 , the series will start to converge when @xmath108 .
this happens at @xmath109 .
for the case @xmath110 , this implies that one needs to have more than 100 !
stochastic configurations in the monte carlo integration in eq .
( [ trln ] ) in order to have a convergent estimate . even then
, the error bar will be very large . to avoid this difficulty
, one can implement the following strategy .
first , one notes that since the metropolis accept / reject involves the ratio of exponentials , one can subtract a universal number @xmath111 from the exponent @xmath112 in the taylor expansion without affecting the ratio .
second , one can use a specific form of the exponential to diminish the value of the exponent .
in other words , one can replace @xmath113 with @xmath114 to satisfy @xmath115 .
the best choice for @xmath111 is @xmath116 , the mean of @xmath112 . in this case , the variance of @xmath113 becomes @xmath117 .
now we shall discuss a very efficient way of estimating the fermion determinant stochastically @xcite . the starting point for the method
is the pad approximation of the logarithm function .
the pad approximant to @xmath118 of order @xmath119 $ ] at @xmath120 is a rational function @xmath121 where deg @xmath122 = deg @xmath123 = @xmath124 , whose value and first @xmath125 derivatives agree with @xmath126 at the specified point @xmath120 .
when the pad approximant @xmath121 is expressed in partial fractions , we obtain @xmath127 whence it follows @xmath128 the pad approximation is not limited to the real axis .
as long as the function is in the analytic domain , i. e. away from the cut of the log , say along the negative real axis , the pad approximation can be made arbitrarily accurate by going to a higher order @xmath119 $ ] and a judicious expansion point to cover the eigenvalue domain of the problem
. exact computation of the trace inverse for @xmath129 matrices is very time consuming for matrices of size @xmath130 .
however , the complex z@xmath0 noise method has been shown to provide an efficient stochastic estimation of the trace @xcite .
in fact , it has been proved to be an optimal choice for the noise , producing a _ minimum _ variance @xcite .
the stochastic error of the complex z@xmath0 noise estimate results only from the off - diagonal entries of the inverse matrix ( the same is true for z@xmath138 noise for any n ) .
however , other noises ( such as gaussian ) have additional errors arising from diagonal entries .
this is why the z@xmath0 noise has minimum variance .
for example , it has been demonstrated on a @xmath139 lattice with @xmath66 and @xmath140 for the wilson action that the z@xmath0 noise standard deviation is smaller than that of the gaussian noise by a factor of 1.54 @xcite . applying the complex z@xmath0 estimator to the expression for the @xmath141 in eq .
( [ tr_apprx ] ) , we find @xmath142 where @xmath143 are the solutions of @xmath144 since @xmath145 are shifted matrices with constant diagonal matrix elements , eq .
( [ col_inv1 ] ) can be solved collectively for all values of @xmath146 within one iterative process by several algorithms , including the quasi - minimum residual ( qmr ) @xcite , multiple - mass minimum residual ( m@xmath147 r ) @xcite , and gmres@xcite .
we have adopted the m@xmath147 r algorithm , which has been shown to be about 2 times faster than the conjugate gradient algorithm , and the overhead for the multiple @xmath146 is only 8% @xcite . the only price to pay
is memory : for each @xmath146 , a vector of the solution needs to be stored .
furthermore , one observes that @xmath148 .
this improves the conditioning of @xmath149 since the eigenvalues of @xmath150 have positive real parts .
hence , we expect faster convergence for column inversions for eq .
( [ col_inv1 ] ) . in the next section ,
we describe a method which significantly reduces the stochastic error .
we now turn to the question of choosing suitable traceless matrices @xmath151 to use in the modified estimator .
one possibility for the wilson fermion matrix @xmath159 is suggested by the hopping parameter @xmath160 expansion of the inverse matrix , @xmath161 this suggests choosing the matrices @xmath151 from among those matrices in the hopping parameter expansion which are traceless : @xmath162 it may be verified that all of these matrices are traceless . in principle , one can include all the even powers which entails the explicit calculation of all the allowed loops in @xmath163 . in this manuscript
we have only included @xmath164 , @xmath165 , and @xmath166 .
our numerical computations were carried out with the wilson action on the @xmath65 ( @xmath60 = 73728 ) lattice with @xmath168 .
we use the hmc with pseudofermions to generate gauge configurations . with a cold start , we obtain the fermion matrix @xmath169 after the plaquette becomes stable .
the trajectories are traced with @xmath170 and 30 molecular dynamics steps using @xmath67 .
@xmath171 is then obtained from @xmath169 by an accepted trajectory run .
hence @xmath169 and @xmath171 differ by a continuum perturbation , and @xmath172 \sim { \cal o}(1)$ ] .
we first calculate @xmath173 with different orders of pad expansion around @xmath174 and @xmath175 .
we see from table 1 that the 5th order pad does not give the same answer for two different expansion points , suggesting that its accuracy is not sufficient for the range of eigenvalues of @xmath169 .
whereas , the 11th order pad gives the same answer within errors .
thus , we shall choose p[11,11]@xmath176 with @xmath177 to perform the calculations from this point on . with 50 noises as a function of the order of subtraction and compared to that of unimproved estimate with 10,000 noises .
the dashed lines are drawn with a distance of 1 @xmath178 away from the central value of the unimproved estimate . ] in table 2 , we give the results of improved estimations for @xmath179 .
we see that the variational technique described above can reduce the data fluctuations by more than an order of magnitude .
for example , the unimproved error @xmath180 in table 2 for 400 z@xmath0 noises is reduced to @xmath181 for the subtraction which includes up to the @xmath182 matrix .
this is 37 times smaller . comparing the central values in the last row ( i.e. the @xmath183 order improved ) with that of unimproved estimate with 10,000 z@xmath0 noises
, we see that they are the same within errors . this verifies that the variational subtraction scheme that we employed does not introduce biased errors .
the improved estimates of @xmath179 from 50 z@xmath0 noises with errors @xmath184 from table 2 are plotted in comparison with the central value of the unimproved estimate from 10,000 noises in fig .
we have recently implemented the above noisy monte carlo algorithm to the simulation of lattice qcd with dynamical fermions by incorporating the full determinant directly @xcite .
our algorithm uses pure gauge field updating with a shifted gauge coupling to minimize fluctuations in the trace log is the wilson dirac matrix .
it gives the correct results as compared to the standard hybrid monte carlo simulation .
however , the present simulation has a low acceptance rate due to the pure gauge update and results in long autocorrelations .
we are in the process of working out an alternative updating scheme with molecular dynamics trajectory to include the feedback of the determinantal effects on the gauge field which should be more efficient than the pure gauge update .
after reviewing the finite density algorithm for qcd with the chemical potential , we propose a canonical approach by projecting out the definite baryon number sector from the fermion determinant and stay in the sector throughout the monte carlo updating .
this should circumvent the overlap problem . in order to make the algorithm practical
, one needs an efficient way to estimate the huge fermion determinant and a monte carlo algorithm which admits unbiased estimates of the probability without upper unitarity bound violations .
these are achieved with the pad - z@xmath0 estimate of the determinant and the noisy monte carlo algorithm .
so far , we have implemented the kentucky noisy monte carlo algorithm to incorporate dynamical fermions in qcd on a relatively small lattice and medium heavy quark based on pure gauge updating . as a next step
, we will work on a more efficient updating algorithm and project out the baryon sector to see if the finite density algorithm proposed here will live up to its promise .
this work is partially supported by the u.s .
doe grant de - fg05 - 84er40154 .
the author wishes to thank m. faber for introducing the subject of the finite density to him and shao - jing dong for providing some unpublished figures .
fruitful discussions with m. alford , i.barbour , u .- j .
wiese , and r. sugar are acknowledged .
he would also thank the organizer , ge mo - lin for the invitation to attend the conference and his hospitality .
0 m. alford , k. rajagopal , and f. wilczek , _ phys .
* b422 * , 247 ( 1998 ) ; r. rapp , t. schaefer , e.v .
shuryak , and m. velkovsky , _ phys .
lett . _ * 81 * , 53 ( 1998 ) ; for a review , see for example , k. rajagopal and f. wilczek , hep - ph/0011333 .
i. m. barbour , s. e. morrison , e. g. klepfish , j. b. kogut , and m .-
lombardo , _ nucl .
( proc . suppl . )
_ * 60a * , 220 ( 1998 ) ; i.m .
barbour , c.t.h .
davies , and z. sabeur , _ phys .
* b215 * , 567 ( 1988 ) . | i will review the finite density algorithm for lattice qcd based on finite chemical potential and summarize the associated difficulties .
i will propose a canonical ensemble approach which projects out the finite baryon number sector from the fermion determinant .
for this algorithm to work , it requires an efficient method for calculating the fermion determinant and a monte carlo algorithm which accommodates unbiased estimate of the probability .
i shall report on the progress made along this direction with the pad - z@xmath0 estimator of the determinant and its implementation in the newly developed noisy monte carlo algorithm .
* uk/02 - 02 * + jan .
2002 + * finite density algorithm in lattice qcd + a canonical ensemble approach * keh - fei liu _ dept . of physics and astronomy , university of kentucky , lexington ,
ky 40506 _ |
the families of high - transition - temperature ( high-@xmath4 ) superconductors are fundamentally composed of two - dimensional copper oxygen planes into which charge carriers , either holes or electrons , are doped . prior to doping charge carriers ,
the parent compounds of the high-@xmath4 copper oxides are antiferromagnetically ordered insulators whose spin dynamics are well modeled by a two - dimensional heisenberg antiferromagnetic ( af ) formalism with an anomalously large nearest - neighbor exchange coupling ( @xmath5 mev ) between the cu sites within the cuo@xmath6 plane @xcite . as spin fluctuations may play a crucial role in the mechanism of high-@xmath4 superconductivity @xcite ,
it is imperative to have a comprehensive picture on how the spin dynamics of the undoped af parent compounds evolve as they are tuned towards optimally - doped superconductivity .
while a comprehensive picture of this carrier induced modification to the spin dynamics has emerged for different classes of hole - doped high-@xmath4 materials @xcite , experiments exploring spin fluctuations in electron - doped copper oxides are just beginning @xcite . as a consequence ,
studies of electron - doped materials provide a unique litmus for testing the electron - hole symmetry in spin dynamical properties .
if spin fluctuations are fundamental to the mechanism of high-@xmath4 superconductivity , they should have universal features for all copper - oxide systems . in the case of hole - doped high-@xmath4 superconductors such as yba@xmath6cu@xmath7o@xmath8 ( ybco ) ,
the most prominent feature in its spin excitation spectrum is a sharp magnetic excitation termed ` resonance ' observed by inelastic neutron scattering .
the resonance is centered at the af ordering wavevector @xmath9 in the two - dimensional reciprocal space of the cuo@xmath6 planes ( see inset of fig .
1a ) and is intimately related to superconductivity @xcite .
our recent discovery of the resonance mode in the electron - doped pr@xmath0lace@xmath1cuo@xmath2 ( @xmath10 k ) demonstrates that the resonance is a universal feature of high-@xmath4 copper oxides regardless of carrier type @xcite . this unifying feature in the spin excitation spectrum of both hole- and electron - doped cuprates contrasts the prevalent asymmetry known between the commensurate low energy spin fluctuations centered at @xmath9 in the electron - doped cuprate systems @xcite and the observed incommensurate spin excitations at @xmath11 in several classes of hole - doped materials @xcite .
while the doping dependence of the resonance excitation has been observed to follow @xmath12 regardless of carrier type @xcite and the doping evolution of the incommensurate spin fluctuations in hole - doped materials is also controlled by @xmath4 @xcite , the doping dependence of the low energy commensurate spin fluctuations in electron - doped materials remains unexplored .
lace@xmath1cuo@xmath2 ( plcco ) and magnetic bulk susceptibility measurements for plcco samples studied .
a ) phase diagram of plcco from kang _
@xcite with the @xmath13 k sample added .
the blue arrows denote the sample compositions and locations on the phase diagram that were probed in the neutron experiments .
the inset shows the direction of @xmath14-scans with the circle denoting @xmath15 and the cross indicating where the offset background was measured @xcite .
b ) bulk susceptibility measurements showing the superconducting phase transitions for the samples studied . both the onset temperature of superconductivity and the point of 90@xmath16 of the normal state response ( shown as dashed lines ) are reported .
susceptibilities have been normalized to one another for plotting purposes . ] for this reason , we chose to investigate the evolution of low frequency spin fluctuations in the electron - doped cuprate pr@xmath0lace@xmath1cuo@xmath2 ( plcco ) . compared with the prototypical electron - doped copper oxide nd@xmath17ce@xmath18cuo@xmath2 ( ncco ) , studying spin fluctuations in plcco offers several distinct advantages : first , crystalline electric field ( cef ) levels of pr@xmath19 in the tetragonal unit cell of plcco have a nonmagnetic , singlet ground state , which avoids complications arising from the magnetic ground state of nd@xmath19 in ncco @xcite .
second , plcco can be tuned from an as - grown nonsuperconducting ( nsc ) antiferromagnet to an optimally doped superconductor , without static af order , through an annealing process @xcite ; whereas static af order coexists with superconductivity in ncco even at optimal doping @xcite .
finally , the cubic ( pr , la , ce)@xmath6o@xmath7 impurity phase arising from the annealing process @xcite in plcco has a nonmagnetic ground state as compared to the magnetic ground state of the ( nd , ce)@xmath6o@xmath7 impurity phase in ncco @xcite . in this article , we present a comprehensive study of the low energy spin dynamics in plcco from 0.5 mev @xmath20 5 mev in a variety of samples as plcco is tuned from a nsc antiferromagnet into an optimally doped superconductor ( @xmath10 k ) through the annealing process @xcite .
electron - doped copper oxides differ from the their hole - doped counterparts in that they require a post - growth annealing treatment in a low - oxygen atmosphere to remove excess oxygen and achieve superconductivity @xcite .
this means that the phase diagram of electron - doped copper oxides is three - dimensional as a function of both ce and oxygen concentration ( fig .
there are therefore two distinct ways to traverse the phase diagram of plcco : samples can be prepared at a fixed annealing condition but grown with variable ce doping levels @xcite , or alternatively , samples can be grown with a fixed ce concentration but with variable annealing treatment @xcite . for our studies ,
we utilize the latter method with a fixed ce concentration of @xmath21 and the resulting phase diagram is shown in fig .
it is clear that the annealing process induces a rapid suppression of the long - range af order and the eventual emergence of a superconducting phase transition . with continued oxygen removal , the superconducting phase
is enhanced to an optimum @xmath4 where the static af order is completely suppressed . for samples which exhibit a superconducting phase , an additional quasi - two - dimensional spin density wave ( sdw ) order appears at the disallowed three - dimensional af ordering wave vector @xmath22@xmath23 in the cuo@xmath6 plane @xcite .
the sdw order has an onset of approximately @xmath24 for underdoped superconducting plcco samples .
the rest of this article is organized as follows : the next section describes the experimental procedure , including details of sample preparation and neutron spectrometer setup . in section iii , we discuss low energy magnetic excitations in the as - grown , af ordered nsc plcco .
subsequent sections ( iv - vi ) cover spin fluctuations in superconducting samples annealed gradually toward optimal superconductivity . as af order
is suppressed in the system , the low energy excitations transition from regimes coupled to the onset of the af phase into a virtually temperature independent regime similar to those observed in several heavy fermion systems known to be near a quantum critical point ( qcp ) in the phase diagram @xcite . in section vii , we discuss the applicability of quantum critical scaling in describing the spin dynamics of plcco with different transition temperatures .
finally , in section viii we briefly summarize the key conclusions of the work .
for our experiments , we grew high quality single crystal plcco samples in an infrared mirror image furnace using the traveling solvent floating zone technique .
all samples were confirmed to have a mosaic of @xmath25 .
following their growth , samples were annealed in either a high vacuum environment ( @xmath26 mbar ) or in an argon gas environment at variable temperatures .
the resulting magnetic susceptibility and superconducting phase transition were measured in a squid magnetometer and are shown for each sample in fig .
the superconducting transition temperature is indicated by both the onset of the diamagnetic signal , @xmath4 ( onset ) , and through the dashed lines showing the 90% position of normal - state bulk susceptibility , @xmath4 ( 90% ) . for the remainder of the paper
, we will reference these samples by their respective onset temperature of superconductivity , @xmath4 ( onset ) .
the as - grown sample , with a mass of @xmath273.0 g , is nonsuperconducting and is labeled as nsc .
the masses of the other plcco s are @xmath271.8 g , @xmath274.8 g , and @xmath273.0 g for the @xmath28 k , @xmath13 k , and @xmath10 k samples , respectively .
an argon annealing atmosphere was used in treating the @xmath28 k sample at 940 @xmath29c for 24 hours @xcite .
the other two samples were annealed for four days in a high vacuum environment at 765 @xmath29c and 775 @xmath29c for the @xmath13 k , and @xmath10 k samples , respectively .
k samples .
a ) order parameter for the n@xmath30el phase in the nsc plcco .
all points are integrated intensities of the * q*=(0.5 , 1.5 , 0 ) af bragg reflection .
b ) representative elastic @xmath14-scans through * q*=(0.5 , 1.5 , 0 ) along the @xmath31 $ ] direction both below @xmath32 at 40 k and above @xmath32 at 218 k. c ) magnetic af order parameter for the @xmath13 k sample .
the sample is aligned in the @xmath33 $ ] scattering zone .
points are the peak intensities collected at the @xmath34 bragg position .
d ) elastic @xmath14-scans through the @xmath35 position along the @xmath36 $ ] direction .
a weak af bragg reflection appears below @xmath32 at 2 k and disappears above @xmath32 at 30 k. ] we reference positions in reciprocal space at wave vector * q*@xmath37 in @xmath38 using @xmath39 ( r.l.u . )
notation , where @xmath40 for the tetragonal plcco unit cell ( space group : @xmath41 @xmath42 , @xmath43 ) .
neutron scattering experiments were performed at the nist center for neutron research on the spins and bt-9 triple - axis spectrometers .
data was collected on the cold - neutron spectrometer spins with a fixed final energy of @xmath44 mev and collimations of open-80@xmath45-sample-80@xmath45-open - detector for nsc , @xmath28 k , and @xmath13 k samples , while an @xmath46 mev with identical collimations was used for spins experiments on the @xmath10 k samples . for data collected from bt-9 , an @xmath47 mev was used with collimations of 40@xmath45 - 60@xmath45-sample-80@xmath45 -open - detector . unless otherwise stated , plcco samples were aligned within the @xmath48 $ ] scattering plane for experiments on spins , and within the @xmath33 $ ] scattering zone for bt-9 .
alignment in the @xmath48 $ ] scattering zone allows for an effective integration along the @xmath49-axis of the quasi - two - dimensional , inelastic scattering from the cu spins in uncorrelated cuo@xmath6 planes ( through relaxed out - of - plane resolution ) , whereas alignment in the @xmath33 $ ] scattering zone allows an accurate determination of the onset of the three - dimensional af phase in the system .
masses for the samples used in these neutron experiments correspond to the masses stated above in the bulk magnetization measurements with the exception of experiments probing the @xmath10 k system . for experiments on this optimally - doped plcco system ,
a set of three @xmath10 k samples were co - aligned with a total mass of @xmath279 g @xcite .
all samples were mounted and loaded into a liquid - he cooled cryostat , and the experiments were performed in the range from @xmath50 k to @xmath51 k. scans taken at @xmath50 k. the peak signal was collected at @xmath52 ( black diamonds ) and the nonmagnetic background was measured at @xmath53 ( black triangles ) .
the dashed line is a linear fit to the energy dependence of the nonmagnetic background .
the instrumental energy resolution of fhwm@xmath54 mev is determined from vanadium scans and is shown as a solid bar under the elastic incoherent peak .
b - d ) constant-@xmath55 scans along the @xmath56 $ ] direction at @xmath57 , 1.5 , and 4.0 mev and collected at @xmath582 k. peaks are fitted by gaussians centered at @xmath59 and are resolution - limited . ]
we first opted to study the behavior of as - grown plcco , prior to any annealing treatment . for this untreated system
, long - range af order persists up to a @xmath60 k as shown in fig .
the rapid intensity increase below about 80 k arises from the induced pr@xmath19 moment through cu@xmath61-pr@xmath19 interaction @xcite .
scans through the allowed three - dimensional af ordering wave vector @xmath22@xmath62 show the appearance of a resolution - limited af bragg reflection centered at @xmath22@xmath62 which diminishes above @xmath32 ( fig .
due to the noncollinear structure of the zero field spin arrangement in plcco , the in - plane af bragg reflection at @xmath59 is disallowed , resulting in a gapped excitation spectrum at this wave vector @xcite .
constant-@xmath55 scans along the @xmath63 $ ] direction show this low energy gap in figs .
3b - d . in fig .
3b , @xmath64 scans at @xmath50 k through the @xmath59 position at @xmath57 mev show no magnetic scattering , whereas for @xmath65 mev clear resolution - limited spin - wave peaks are observed centered at @xmath59 .
these resolution - limited spin - wave peaks are well fit by gaussians and give minimum in - plane correlation lengths of @xmath66 at @xmath67 mev and @xmath68 at @xmath69 mev @xcite . now plotting in fig .
3a the constant-@xmath14 scans at the @xmath59 position , this low energy gap becomes clearer .
background points taken at @xmath70 are overplotted with raw data taken at the peak @xmath59 position , showing the presence of the low energy gap to be @xmath71 mev .
this gap energy along with the enhancement in the observed magnetic scattering for @xmath72 mev is consistent with previous studies of the parent compound pr@xmath6cuo@xmath6 in which the opening of an in - plane anisotropy gap was observed in addition to the lower interplane gap @xcite .
the temperature dependence of the low energy spin waves in this nsc sample is shown in figs .
4a - c . for energies above the gap , the spin - wave excitations increase in intensity with increasing temperature ( for @xmath73 ) following the bose population factor @xmath74 = 1/(1-exp(-\hbar \omega/{k_b t}))$ ] . removing the nonmagnetic background contributions to the scattering , we obtain @xmath75@xmath22@xmath76 and determine the imaginary part of the dynamic susceptibility , @xmath77@xmath22@xmath76 , using @xmath75@xmath22@xmath78\chi^{\prime \prime}({\bf q},\omega)$ ] .
now plotting @xmath79 in figs .
4d - f , the @xmath80-independence of the dynamic susceptibility for @xmath81 becomes clear .
the peaks in @xmath82 at these energies remain @xmath80-independent consistent with the excitations simply following the bose statistics expected for spin - wave excitations .
the presence of bose - populated spin wave excitations in this nsc sample is similar to those observed in parent compounds la@xmath6cuo@xmath83 and pr@xmath6cuo@xmath83 @xcite . above @xmath24 , however , the low energy spin gap closes and spin fluctuations appear in the @xmath84 mev channel .
this is most likely due to the emergence of classical critical fluctuations arising from the suppression of af order at @xmath85 k , or due to a crossover from three - dimensional to two - dimensional spin fluctuations as the weak out - of - plane cu exchange coupling breaks down above @xmath32 . the overall picture of spin dynamics in this nsc sample is well described by spin waves arising from the long - range af order in this system , which now provides a baseline for studying how these spin excitations evolve as the system is tuned into superconductivity .
-scans along the @xmath56 $ ] direction for various temperatures at @xmath57 , 1.5 , and 4.0 mev .
peaks in scattering are fitted by gaussians on a linear background .
a clear gap in scattering at @xmath57 mev persists above 40 k until the three - dimensional order breaks down above @xmath32 .
d - e ) measured dynamic susceptibility at various temperatures for @xmath57 , 1.5 , and 4.0 mev .
dashed lines are gaussian fits to @xmath80=218 k @xmath86 mev ) in panel ( d ) , @xmath80=2 k @xmath87 mev ) in panel ( e ) , and @xmath80=2 k @xmath88 mev ) in panel ( f ) . ]
a ) representative constant-@xmath14 scans at 2 k , 40 k , and 70 k collected at @xmath59 ( circles ) , where the solid black line shows the background scattering collected at @xmath70 ( triangles ) . at @xmath89
k , a clear enhancement can be seen around @xmath90 mev .
the solid bracket underneath the incoherent elastic line is the measured energy resolution by a vanadium standard .
b ) the temperature dependence of the @xmath91 mev and @xmath92 mev scattering at @xmath93 .
the arrows mark the onset of @xmath24 and @xmath94 .
c - e ) @xmath14-scans of the @xmath28 k plcco along @xmath63 $ ] for @xmath92 , 1.5 , and 4 mev and @xmath89 k , 40 k , 70 k and 118 k. center brackets are instrumental resolutions measured by resolution - limited spin - wave peaks from the long - range af ordered as - grown nsc sample . the inset in panel ( e ) shows @xmath14-scans along the @xmath49-axis , where solid line is the cu@xmath61 form factor squared and the dashed line marks the background . ] we now turn to study an underdoped , superconducting plcco sample that is annealed to a @xmath28 k with a coexisting @xmath95 k @xcite .
the resulting low energy fluctuations around the af ordering wave vector for this sample are plotted in fig .
5 . raw scattering intensities from energy scans at @xmath59 from @xmath96 mev
are shown in fig .
5a for temperatures above and below both @xmath4 and @xmath32 as solid circles , whereas the nonmagnetic background collected at @xmath22=(0.53 , 0.53 , 0 ) is overplotted as solid triangles . one immediate difference appearing between the spectra of this underdoped superconductor and the nsc sample is the appearance of a peak in the low-@xmath80 dynamic susceptibility along with the absence of a low energy spin - gap .
figures 5c - e show this gapless peak through @xmath14-scans at @xmath97=0.5 , 1.5 , and 4 mev , where the @xmath80=2 k response displays a clear enhancement at @xmath97=1.5 mev .
solid lines show that gaussian fits centered at @xmath15 on linear backgrounds can well describe the observed peaks in figs .
the @xmath50 k excitations exhibit widths much broader than the resolution - limited spin - wave peaks of the nsc system ( figs . 3c and 3d ) .
calculating the fourier transforms of these gaussian peaks yields the minimum in - plane dynamic correlation lengths of @xmath98 at @xmath57 mev , @xmath99 at @xmath67 mev , and @xmath100 at @xmath69 mev . using the resolution - limited widths of the nsc plcco as reference , the true in - plane dynamic spin correlation lengths , @xmath101 ,
are calculated to be @xmath102 at @xmath67 mev and @xmath103 at @xmath69 mev for the @xmath28 k plcco .
the substantially broader @xmath14-widths in figs .
5c - e than those of the resolution - limited spin waves in the nsc plcco suggest that these excitations can not arise from the classical spin - wave scattering from the three - dimensional static n@xmath104el ordered phase in the sample @xcite . upon warming ,
the low-@xmath80 peak in the dynamic susceptibility vanishes and excitations at @xmath105 mev populate upward until the system is warmed above @xmath32 . above @xmath32 , however , there is a crossover in the magnetic response , and the spectral weight for @xmath105 mev begins to decrease with increasing temperature .
this crossover is also shown in fig .
5b in which the intensity of the @xmath84 mev excitations and the static @xmath106 mev sdw moment are both plotted as a function of temperature . with increasing temperature from 2 k , the @xmath84 mev fluctuations increase in intensity until the breakdown of the static sdw order near @xmath32 . on further warming , these @xmath84 mev fluctuations begin to damp until disappearing for temperatures above 118 k ( figs . 5b and 5c ) .
these excitations are also found to be quasi - two - dimensional within the cuo@xmath6 layers through experiments which orient the crystal in the @xmath107 $ ] scattering plane .
this facilitates @xmath108 $ ] scans in the inset of fig .
5e , where the @xmath109-dependence of the spin excitations at @xmath84 , 1.5 , 4 mev is rod - like and simply decays following the cu@xmath61 magnetic form factor ( solid line ) .
this weak correlation perpendicular to the cuo@xmath6 planes indicates that spin excitations are uncorrelated along the @xmath49-axis and reflect the highly anisotropic exchange coupling between cu sites within the cuo@xmath6 plane ( @xmath110 mev ) and the much weaker out - of - plane exchange @xcite . at the @xmath59 position for @xmath111 , 1.5 , and 4.0 mev .
a - c ) @xmath112 for @xmath113 ( @xmath50 k and 40 k ) .
d - f ) @xmath112 for @xmath114 ( @xmath115 k and 118 k ) . in all panels ,
solid lines are gaussian fits to the @xmath50 k dynamic susceptibility at the corresponding energy . ] for @xmath116 mev , the magnetic scattering intensity at @xmath117 is @xmath80-independent from @xmath50 k to 120 k as shown in figs . 5a and 5e . in contrast ,
the spin scattering at @xmath118 mev is strongly coupled to the appearance of the af phase .
hence , this regime of @xmath80-independent @xmath119 provides additional evidence of a crossover in the spin dynamics that is now instead reflected in the energy scale of the observed spin fluctuation spectrum .
a changeover in the response of the system can also be tested by plotting @xmath120 at various energies for temperatures both above and below @xmath32 as shown in fig .
the top panels in figs .
6a - c show the measured dynamic susceptibility below @xmath32 .
from these , the population of the lowest energy excitations at @xmath57 mev can be seen to simply follow bose statistics with @xmath121@xmath22@xmath122 ) remaining constant . at @xmath67 mev
, there exists a slight decrease in the susceptibility upon warming to 40 k , most likely due to the vanishing peak in the susceptibility ( which is present at 2 k , see fig . 5 ) .
the susceptibility at @xmath69 mev , however , recovers this behavior of following the bose population factor with no @xmath80-dependence in @xmath120 up to @xmath123 k. for temperatures above @xmath32 , the dynamic susceptibility decreases sharply with increasing @xmath80 for all energies studied as shown in figs .
this is simply reflective of the decrease in magnetic scattering with increasing @xmath80 ( for @xmath124 ) at @xmath125 and 1.5 mev .
the decrease in @xmath126 at @xmath69 mev , instead , arises from the observed @xmath80-independent @xmath119 for temperatures up to 120 k. this is in sharp constrast to the @xmath127 mev spin - wave excitations in as - grown plcco , where the intensity of @xmath119 is entirely controlled by the bose statistics ( figs . 4c and 4f ) .
an interesting question then arises : how do these regimes of energy and temperature @xmath128 that couple to either the af phase or the paramagnetic phase ( giving rise to this @xmath80-independent response in @xmath129 ) evolve with increased doping ?
we address this question below through both the experimental observation ( sections v and vi ) and data analysis ( section vii ) .
we now describe results on an plcco sample with @xmath13 k ( fig .
experiments probing the static magnetic ordering of this system show an af bragg reflection at the @xmath130 position with a @xmath131 25 k ( fig . 2c ) .
@xmath14-scans through this af ordering wave vector along the @xmath108 $ ] direction reveal that this weak reflection disappears for @xmath132 k ( fig .
2d ) . in fig .
7a , the measured scattering at @xmath59 for @xmath133 mev is plotted as crossed - boxes , while nonmagnetic background collected at @xmath134 is plotted as solid triangles . examining these low-@xmath55 fluctuations , substantial differences appear between the spectra of this @xmath13 k sample and those of the @xmath28 k system .
the excitations at the @xmath59 position remain gapless ; however , they lack a clearly defined peak in the low-@xmath80 susceptibility at @xmath50 k. a comparison of the @xmath50 k magnetic excitations at @xmath111 , 1.5 , and 4.0 mev is shown between the @xmath13 k and @xmath28 k plcco samples in figs . 7 c - e .
the dashed lines show gaussian fits to the scattering observed in the @xmath28 k sample whose enhancement at @xmath67 mev contrasts the continual decrease of scattering intensity with increasing energy transfer seen in the @xmath13 k plcco ( crossed - box symbols ) .
additionally , these plots show that the excitations observed in the @xmath13 k system have broadened in @xmath14 at all energies relative those observed in the @xmath28 k sample .
the fourier transforms of the gaussian fits in figs .
7c - e give minimum in - plane dynamic spin correlation lengths of @xmath135 at @xmath57 mev , @xmath136 at @xmath67 mev , and @xmath137 at @xmath69 mev .
correcting for the instrumental resolution yields @xmath138 at @xmath67 mev and @xmath139 at @xmath69 mev for the @xmath13 k plcco .
a ) constant-@xmath14 scans at @xmath117 and 2 k , 30 k , 60 k and 100 k. the solid black line shows the background scattering at @xmath140 and the horizontal bar beneath the incoherent elastic peak is the instrumental resolution .
b ) temperature dependence of the inelastic ( @xmath84 mev ) scattering at @xmath117 and elastic ( @xmath106 mev ) scattering at @xmath130 bragg position also shown in fig .
2c demonstrating a magnetic order at @xmath141 k. c - e ) @xmath14-scans of the @xmath13 k plcco along the @xmath63 $ ] direction for @xmath92 , 1.5 , and 4 mev , respectively , at @xmath89 k. the dashed lines show identical scans from the @xmath28 k sample with its @xmath142 bragg intensity normalized to that of the @xmath13 k sample .
center brackets are instrumental resolutions and solid lines are gaussian fits on flat backgrounds . ] the temperature dependence of the magnetic excitations in fig .
7 a shows a similar type of crossover in the spin dynamics to those observed in the under - doped @xmath28 k system .
the regimes of these two types of spin dynamics are again determined by the energy and temperature scale of the af order in the system , now with a @xmath131 25 k. for @xmath73 , there exists a slight enhancement upon warming of the spin fluctuations at @xmath143 mev , whereas for @xmath124 and @xmath144 mev the observed scattering intensity remains @xmath80-independent over a broad temperature range ( @xmath50 k @xmath145 60 k ) .
the relative intensity of the magnetic scattering at @xmath146 k is difficult to determine due to uncertainties in the nonmagnetic background contributions , whereas there was no observed change in background scattering between @xmath50 k and 60 k. the enhancement coupled to the af order is significantly damped and no longer strictly follows the bose population factor . instead ,
a weak increase in the population of the @xmath84 mev fluctuations is observed ( as shown in fig .
7b ) as the system is warmed towards @xmath32 .
the coupling of these low energy fluctuations to @xmath32 as a function of @xmath80 is also significantly broadened ( fig .
7b ) , in contrast to that of the @xmath28 k plcco ( fig .
5b ) . on the other hand , the temperature and energy region in which the magnetic scattering intensity is temperature independent increases , reflecting the much weaker af phase in this sample along with the reduced energy scale of @xmath32 .
a - c ) constant-@xmath55 scans at @xmath50 , 30 , 55 , 80 , and 95 k through the @xmath59 position along the @xmath56 $ ] direction .
solid lines are gaussian fits on linear backgrounds .
the @xmath50 , 30 , and 55 k data are replotted for comparison from ref .
solid brackets show the resolution of the spectrometer .
the increased background scattering with increasing temperature at @xmath84 and 1.5 mev may arise from single / multi - phonon scattering and/or air scattering .
d - f ) @xmath112 at @xmath50 , 30 , 55 , 80 , and 95 k. dashed lines are gaussian fits to the @xmath50 k dynamic susceptibility . ]
turning now to the final optimally doped @xmath10 k sample , previous experiments have shown that there exists no static af order co - existing with superconductivity in this sample down to 600 mk @xcite .
the low energy excitations for this system were first reported in ref .
11 and are expanded upon in fig .
figures 8a - c show the raw scattering intensities observed for @xmath57 , 1.5 , and 3.5 mev at various temperatures .
the spin excitations in these energies are fitted by gaussians on linear backgrounds and give minimum in - plane dynamic spin correlation lengths of @xmath149 at @xmath57 mev , @xmath150 at @xmath151 mev , and @xmath152 at @xmath153 mev at @xmath50 k @xcite .
these widths are within error to those observed in the @xmath13 k sample when neglecting the slight change in instrumental resolution on changing @xmath154 from 5.0 mev to 3.7 mev .
this is reasonable since the measured @xmath14-widths are appreciably larger than the instrument resolution in both geometries @xcite . upon warming from @xmath50 k to @xmath155 k ,
there is no change in the measured magnetic scattering intensity at all energies down to @xmath57 mev .
further increase in temperature to @xmath156 k renders a slight reduction in scattering at @xmath84 mev and 1.5 mev , with no change measured in the 3.5 mev excitations . at @xmath157
k , the peak at @xmath57 mev has completely vanished .
there exists a strong suppression of the @xmath158 mev fluctuations at @xmath159 k. in contrast , there are no changes in the @xmath160 mev fluctuations up to 80 k. figures 8d - e show the measured @xmath161 for the same energies reflecting a continued decrease in the susceptibility with increasing temperature .
the dashed lines show gaussian fits centered at @xmath162 for the @xmath50 k susceptibility to highlight the dramatic decrease in @xmath161 at different energies ( figs .
8d - f ) . at the highest measured temperature ( @xmath157 k ) , the system becomes gapped at @xmath57 mev similar to the two other underdoped plcco .
the absence of any regime in which the dynamic susceptibility remains constant , reflective of bosonic excitations similar to those observed in the @xmath28 k samples , demonstrates a drastic deviation from the two - distinct regimes of magnetic response observed in the @xmath28 k plcco .
therefore , low energy spin excitations in plcco evolve from coupling to the onset of the af phase in underdoped materials to essentially temperature independent from 2 k to 30 k for the optimally doped sample .
recently , we have discovered that optimally doped plcco has a resonance @xcite .
similar to hole - doped materials @xcite , the resonance in electron - doped plcco increases in intensity below @xmath4 and its energy scales with @xmath163 forming a universal plot for all superconducting copper oxides irrespective of electron- or hole - doping @xcite .
however , in contrast to hole - doped materials , magnetic excitations below the resonance in electron - doped plcco form commensurate and gapless scattering ( fig . 8)
therefore , the hour - glass shaped dispersion in the magnetic scattering of hole - doped superconducting materials @xcite may not be a universal feature of all high-@xmath4 superconductors . instead , the resonance itself appears to be a universal property of the superconducting copper oxides .
scaling in the dynamic susceptibility of plcco with @xmath28 k and @xmath164 k. a ) the results from a minimization of @xmath165 for all @xmath166 data with 80 k @xmath167 30 k and for data with @xmath168 k and @xmath169 mev .
the bin size of 0.5 along the log(@xmath170 scale is the only assumption made in obtaining @xmath171 .
b ) the log plot of @xmath172 as a function of @xmath173 shows three distinct regions : 1 ) data with 80 k @xmath167 30 k and with @xmath174 30 k and @xmath169 mev are within the quantum critical scaling regime and collapse onto a universal curve .
2 ) the data at 118 k show a clear high temperature departure from this quantum critical scaling behavior .
3 ) the data with @xmath168 k and @xmath175 mev show a low temperature departure from the scaling regime .
the breakdown of scaling at 120 k could be due to uncertainties in determining the backgrounds .
c ) summary of the @xmath173 scaling regime .
d ) @xmath172 as a function of @xmath173 now plotted with only the data in the quantum critical scaling regime shown in panel ( c ) .
this provides a clearer picture of the universal collapse for data within the valid scaling regime . ]
the identification of the continuous suppression of @xmath24 as the optimal superconductivity is approached in plcco as a function of annealing process @xcite suggests the possibility of a magnetic qcp , regardless of the precise nature of the magnetic structure ( homogeneous or inhomogeneous ) . to fully establish the existence of such
a qcp requires the study of magnetic dynamics , which we have shown above using inelastic neutron scattering .
we focus on two signatures of a qcp .
first , a qcp is accompanied @xcite by a quantum critical region at finite temperatures and finite energies bounded below by a scale which gradually goes to zero as the qcp is reached where the dynamics manifests the excitations of the qcp .
second , in this quantum critical regime , the dynamics are scale - invariant , a particular form of which is an @xmath173 scaling @xcite ; such an @xmath173 scaling has been systematically studied in heavy fermion metals @xcite in which the existence of a magnetic qcp is not in doubt @xcite . in the following , we will determine how our data can be described by the standard qcp theory . scaling in the dynamic susceptibility of plcco with @xmath13 k , @xmath176 k and @xmath10 k ( with no coexisting af order ) .
a ) a log plot of @xmath172 as a function of @xmath173 for the @xmath13 k sample over - plotted on the universal curve from fig . 9b . by normalizing the @xmath142 bragg intensities of both samples , we find @xmath177 which indicates that magnetic scattering at @xmath117 is considerably weaker in the @xmath13 k plcco .
this scaling plot reflects the modified scaling regimes in this @xmath13 k sample as discussed in the text .
b ) @xmath172 plotted as a function of @xmath173 for the @xmath10 k sample again over - plotted on the universal curve from fig .
c ) scaling relation now plotted for all three superconducting samples , @xmath28 k , @xmath13 k , and @xmath10 k , and their respective scaling regimes showing a universal collapse onto the curve from fig . 9b . ] to perform a scaling analysis for the @xmath28 k plcco , we note that the @xmath173 scaling can be written as @xmath178 , where the scaling exponent @xmath179 and the scaling function @xmath180 are determined through the best - observed collapse of the data onto one universal curve @xcite .
the quantum critical scaling is different from the critical scattering from classical af second order phase transition as the former is controlled by @xmath80 itself while the characteristic energy scale @xmath181 in the latter case is determined by reduced temperature @xmath182 ( @xmath183 ) @xcite . for the heisenberg antiferromagnet with @xmath164
k , @xmath181 should fall within the quasi - elastic part of the energy spectra and thus not contribute to the observed magnetic scattering in fig .
5 . by plotting all data above the 30 k energy scale ( @xmath184 mev ) as a function of @xmath173 and minimizing @xmath185 ( ref .
38 ) , we find @xmath186 independent of the functional form of @xmath180 ( fig .
using this value for the exponent leads to a scaling plot of @xmath187 as a function of @xmath173 ( fig .
such a plot also identifies the scaling regime , which is bounded at low energy ( 2 mev ) and temperatures ( fig .
9c ) . the data within the scaling regime in fig .
9c clearly show a collapse onto a universal curve as plotted in fig .
we fit the latter using @xmath188 and find @xmath189 .
this collapse of the data over more than two decades of @xmath173 strongly suggests the presence of universal dynamics .
if these universal dynamics indeed reflect a nearby qcp , the scaling behavior must break down at low energies and temperatures , with the cut - off scale determined by the distance to the qcp .
indeed , for small @xmath190 the @xmath173 scaling is observed only at temperatures above about 30 k ( fig .
the fact that this temperature scale is of the order of @xmath24 suggests that the cut - off to scaling is connected with the development of the af order . likewise , at the lowest measured temperature , @xmath89 k , the @xmath173 scaling is seen only at @xmath191 above about 2 mev ( fig .
9b ) , which , within the error bars , is equal to the low-@xmath190 cut - off temperature multiplied by the universal constant @xmath192 .
now turning to how this scaling regime evolves in samples tuned closer to optimal doping , the same type of scaling analysis can be performed for the low - energy magnetic spectra of the @xmath13 k plcco . since our neutron diffraction measurements at @xmath130 show a @xmath193 k ( fig .
2c ) , this sample should be much closer to the magnetic qcp than the @xmath28 k ( @xmath164 k ) plcco .
figure 10a shows the summary of the @xmath187 of the @xmath13 k plcco over - plotted on the universal scaling fit to the @xmath28 k sample ( dashed line ) . a scale factor , @xmath194 ,
was used to normalize the @xmath195 at @xmath117 for the @xmath13 k plcco to that of the @xmath28 k plcco , which reflects the fact that the @xmath14-width in the former is much broader ( figs .
7c - e ) . besides the overall scale factor
, we find that all data with @xmath196 mev and/or @xmath197 k fall on the universal curve , consistent with the @xmath13 k plcco being closer to the qcp point .
the @xmath173 scaling regime is seemingly correlated with the gradual suppression of the af order as optimal superconductivity is approached , indicating the presence of a magnetic quantum critical point in electron - doped superconductors . however , the dynamic in - plane spin correlation length as stated earlier is observed to decrease with increasing @xmath4 and decreasing @xmath32 , suggesting that this qcp in plcco can not arise from criticality towards a three - dimensional long - range af order at 0 k @xcite . instead , the data are similar to what happens in the li - doped la@xmath6cuo@xmath83 @xcite and in certain spin - glass qcp heavy fermion systems @xcite .
although there is at present no comprehensive theory for the expected properties of a metallic spin - glass qcp , the reduced magnetic transition temperature and the continuous nature of the transition ( fig .
2c ) imply that the scaling behavior in the @xmath13 k plcco should persist down to lower energies and temperatures than that of the @xmath28 k sample . continuing this analysis on the magnetic fluctuations observed in the @xmath10 k sample , an identical scaling plot is shown in fig .
a constant scale factor was again used to normalize the scaling fit from the @xmath28 k sample to overlay with the data from the @xmath10 k sample .
as the system is tuned closer to the qcp , there should now be larger ranges of energies and temperatures @xmath128 in which the scaling remains valid . indeed
, now all energies and temperatures ( for which a suitable nonmagnetic background was measured ) are shown to collapse on the same universal curve determined for the @xmath28 k sample ( dashed line in fig .
this encompasses the entire range of energy and temperature probed from @xmath198 mev and @xmath199 k with the exception of a notable divergence at the @xmath84 mev spin fluctuations measured at @xmath50 k in fig .
this divergence from the scaling fit for spin dynamics with @xmath200 mev may result from the system being tuned slightly beyond the qcp . in this case , the low energy / temperature spin dynamics crossover into the quantum disordered regime and no longer obey the scaling relation of quatum critical excitations . since the system does not exhibit af order to at least 600 mk
, we can not determine how close the sample is to the @xmath201 k qcp .
nevertheless , the overall trend for the three superconducting plcco investigated is that of increasing regimes of validity for the dynamic @xmath173 scaling as the system is tuned closer to the qcp itself .
finally , the valid scaling regimes for all three sc plcco samples are overplotted in fig .
10c showing a universal collapse onto a common function , with the exception of the lowest energy excitations at @xmath50 k for the @xmath10 k sample as discussed earlier . the universal collapse signifying @xmath173 scaling and the systematic evolution of scaling regime as af order
is suppressed with annealing strongly suggest the presence of a magnetic qcp in the electron - doped plcco .
recent transport and optical measurements on electron - doped pr@xmath17ce@xmath202cuo@xmath203 ( pcco ) have identified singular behaviors , which appear compatible with the influence of a qcp @xcite .
our results suggest that such a qcp may have a magnetic origin , possibly due to the presence of a spin - glass qcp .
we note that previous work has shown the presence of a spin - glass qcp in the hole - doped cuprates @xcite , thus suggesting that this magnetic qcp might be a common feature in high-@xmath4 superconductors regardless of doped - carrier type .
we have systematically measured the doping evolution of the low energy spin fluctuations in the electron - doped cuprate , plcco , as the system is tuned from an as - grown nsc antiferromagnet into a phase - pure optimally doped superconductor .
the as - grown , semiconducting plcco system exhibits gapped low energy spin fluctuations consistent with those observed in the parent compound pr@xmath6cuo@xmath83 . tuning the system into an optimally doped superconductor changes these low energy spin excitations dramatically . instead of a gapped spin wave spectrum , the superconducting plcco samples exhibit gapless low energy spin dynamics which exhibit two differing regimes of response in @xmath128 : the first of these coupling to the onset of the af order in the system and the second appearing as spin dynamics whose scattering is weakly temperature dependent over large temperature ranges .
the regime that couples to the af order in these samples is seen to evolve from @xmath204 k and @xmath205 mev in the @xmath28 k sample to @xmath206 k and @xmath207 mev in the @xmath13 k plcco . at optimal doping ( @xmath10 k ) , where the static af phase is suppressed to below 600 mk , the observed magnetic scattering for @xmath208 k or @xmath209 mev is @xmath80-independent , suggestive of the possible influence of quantum critical fluctuations in the system .
subsequent scaling analysis of the data revealed a collapse of appropriate regions of @xmath128 onto a universal curve for all three superconducting plcco systems studied , thereby providing microscopic evidence for a qcp in the electron doped plcco .
additionally , the first system discussed , @xmath28 k plcco , displays a peak in the low-@xmath80 susceptibility centered at @xmath210 mev which vanishes upon warming , whereas systems tuned closer to optimal doping ( @xmath13 k and @xmath10 k ) display only continuously decreasing magnetic response with increasing energy transfer .
our experiments described here have provided a systematic investigation into the evolution of the low energy spin dynamics in an electron - doped copper oxide , plcco , thereby providing valuable constraints on microscopic theories seeking to model the spin excitations as the materials evolve from a long - range ordered antiferromagnet into an optimally doped superconductor .
we are grateful to hong ding , j. w. lynn , h. a. mook , tai kai ng , and f. c. zhang for helpful discussions .
the neutron scattering part of this work is supported by the u.s .
nsf dmr-0453804 .
the plcco sample growth is supported in part by u.s .
doe de - fg02 - 05er46202 .
ornl is supported by de - ac05 - 00or22725 with ut / battelle llc .
lanl is supported by u.s .
spins is supported by the u.s .
nsf through grant no .
the part of the work done in japan was supported by the grant - in - aid for science provided by the japan society for the promotion of science .
is supported by nsf dmr-0424125 and the robert a. welch foundation .
s. m. hayden , g. aeppli , h. a. mook , t. g. perring , t. e. mason , s .- w .
cheong , and z. fisk , phys .
lett . * 76 * , 1344 ( 1996 ) .
p. bourges , h. casalta , a. s. ivanov , and d. petitgrand , phys .
lett . * 79 * , 4906 ( 1997 ) .
r. coldea , s .
m. hayden , g. aeppli , t. g. perring , c. d. frost , t. e. mason , s .- w cheong , and z. fisk , phys .
lett . * 86 * , 5377 ( 2001 ) .
d. j. scalapino , science * 284 * , 1282 ( 1999 ) .
pengcheng dai , h. a. mook , s. m. hayden , g. aeppli , t. g. perring , r. d. hunt , and f. do@xmath211an , science * 284 * 1344 ( 1999 ) .
hyungje woo , pengcheng dai , s. m. hayden , h. a. mook , t. dahm , d. j. scalapino , t. g. perring , f. do@xmath211an , cond - mat/0608280 ( nat . phys . in press ) .
s. m. hayden , h. a. mook , pengcheng dai , t. g. perring , and f. do@xmath211an , nature ( london ) * 429 * , 531 ( 2004 ) .
j. m. tranquada , h. woo , t. g. perring , h. goka , g. d. cu , g. xu , m. fujita , and k. yamada , nature ( london ) * 429 * , 534 ( 2004 ) . n. b. christensen , d. f. mcmorrow , h. m. ronnow , b. lake , s. m. hayden , g. aeppli , t. g. perring , m. mangkorntong , m. nohara , and h. takagi , phys . rev . lett . * 93 * , 147002 ( 2004 ) .
john m. tranquada , cond - mat/0512115 ( 2005 ) . k. yamada , k. kurahashi , t. uefuji , m. fujita , s. park , s .- h .
lee , and y. endoh , phys .
lett . * 90 * , 137004 ( 2003 ) .
stephen d. wilson , pengcheng dai , shiliang li , songxue chi , h. j. kang , and j. w. lynn , nature ( london ) * 442 * , 59 ( 2006 ) .
stephen d. wilson , shiliang li , hyungje woo , pengcheng dai , h. a. mook , c. d. frost , seiki komiya , and yoichi ando , phys .
lett . * 96 * , 157001 ( 2006 ) .
k. yamada , c. h. lee , k. kurahashi , j. wada , s. wakimoto , s. ueki , h. kimura , y. endoh , s. hosoya , g. shirane , r. j. birgeneau , m. greven , m. a. kastner , and y. j. kim , phys .
b * 57 * , 6165 ( 1998 ) .
pengcheng dai , h. a. mook , r. d. hunt , and f. do@xmath211an , phys .
b * 63 * , 054525 ( 2001 ) .
a. t. boothroyd , s. m. doyle , d. mck .
paul , and r. osborn , phys .
b * 45 * , 10075 ( 1992 ) .
i. w. sumarlin , j. w. lynn , t. chattopadhyay , s. n. barilo , d. i. zhigunov , and j. l. peng , phys .
b * 51 * , 5824 ( 1995 ) .
m. fujita , t. kubo , s. kuroshima , t. uefuji , k. kawashima , k. yamada , i watanabe , and k. nagamine , phys .
b * 67 * , 014514 ( 2003 ) .
pengcheng dai , h. j. kang , h. a. mook , m. matsuura , j. w. lynn , y. kurita , s. komiya , and y. ando , phys .
b * 71 * , 100502(r ) ( 2005 ) . h. j. kang , pengcheng dai , h. a. mook , d. n. argyriou , v. sikolenko , j. w. lynn , y. kurita , s. komiya , and y. ando , phys .
b * 71 * , 214512 ( 2005 ) .
t. uefuji , t .
kubo , k. yamada , m. fujita , k. kurahashi , i. watanabe , and k. nagamine , physica c * 357 - 360 * , 208 ( 2001 ) . h. j. kang , pengcheng dai , j. w. lynn , m. matsuura , j. r. thompson , shou cheng zhang , d. n. argyriou , y. onose , and y. tokura , nature ( london ) * 423 * , 522 ( 2003 ) . m. matsuura , pengcheng dai , h. j. kang , j. w. lynn , d. n. argyriou , k. prokes , y. onose , and y. tokura , phys . rev .
b * 68 * , 144503 ( 2003 ) .
y. tokura , h. takagi , and s. uchida , nature ( london ) * 337 * , 345 ( 1989 ) .
a. schroder , g. aeppli , r. coldea , m. adams , o. stockert , h. v. lohneysen , e. bucher , r. ramazashvili , p. coleman , nature ( london ) * 407 * , 351 ( 2000 ) .
m. c. aronson , r. osborn , r. a. robinson , j. w. lynn , r. chau , c. l. seaman , and m. b. maple , phys .
lett . * 87 * , 197205 ( 2001 ) .
stephen d. wilson , pengcheng dai , d. t. adroja , s .- h .
lee , j .- h .
chung , j. w. lynn , n. p. butch , and m. b. maple , phys .
lett . * 94 * , 056402 ( 2005 ) .
g. r. stewart , rev .
* 73 * , 797 ( 2001 ) .
a. n. lavrov , h. j. kang , y. kurita , t. suzuki , s. komiya , j. w. lynn , s .- h .
lee , pengcheng dai , and y. ando , phys .
. lett . * 92 * , 227003 ( 2004 ) .
d. petitgrand , s. v. maleyev , ph .
bourges , and a. s. ivanov , phys .
b * 59 * , 1079 ( 1999 ) .
j. a. hertz , phys .
b * 14 * , 1165 ( 1976 ) .
sudip chakravarty , b. i. halperin , and d. r. nelson , phys .
b. * 39 * , 2344 ( 1989 ) .
s. sachdev and j. w. ye , phys .
lett . * 69 * , 2411 ( 1992 ) .
a. j. millis , phys .
b * 48 * , 7183 ( 1993 ) . c. m. varma , phys . rev .
lett . * 83 * , 3538 ( 1999 ) .
a. abanov and a. chubukov , phys .
rev . lett .
* 84 * , 5608 ( 2000 ) .
qimiao si , s. robello , k. ingersent , j. lleweilun smith , nature ( london ) * 413 * , 804 ( 2001 ) .
s. sachdev , rev .
phys . * 75 * , 913 ( 2003 ) .
w. bao , y. chen , y. qiu , and j. l. sarrao , phys .
lett . * 91 * , 127005 ( 2003 ) . m. f. collins , magnetic critical scattering ( oxford university press , oxford , 1989 ) page 46 .
y. dagan , m. m. qazilbash , c. p. hill , v. n. kulkarni , and r. l. greene , phys .
lett . * 92 * , 167001 ( 2004 ) .
a. zimmers , j. m. tomczak , r. p. s. m. lobo , n. bontemps , c. p. hill , m. c. barr , y. dagan , r. l. greene , a. j. millis , and c. c. homes , europhys . lett . * 70 * , 225 ( 2005 ) . c. panagopoulos , j. l. tallon , b. d. rainford , t. xiang , j. r. cooper , and c. a. scott , phys .
b * 66 * , 064501 ( 2002 ) . while the scattering at @xmath212 and @xmath213 ( depending on the system ) may include contributions from multiphonon and incoherent magnetic scattering , it does not have correlated magnetic scattering contribution from @xmath214 and thus can be regarded as background scattering . | we use inelastic neutron scattering to explore the evolution of the low energy spin dynamics in the electron - doped cuprate pr@xmath0lace@xmath1cuo@xmath2 ( plcco ) as the system is tuned from its nonsuperconducting , as - grown antiferromagnetic ( af ) state into an optimally - doped superconductor ( @xmath3 k ) without static af order .
the low temperature , low energy response of the spin excitations in under - doped samples is coupled to the presence of the af phase , whereas the low - energy magnetic response for samples near optimal @xmath4 exhibits spin fluctuations surprisingly insensitive to the sample temperature .
this evolution of the low energy excitations is consistent with the influence of a quantum critical point in the phase diagram of plcco associated with the suppression of the static af order .
we carried out scaling analysis of the data and discuss the influence of quantum critical dynamics in the observed excitation spectrum . |
this paper is devoted to the systematic studies of the asymptotic solutions of advection equation with time - oscillating velocity field .
we employ the two - timing method in the form introduced by @xcite , which allows us to produce the classification of an infinite number of distinguished limits and solutions .
we have obtained several new versions of averaged equations for a passive scalar and have presented several examples .
the main purpose of this paper is the introduction of a new systematic and general viewpoint on the hierarchy of distinguished limits , drifts , and pseudo - diffusion .
this viewpoint can help to organize and to unify a number of results , including that of @xcite .
we have chosen the advection equation as the simplest interesting example , with the hope that similar studies can be performed for other interesting equations .
the two - timing method has been used by many authors , see @xcite , however our analysis goes well beyond the usual calculations of the main approximations in various special cases .
our analytic calculations are rigorous and straightforward by their nature , however they include a large number of integration by parts and algebraic transformations ; the performing of such calculations represents a ` champion - type ' result by itself .
the details of calculations are fully presented in the appendix .
the key point of asymptotic theory for odes or pdes ( with two or more small parameters ) is the choice of asymptotic path in the space of these parameters . in other words ,
all available small parameters should be expressed through a single one ; it allows one to compare different terms in asymptotic series .
the related literature is very diverse and short of exact mathematical formulations .
usually , during the studies of a particular problem , some ` lucky ' asymptotic path is chosen on the basis that it gives the successful derivation of an averaged equation or successful calculation of the main term in asymptotic solution , see _
e.g. _ @xcite .
such ` lucky ' asymptotic paths are called _ distinguished limits _ ( dl s ) , while related solutions are called _ distinguished limit solutions _ ( dls s ) .
any studies of the multiplicity of distinguished paths ( as well as the multiplicity of averaged equations and related solutions ) and interrelations between them are always avoided .
a different approach is given by @xcite , where the concept of ` strong ' , ` moderate ' , and ` weak ' oscillations is introduced .
however , this approach overlooks some asymptotic solutions and could be seen as being ` too rigid ' ; therefore we propose its development in this paper .
our asymptotic procedure is based on the simultaneous scaling ( with the use of strouhal number @xmath0 ) a slow time - scale and vibrational velocity amplitude .
this procedure is flexible , self - consistent , covers broad classes of asymptotic solutions , and leads to a natural classification of solutions .
the considered advection equation is a hyperbolic pde of the first order with time - oscillating coefficient , which represents the _ prescribed _ velocity field .
we emphasize that the velocity field has no connection to any particular dynamics . indeed
, the governing equations for a velocity can be chosen very differently : they could represent an inviscid or viscous fluid , a fluid with any rheology , liquid crystals , blood , elastic and plastic medium , _ etc .
_ the advection equation can be also a part of the collisionless boltzmann equation or vlasov equation ( see _ e.g. _ @xcite ) .
the introduction of dimensional variables into the advection equation and the use of the two - timing assumption leads us to the problem with _ two independent small scaling parameters that represent the ratio of two time - scales and amplitude of oscillations_. the given set of parameters produces the _ strouhal number @xmath0 as a basic large dimensionless parameter _ ; the appearance of @xmath0 means that the scaling is not unique . in agrement with the general _ ansatz _ of asymptotic theory we can use @xmath0 to vary all the characteristic scales ; in our approach we use @xmath0 to introduce the appropriate scales of velocity and slow time .
the obtained dimensionless equation has been used for the identifying of distinguished limits .
we have found the infinite sequence of the distinguished limits , which are corresponding to the successive degeneration of a drift velocity .
the _ order _ of this degeneration is chosen to enumerate the distinguished limits .
the derivation of averaged equations produce the successive approximations for a drift velocity as well as a qualitatively new ` diffusion - like ' term , which we call _ pseudo - diffusion_. remarkably , all the coefficients in the obtained averaged equations are universal ; they are the same in dls s of different orders and only ` moving ' to lower approximations with the increasing of the order of dls . as the next important step we discover _ one - parametric _ families of solutions ( infinite number of solutions corresponding to each dls ) ,
different parametric solutions contain different slow time - scales and different velocity amplitudes .
our results are particularly striking for the case of purely periodic velocity oscillations . here
we address an intriguing question : how to find the related slow time - scale while it does not appear in the coefficients of the equation ?
our answer is : the slow time - scale is uniquely determined by the magnitude of a given velocity in terms of @xmath0 . for a given velocity
we obtain the unique slow time - scale .
we have established that the solutions corresponding to different orders of velocity are physically different even if they have the same functional form .
the central notion which appears in this paper is a _
drift velocity_. our study can be seen as a general view on the studies of drift velocities with the emphasis at their different appearances in averaged equations ; it is complemented by an unexpected but inevitable appearance of pseudo - diffusion .
we put the studies of drifts into the general context of the asymptotic theory , that has never been done before .
in addition , there is a difference between the _ classical drift _ and _ eulerian drift_. the classical ( or lagrangian ) drift appears as an averaged velocity for fixed lagrangian coordinates ; hence the classical drift represents an average velocity of material particles .
the classical drift , in its main approximation , is a well - studied phenomenon , see @xcite . in our consideration ( as well as in @xcite )
the average operation takes place for _ fixed eulerian coordinates _ , hence our drift represents an average velocity at a given eulerian position and we call it an _ eulerian drift_. the relation between the classical drift and eulerian drift represents an open question to study , while their leading terms ( in the amplitude series ) are certainly coincide . in our consideration of eulerian drift
we explicitly show its different options , approximations , and appearances .
we calculate first three approximations for the eulerian drift velocity that is especially useful in the cases of vanishing leading approximation .
situation with _ pseudo - diffusion _ is more intriguing .
first of all , the appearance of this ` diffusion - like ' term in averaged equations is surprising by itself .
another surprise is the form of the pseudo - diffusion matrix which appears as a _ lie - derivative _ of an averaged _ quadratic displacement tensor_. for different flows , pseudo - diffusion can correspond to diffusion or ` anti - diffusion ' or to some intermediate cases .
we make the first step in the understanding of the meaning of pseudo - diffusion by considering some simple example of rigid - body oscillations .
we arrive to a surprising conclusion that the combination of the advection due to rigid - body oscillations and the eulerian averaging procedure can produce the evolution of the averaged field , which formally coincides with physical diffusion .
we present five examples of flows aimed to illustrate different appearances of drift and pseudo - diffusion .
these examples cover the general form of modulated oscillatory fields , stokes drift , a spherical ` acoustic ' wave , a class of velocity fields with vanishing main term of the drift , and a velocity field which provides a chaotic averaged ` lagrangian ' dynamics .
it is important , that in this paper we do not use any additional assumptions , except the two - timing hypothesis itself .
all the results appear from the straightforward and rigorous calculations and represent mathematical theorems , although they are not formulated explicitly in this way .
hence , our results can be seen as the test on the validity and reliability of two - timing method . any failure to interpret our results physically or mathematically
can be explained only by the insufficiency of two time - scales , that might mean that three time - scales ( or more ) are required .
at the same time , following to a high mathematical rigour , we avoid any physical interpretations , while such interpretations could be needed in future .
a fluid flow is given by its velocity field @xmath1 , where @xmath2 and @xmath3 are cartesian coordinates and time , asterisks stand for dimensional variables .
we suppose that velocity field is sufficiently smooth , but _
we do not assume that it satisfies any equations of motion_. the dimensional advection equation for a scalar field @xmath4 is @xmath5 this equation describes the motions of a lagrangian marker in either an incompressible or compressible fluid , and the advection of a passive scalar admixture with concentration @xmath4 in an incompressible fluid .
it is also closely relevant to various models , not related to fluid dynamics .
the hyperbolic equation ( [ exact-1 ] ) has characteristics curves ( trajectories ) @xmath6 described by an ode @xmath7 where @xmath8 and @xmath9 are eulerian and lagrangian coordinates .
the _ classical drift _ motion follows after the averaging of ( [ exact-1a ] ) , which is automatically performed for fixed lagrangian coordinates .
in contrast , in this paper we consider only the equation ( [ exact-1 ] ) , which is subject of averaging for fixed eulerian coordinates , the related drift motion can be called _ eulerian drift . _ such an approach significantly simplifies calculations and is reacher on results , however the link between the classical ( lagrangian ) drift and the considered below eulerian drift emerges as a new problem to study .
we accept that the field @xmath1 is oscillating in time and possesses the characteristic scales of velocity @xmath10 , length @xmath11 , and frequency @xmath12 .
these three parameters give a dimensionless parameter
strouhal number @xmath13 , hence the dimensionless variables and parameters ( written below without asterisks ) _ are not unique _ ; in this paper we use the following set @xmath14 while @xmath15 has the ` two - timing ' functional form @xmath16 with constants @xmath17 we note that the scale of velocity is @xmath18 , not just @xmath10 ; such kind of ` combined ' scaling is usual for asymptotic analysis and for fluid dynamics ( for example , for viscous flows the length - scale can be chosen as @xmath19 , where @xmath20 is reynolds number ) .
in fact , the accepted scaling for a velocity is absolutely required for the further consideration , otherwise ( say , for @xmath21 ) the asymptotic solutions for some important cases do not exist ( see discussion ) .
the appearance of two constants @xmath22 and @xmath23 reflects the fact that we have two small parameters in the equation .
different options for the values of @xmath22 and @xmath23 will be introduced later .
physically , the restriction @xmath24 makes the variable @xmath25 ` slow ' in comparison with @xmath26 , while @xmath27 gives the smallness of a vibrational spatial amplitude . in the accepted dimensionless variables ( after the use of the chain rule ) our main equation ( [ exact-1 ] )
takes the form @xmath28 where subscripts @xmath26 and @xmath25 stand for the related partial derivatives ; @xmath25 and @xmath26 represent two _ mutually dependent _ variables , which are called _ slow time _ and _ fast time_. equation ( [ exact-6 ] ) can be rewritten in the form containing two independent small parameters @xmath29 hence , we study the asymptotic limit in the plane @xmath30 @xmath31 where different asymptotic paths can be prescribed by different choices of @xmath22 and @xmath23 in .
each such a path may produce different solutions ; the paths which produce valid asymptotic solutions are called _ distinguished limits_.
the key suggestion of the two - timing method is @xmath32 as a result , we convert , from pde with independent variables @xmath33 and @xmath34 ( where @xmath26 and @xmath25 are mutually dependent and expressed in terms of @xmath33 ) into pde with the extended number of independent variables @xmath35 and @xmath34 . then the solutions of , must have a functional form : @xmath36 it should be emphasized , that without the functional form of solutions can be different from ; indeed the presence of @xmath0 allows us to build an infinite number of different time - scales , not just @xmath26 and @xmath25 . in this paper
we accept and analyse the related averaged equations and solutions in the functional form . in order to make further analytical progress
, we introduce a few convenient notations and agreements . in this paper
we assume that _ any dimensionless function _
@xmath37 has the following properties : \(i ) @xmath38 and all its @xmath34- , @xmath25- , and @xmath26-derivatives of the required below orders are also @xmath39 ; \(ii ) @xmath40 is @xmath41-periodic in @xmath26 , _
i.e. _ @xmath42 ( about this technical simplification see discussion ) ; \(iii ) @xmath40 has an average given by @xmath43 \(iv ) @xmath40 can be split into averaged and purely oscillating parts @xmath44 where _ tilde - functions _ ( or purely oscillating functions ) are such that @xmath45 and the _ bar - functions _
@xmath46 ( or the averaged functions @xmath47 ) are @xmath26-independent ; \(v ) we introduce a special notation @xmath48 ( with a superscript @xmath26 ) for the _ tilde - integration _ of tilde - functions , such integration keeps the result in the tilde - class .
we notice that the integral of a tilde - function @xmath49 often does not belong to the tilde - class . in order to keep the result of integration in the tilde - class
we should subtract the average @xmath50 , which can be rewritten as @xmath51
our calculations have shown that there is a series of _ distinguished limits _ for the equation ( [ exact-6a ] ) with independent variables @xmath52 ; each distinguished limit represents the one - parametric path @xmath53 , @xmath54 ( with an integer @xmath55 ) in the plane @xmath56 ; hence ( [ exact-6a ] ) yields : @xmath57 we denote the distinguished limits as dl(@xmath58 ) ; our calculations have been performed for @xmath59 .
the detailed calculations for the most interesting case @xmath60 are given in appendix .
other cases @xmath61 are very similar and contain the same coefficients of the averaged equations and the same blocks of calculations as the case @xmath60 ; therefore for @xmath61 we present below some final results only . in all cases
we are looking for solutions in the form of regular series @xmath62 analytical calculations contain the following two steps : ( i ) writing the equations of successive approximations and splitting each such equation into its ` bar ' and ` tilde ' parts , see ; ( ii ) obtaining the closed systems of equations for the ` bar ' parts ; during this step one should perform a large number of integrations by parts and algebraic transformations .
the results give full solutions @xmath63 in any considered approximation .
the described above steps can be performed only for distinguished limits ; all other asymptotic paths produce controversial systems of equations or they are not leading to closed systems of equations for successive approximations .
the first four dl s correspond to the variables @xmath25 and velocity @xmath64 as following : dl(1 ) : @xmath65 ; @xmath66 , where both ` bar ' and ` tilde ' parts of @xmath64 are not zero ; dl(2 ) : @xmath67 ; @xmath68 , where @xmath69 ; dl(3 ) : @xmath70 ; @xmath71 , where @xmath69 , and @xmath72 ; dl(4 ) : @xmath73 ; @xmath71 , where @xmath69 , @xmath72 , and @xmath74 .
the accepted notations are @xmath75,\quad \overline{{\boldsymbol{v}}}_1\equiv\frac{1}{3}\langle[[\widetilde{{\boldsymbol{u}}},\widetilde{{\boldsymbol{\xi}}}],\widetilde{{\boldsymbol{\xi}}}]\rangle,\quad \widetilde{{\boldsymbol{\xi}}}\equiv\widetilde{{\boldsymbol{u}}}^\tau ; \label{4.18}\end{aligned}\ ] ] where the square brackets stand for a commutator of any two vector - functions @xmath76 and @xmath77 @xmath78\equiv({\boldsymbol{g}}\cdot\nabla){\boldsymbol{f}}-({\boldsymbol{f}}\cdot\nabla){\boldsymbol{g}}\ ] ] from the above list , one can see that the increasing of @xmath58 corresponds to the successive degenerations of velocity @xmath64 and drift velocity @xmath79 .
the case dl(1 ) with @xmath80 corresponds to the advection speed of order one , hence @xmath81 and the averaged equation of zero approximation is @xmath82 one can see that in the main approximation we have the advection of an averaged scalar field with the averaged velocity .
however , in this paper we concentrate our attention to the cases when the oscillatory part of a velocity enters the averaged equations of the main approximation .
therefore , we do not consider dl(1 ) in detail here , we just state that all the coefficients in the averaged equations are similar to that in dl(2 ) ( see below ) ; more exact : the same coefficients appear in the dl(1 ) averaged equations of the next approximation in @xmath83 ( in comparison with dl(2 ) ) .
dl(2 ) is the most instructive and physically interesting case , therefore we consider it in detail . here
, we have @xmath84 , hence , the speed of advection is described by a longer ( than in dl(1 ) ) slow time - scale @xmath67 .
the averaged equations of the first three successive approximations are ( see the detailed derivation in appendix ) @xmath85 with notations @xmath86,\widetilde{{\boldsymbol{\xi}}}]\rangle + \frac{1}{2}\langle[\widetilde{{\boldsymbol{v}}}_0,\widetilde{{\boldsymbol{v}}}_0^\tau]\rangle+ \frac{1}{2}\langle[\widetilde{{\boldsymbol{\xi}}},\widetilde{{\boldsymbol{\xi}}}_s]\rangle + \frac{1}{2}\langle\widetilde{{\boldsymbol{\xi}}}{\mathrm{div}\,}\widetilde{{\boldsymbol{u}}}'+ \widetilde{{\boldsymbol{u}}}'{\mathrm{div}\,}\widetilde{{\boldsymbol{\xi}}}\rangle,\label{4.19 } \\ & & \widetilde{{\boldsymbol{u}}}'\equiv\widetilde{{\boldsymbol{\xi}}}_s-[\overline{{\boldsymbol{v}}}_0,\widetilde{{\boldsymbol{\xi } } } ] , \label{4.20a}\\ & & 2\overline{\chi}_{ik}\equiv\langle\widetilde{u'}_i\widetilde{\xi}_k+\widetilde{u'}_k\widetilde{\xi}_i\rangle= \mathfrak{l}_{\overline{{\boldsymbol{v}}}_0}\langle\widetilde{\xi}_i\widetilde{\xi}_k\rangle,\label{4.20}\\ & & \mathfrak{l}_{\overline{{\boldsymbol{v}}}_0}\overline{f}_{ik}\equiv \left(\partial_s+ \overline{{\boldsymbol{v}}}_0\cdot\nabla\right)\overline{f}_{ik}-\frac{\partial\overline{v}_{0k}}{\partial x_m}\overline{f}_{im}- \frac{\partial\overline{v}_{0i}}{\partial x_m}\overline{f}_{km } \label{4.20b}\end{aligned}\ ] ] where the operator @xmath87 is such that @xmath88 represents the condition for tensorial field @xmath89 to be ` frozen ' into @xmath90 ( @xmath87 is also known as the _ lie derivative _ of a vector field ) .
the summation convention is in use everywhere in this paper
. three equations ( [ 4.15])-([4.17 ] ) can be written as a single advection-`diffusion ' equation ( with an error @xmath91 ) @xmath92 eqn .
( [ 4.21 ] ) shows that the averaged motion represents a drift with velocity @xmath93 and _ pseudo - diffusion _ with matrix coefficients @xmath94 . for dl(3 )
we impose a restriction @xmath72 and ( similarly to the calculations in appendix ) derive the equations : @xmath95,\widetilde{{\boldsymbol{\xi}}}]\rangle + \frac{1}{2}\langle[\widetilde{{\boldsymbol{v}}}_0,\widetilde{{\boldsymbol{v}}}_0^\tau]\rangle+ \frac{1}{2}\langle[\widetilde{{\boldsymbol{\xi}}},\widetilde{{\boldsymbol{\xi}}}_s]\rangle + \frac{1}{2}\partial_s\langle\widetilde{{\boldsymbol{\xi}}}{\mathrm{div}\,}\widetilde{{\boldsymbol{\xi}}}\rangle,\label{4.19a}\\ & & 2\overline{\chi}_{ik}= \partial_s\langle\widetilde{\xi}_i\widetilde{\xi}_k\rangle\label{4.20bb}\end{aligned}\ ] ] in fact , one can see that the equations - are very similar to that of dl(2 ) ( [ 4.15])-([4.17 ] ) ; the difference is : the same coefficients as in dl(2 ) appear in the previous approximations of dl(3 ) . for dl(4 ) we impose two restrictions @xmath72 and @xmath74 and derive ( by a similar procedure ) the equation @xmath96 with the same @xmath97 and @xmath98 as in dl(3 ) . in general , the comparison between the averaged equations for dl(1)-(4 ) shows that the same coefficients in dl(@xmath58 ) appear in the equations of the next order ( in @xmath83 ) in dl(@xmath99 ) . the higher approximations
dl(5 ) , _ etc .
_ can be derived similarly , however the calculations become too cumbersome .
one can see , that , in all presented above distinguished limit equations , the meaning of key _
parameter @xmath83 is not uniquely defined in terms of @xmath100 or @xmath0_. in dl(2 ) @xmath53 , @xmath101 in ( [ exact-6a ] ) , which imposes the link @xmath22 , @xmath23 and @xmath83 : @xmath102 hence , each distinguished limit solution of ( [ exact-6a ] ) , obtained in terms of @xmath83 , produces an infinite number of _ parametric solutions _
, one solution for any real number @xmath27 ( or for any real number @xmath24 ) .
those solutions are mathematically similar but physically different , since they correspond to different magnitudes of given velocities @xmath103 and different slow - time variables @xmath104 .
there are two ways of choosing @xmath22 and @xmath23 .
given @xmath22 : when the slow time - scale @xmath25 is given in the prescribed function @xmath105 , then it defines @xmath22 ; then @xmath23 follows from .
given @xmath23 : alternatively , when the velocity amplitude of @xmath106 is given , then it defines @xmath23 and we have to calculate @xmath22 from .
some interesting sets of @xmath22 and @xmath23 are : \(i ) if @xmath107 is given as the function of variables @xmath108 and @xmath81 , then @xmath109 , @xmath110 ; hence the velocity @xmath111 ( [ variables ] ) , and in @xmath112 .
\(ii ) the most frequently considered case is @xmath113 , then @xmath114 , @xmath115 , @xmath116 , and @xmath117 .
\(iii ) rather exotic possibility corresponds to the case @xmath118 . here ,
@xmath119\varepsilon$ ] , @xmath120 and velocity @xmath121\omega)$ ] .
such a scaling is required , for example , if a particular slow time - scale @xmath119\omega$ ] is prescribed in @xmath107 .
\(iv ) the above results for dl(2 ) are particularly striking for the case of @xmath122 , when the given velocity does not contain any slow time - scale @xmath25 and represents purely periodic velocity oscillations .
an intricate question arises : how to choose any particular scale of @xmath25 ?
our answer is : in this case the slow time - scale is uniquely determined by the magnitude @xmath123 of @xmath124 . for every given value of @xmath23 we obtain the related slow time variable @xmath125 .
it should be accepted that solutions with different @xmath23 are physically different since they correspond to different orders of a prescribed velocity field and different slow time - scales @xmath25 .
\(v ) the functional dependence between @xmath22 and @xmath23 is physically natural : the increasing of the amplitude of a given velocity ( increasing @xmath23 ) leads to the decreasing of the slow time - scale ( decreasing @xmath22 ) .
in general , the transformations similar to ( [ main - eq111 ] ) produce an infinite number of _ parametric solutions _ for each case dl(1 - 4 ) .
the inevitable appearance of the _ pseudo - diffusion _ terms in our averaged equations does represent an intriguing result .
we have introduced the term _ pseudo - diffusion _ on the grounds of the following arguments : \(i ) the evolution of @xmath126 is described by an advection - pseudo - diffusion - type equation ( [ 4.21 ] ) , where the matrix of pseudo - diffusion coefficients @xmath98 can be positive , negative or sign - indefinite ; it also can change its sign in space and time ; ( ii ) the equation ( [ 4.21 ] ) is valid only for the regular asymptotic expansions ( [ 4.22])-([4.22aa ] ) , while the classical diffusion is often affiliated with the singular expansions ; \(iii ) for the most interesting case dl(2 ) the pseudo - diffusion term represents a known ` source ' term in the second approximation ; for dl(4 ) the pseudo - diffusion does appear already in the leading approximation , although its evolution is described by ` very ' slow time @xmath127 ; \(iv ) the possibility of the secular growth ( in @xmath25 ) of averaged solutions due to pseudo - diffusion represents a central open question .
such secular terms could appear for the pseudo - diffusion matrix @xmath98 , which is constant or monotonically increasing in @xmath25 , but they are unlikely to appear if @xmath98 is oscillating in @xmath25 .
if such a growth does appear , then the sufficiency of two time - scales only should be reconsidered .
\(v ) the appearance of pseudo - diffusion matrix as a _ lie - derivative of an averaged quadratic displacement tensor _ , represents the result of straightforward calculations , which still requires qualitative understanding .
we propose the following example as a first step of such understanding . _ : _ the simple and instructive example of a prescribed velocity is a rigid - body oscillations of a medium , where @xmath128 . due to the vanishing of all spatial derivatives ,
all approximations of a drift velocity also vanish @xmath129 ( @xmath130 ) and the dl(4 ) equation takes a simplified form @xmath131 there are few observations for this equation : \(a ) when @xmath132 ( purely periodic oscillations of a velocity ) , then @xmath133 , which leads to a steady distribution @xmath134 ; hence , any purely periodic rigid - body oscillations do not affect the averaged field @xmath135 .
\(b ) in order to obtain @xmath136 ( with @xmath137 a constant tensor ) , one has to take @xmath138 , where @xmath139 is an arbitrary tilde - function ; we note that for positive matrix @xmath137 we have a standard diffusion equation , for negative @xmath137 an ` anti - diffusion ' equation , and for a sign - indefinite matrix a pseudo - diffusion equation .
\(c ) one can see , that for positive matrix @xmath137 the related evolution of @xmath135 is described by standard diffusion equation .
however , in a contrast to classical diffusion , the mechanism of this evolution is different : this evolution is due to the growth of a spatial oscillatory amplitude as @xmath140 ; hence , the described phenomenon ( which looks like diffusion ) actually represents an advection ( by the increasing in amplitude oscillations ) , which is ` superimposed ' with the average operation .
hence , we have arrived to the following striking result : the superposition of advection due to rigid - body oscillations and the eulerian averaging operation can lead to the evolution of an averaged solution , that is mathematically equivalent to the classical diffusion .
\(d ) for @xmath141 being a periodic function of @xmath25 ( which represents modulated oscillations ) , the sign of @xmath98 is periodically changing , mathematical properties of such pde are unknown , however one might expect that such an equation has bounded solutions ( secular growth is absent ) .
the case when the eigenvalues of @xmath98 have different signs , has not been systematically studied yet , see @xcite .
\(e ) for understanding of applicability of one should keep in mind that all our dimensionless functions and their derivatives are of order one , hence , the consideration of solutions with large gradients ( _ e.g. _ a point source solution ) represents an invalid operation .
\(f ) we emphasise , that for non - zero drift velocities all the equations , and contain additional spatial derivatives ( in comparison with ) . in particular ,
pseudo - diffusion matrix @xmath98 contains spatial derivatives ( as parts of the lie - derivative ) ; this intriguing appearance of pseudo - diffusion remains completely unexplained from a physical viewpoint .
all the above averaged equations have been derived for arbitrary function @xmath142 , therefore , the number of interesting examples can be significant . in this section
we consider only five instructive classes of @xmath142 . * _ example 1 .
_ _ the superposition of two modulated oscillatory fields : _ * @xmath143 where @xmath144 and @xmath145 are arbitrary bar - functions .
the straightforward calculations lead to @xmath146=[\overline{{\boldsymbol{p}}},\overline{{\boldsymbol{q } } } ] .
$ ] equations ( [ 4.18 ] ) , ( [ 4.19 ] ) , yield @xmath147,\quad \overline{{\boldsymbol{v}}}_1\equiv 0 , \label{example-1 - 4}\\ & & \overline{{\boldsymbol{v}}}_2=\frac{1}{8}\left([\overline{{\boldsymbol{p}}},\overline{{\boldsymbol{p}}}]+[\overline{{\boldsymbol{q}}},\overline{{\boldsymbol{q}}}]\right ) -\frac{1}{4}\left([\overline{{\boldsymbol{p}}}_s,\overline{{\boldsymbol{p}}}]+[\overline{{\boldsymbol{q}}}_s,\overline{{\boldsymbol{q}}}]\right)+\label{example-1 - 4a}\\ & & + \frac{1}{4}\left(\overline{{\boldsymbol{p}}}\,{\mathrm{div}\,}\overline{{\boldsymbol{p}}}'+\overline{{\boldsymbol{q}}}{\mathrm{div}\,}\overline{{\boldsymbol{q } } } ' + \overline{{\boldsymbol{p}}}'{\mathrm{div}\,}\overline{{\boldsymbol{p}}}+\overline{{\boldsymbol{q}}}'{\mathrm{div}\,}\overline{{\boldsymbol{q}}}\right),\nonumber\\ & & \overline{{\boldsymbol{p}}}\equiv[\overline{{\boldsymbol{v}}}_0,\overline{{\boldsymbol{p}}}],\quad \overline{{\boldsymbol{q}}}\equiv[\overline{{\boldsymbol{v}}}_0,\overline{{\boldsymbol{q}}}],\quad \overline{{\boldsymbol{p}}}'\equiv\overline{{\boldsymbol{p}}}_s-\overline{{\boldsymbol{p}}},\quad \overline{{\boldsymbol{q}}}'\equiv\overline{{\boldsymbol{q}}}_s-\overline{{\boldsymbol{q}}},\nonumber\\ & & \langle\widetilde{\xi}_i\widetilde{\xi}_k\rangle= \frac{1}{2}(\overline{p}_i \overline{p}_k + \overline{q}_i \overline{q}_k)\label{example-1 - 5}\end{aligned}\ ] ] the expression for the _ pseudo - diffusion _
matrix @xmath98 follows after the substitution of ( [ example-1 - 5 ] ) into ( [ 4.20 ] ) .
the expression @xmath107 ( [ example-1 - 1 ] ) is general enough to produce any given function @xmath90 .
indeed , in order to obtain @xmath148 and @xmath149 one has to solve the equation @xmath150=(\overline{{\boldsymbol{q}}}\cdot\nabla)\overline{{\boldsymbol{p}}}- ( \overline{{\boldsymbol{p}}}\cdot\nabla)\overline{{\boldsymbol{q}}}=2\overline{{\boldsymbol{v}}}_0({\boldsymbol{x}},s)\end{aligned}\ ] ] which represents an undetermined bi - linear pde - problem for two unknown functions @xmath148 and @xmath149 . * _ example 2 . _ _ stokes drift : _ * the dimensionless plane velocity field for an inviscid incompressible fluid is ( see @xcite ) @xmath151 where for our consideration one can take @xmath152 and @xmath153 , however we choose @xmath154 and @xmath155 as constants ; however , for our consideration they can be chosen as arbitrary functions of @xmath25 .
the fields @xmath156 , @xmath157 ( [ example-1 - 1 ] ) are @xmath158 the calculations of ( [ example-1 - 4 ] ) yield @xmath159 which represent the classical stokes drift and the first correction to it ( which vanishes ) .
for brevity , the explicit formula for @xmath160 is not given here .
further calculations show that @xmath161 one can see that eigenvalues @xmath162 and @xmath163 correspond to strongly anisotropic pseudo - diffusion . the averaged equation ( [ 4.21 ] ) ( with an error @xmath164 )
can be written as @xmath165 where @xmath166 and @xmath167 are @xmath168-components of corresponding velocities ( their @xmath169-components vanish ) .
this equation has an exact solution @xmath170 where @xmath171 is an arbitrary function , which is not affected by pseudo - diffusion .
* _ example 3 . _ _ a spherical ` acoustic ' wave : _ * the main term for velocity potential for an outgoing spherical acoustic wave is @xmath172 where @xmath173 , @xmath174 , and @xmath175 are an amplitude , a wavenumber , and a radius in a spherical coordinate system .
the velocity is purely radial and has a form ( [ example-1 - 1 ] ) @xmath176 where @xmath177 , and @xmath178 are radial components of corresponding vector - fields .
one can find that the fields @xmath179 and @xmath180 $ ] are also purely radial ; the radial component for the commutator is @xmath181 where @xmath182 is the radial component of @xmath179 and subscript @xmath175 stands for the radial derivative .
the drift ( [ example-1 - 4 ] ) is purely radial with @xmath183 it is interesting , that @xmath166 formally coincides with the velocity , caused by a point source in an incompressible fluid , and ( for small @xmath175 ) the value of @xmath167 dominates over @xmath166 , so the series is likely to be diverging at @xmath184 .
further calculations yield @xmath185 where @xmath186 stands for the only nonzero @xmath187-component of @xmath98 .
one can see , that in this case pseudo - diffusion appears as ordinary diffusion .
* _ example 4 .
_ _ @xmath188-drift : _ * it is interesting to consider such flows for which the classical expression for a drift vanishes : @xmath72 , but @xmath189 .
let the velocity field be a superposition of two standing waves of frequencies @xmath100 and @xmath190 : @xmath191=\frac{1}{2}[\overline{{\boldsymbol{p}}},\overline{{\boldsymbol{q}}}](2\cos\tau\sin 2\tau-\cos 2\tau\sin\tau)\label{example-5 - 3}\end{aligned}\ ] ] hence ( [ 4.18 ] ) yields @xmath192\rangle\equiv 0,\quad \overline{{\boldsymbol{v}}}_1=\frac{1}{3}\langle[[\widetilde{{\boldsymbol{u}}},\widetilde{{\boldsymbol{\xi}}}],\widetilde{{\boldsymbol{\xi}}}]\rangle= \frac{1}{8}[[\overline{{\boldsymbol{p}}},\overline{{\boldsymbol{q}}}],\overline{{\boldsymbol{p}}}]\end{aligned}\ ] ] these expressions produce infinitely many examples of the flows with @xmath188-drift .
at the same time shows that for a single standing wave the @xmath188-drift is absent .
* _ example 5 . _
_ chaotic averaged ` lagrangian ' dynamics for @xmath193 : _ * let the solenoidal / incompressible velocity ( [ example-1 - 1 ] ) be @xmath194 where @xmath195 are cartesian coordinates ; @xmath196 are constants .
either of these fields , taken separately , produces simple integrable dynamics of particles .
the calculations yield @xmath197 the computations of the averaged ` lagrangian ' dynamics for @xmath198 ( based on ( [ exact-1a ] ) for dl(2 ) ) were performed by prof .
a.b.morgulis ( private communications ) .
he has shown that this steady averaged flow exhibits chaotic dynamics of particles .
in particular , positive lyapunov exponents have been observed .
hence , the drift created by a simple oscillatory field can produce complex
_ averaged ` lagrangian ' dynamics_.
_ struggle for natural notations : _ the introducing of convenient and logical notations represents an important part of this paper .
this topic is always underestimated , however natural notations are especially helpful in the presence of cumbersome calculations . _ averaged and oscillatory parts of solutions : _ the analysis , presented in the paper , produces not only the averaged parts of solutions but also full solutions .
indeed , in appendix one can find the oscillatory parts of solutions in all considered approximations , see , , , , .
_ does three - timing method really needed ? _ this paper is aimed to create a general viewpoint on distinguished limits , drifts , and pseudo - diffusion based on the rigorous implementation of the two - timing method .
we do not use any additional suggestions and assumptions , hence any contradictions ( either mathematical or physical ) in further results ( which follow from our averaged equations and solutions ) can be explained by insufficiency of two time - scales only .
indeed , the presence of scaling parameters , such as @xmath0 , @xmath199 , and @xmath200 , allows one to introduce an infinite number of additional time - scales .
from this perspective the considered problem can be seen as a test for sufficiency of the two - timing method . in particular , due to the presence of pseudo - diffusion , some secular ( in @xmath25 ) terms could appear in solutions of , and .
if it is proven , then one may suggest that the two - timing method fails at the orders of approximations , where such a secular growth appears . in this case further time - scales ( additional to @xmath26 and @xmath25 ) can be introduced , which requires the systematic development of three - timing _ etc . _
methods . however , any mathematically systematic method , which allows to derive the averaged equations with three or more time - scales from an original pde , is still unknown . _
mathematical justification of the two - timing method : _ one can rewrite the approximate solution ( along with its tilde - parts given in appendix ) back to the original time variable @xmath33 and substitute it into the exact original equation ( [ exact-1 ] ) : then a small residual ( a nonzero right - hand - side in ( [ exact-1 ] ) ) appears .
the two - timing method ( if it used formally ) allows to produce an approximate solution with a residual as small as required by an user .
however , the next logical step is more challenging : one has to prove that the solution of the equation with a small residual ( instead of zero in the right - hand side ) is close to the exact solution .
for the two - timing procedure such proofs had been performed for the leading approximation of solution by @xcite in the problem of vibrational convection .
similar justifications for other equations with oscillatory coefficients are not known yet . _
the interplay between a velocity amplitude and slow time - scale : _ a different ( from the presented in this paper ) approach is given in @xcite where _ an inspection procedure _ for calculation of distinguished paths is proposed and the concept of ` strong ' , ` moderate ' , and ` weak ' oscillations ( or vibrations ) is introduced . the advantage of the approach presented in this paper is the additional possibility to vary the slow time - scale @xmath25 .
it makes the structure of asymptotic solutions more flexible and allows to consider broader classes of asymptotic solutions .
however , from a ` physical viewpoint ' some of our results can appear as ` unexpected ' and ` paradoxical ' .
say , in the case @xmath113 in , which looks ` physically natural ' , one should take @xmath114 , @xmath115 , @xmath116 , and @xmath117 .
then the appearance of the variable @xmath116 may not be seen as ` natural ' . from another side ,
if one takes @xmath81 ( which also looks ` physically natural ' ) , then it must be @xmath201 , that , again , could be viewed as physically ` artificial ' .
such an interplay between the scale of velocity and slow time - scale does create some ample confusions and misunderstandings . _ the most striking case
purely periodic oscillations : _ the results for dl(2 ) are particularly striking for the case of purely periodic velocity oscillations , when @xmath122 is independent of @xmath25 . here
the most intricate question is : how to choose any particular scale of @xmath25 ?
we have an answer : the slow time - scale is uniquely determined by the magnitude @xmath123 of @xmath103 . for every value of @xmath23
we obtain the related dl(2 ) slow time @xmath125 .
it should be accepted that solutions with different @xmath23 are physically different , since they correspond to different orders of the prescribed velocity field @xmath15 .
an additional ` naive ' question , which could be asked here , is : why do we need to introduce any slow time - scale @xmath25 at all ?
the answer to this question is : if for the coefficient @xmath122 we consider only a solution of similar structure @xmath202 , then we can overlook a broad class of asymptotic solutions .
a well - known and clarifying example here is the solutions of mathieu equation , which represents an ode with purely periodic coefficient , see @xcite .
the slow time - scale , which appears there ( say , in the slowly growing solutions , related to the parametric resonance ) , is proportional to the amplitude of frequency modulation .
our case looks somehow similar : the slow time - scale appears to be related to the amplitude of velocity oscillations .
we believe that such a relation between the amplitude of purely oscillating coefficients and the slow time - scale represents a general property of related pdes and odes . _ on the term ` pseudo - diffusion ' : _ this term has been in a diverse use in various applied disciplines , some of them remote from fluid dynamics and mathematical physics , say in chemistry , geology , _
etc_. for example , its definition from geology is : `` mixing of thin superposed layers of slowly accumulated marine sediments by the action of water motion or subsurface organisms '' ( see mcgraw - hill dictionary of scientific and technical terms , 6e , 2003 by the mcgraw - hill companies , inc . ) .
however , there is one purely mathematical definition , where ` pseudo - diffusion ' appears as the ` hyperbolic type ' spatial operator @xmath203 with @xmath204 , see @xcite .
this definition is qualitatively close to one case ( stokes drift ) which appears in this paper . at the same time , we have extended the meaning of ` pseudo - diffusion ' by including all possible cases such as @xmath205 with the function @xmath206 changing its sign in space and time . _
averaged ` lagrangian ' dynamics vs. exact lagrangian dynamics ?
_ _ example 5 _ demonstrates that a drift can produce chaotic averaged dynamics which in this case can not be directly affiliated with material particles .. this result brings on a number of questions , such as : ( i ) what is the relationship between the lagrangian chaotic motions for the original dynamical system and ` lagrangian ' chaotic motions for the averaged one ? ( ii )
how can chaotic averaged ` lagrangian ' motion and pseudo - diffusion complement each other ?
( iii ) how can ` averaged chaos ' , induced by a drift , be used in the theory of mixing ? ( iv ) let the averaged ` lagrangian ' dynamics be chaotic , then how can the related results by @xcite , and many others be used in applications .
another related example could be the abc - flow by @xcite , however , to make it relevant to the topic of this paper we need to find such an oscillating flow which has the abc - flow as its drift .
_ additional reading materials : _ for an interested reader : all the presented in this paper calculations as well as the calculations for different distinguished limits and a number of additional ( to sect.7 ) examples are given in the arxiv papers by @xcite , which are quoted below as i and ii .
_ is our set of distinguished limits dl(n ) complete ?
_ the key question can be asked : does the described in this paper set of distinguished limits and parametric solution represent a full set of asymptotic solutions of , obtainable by the two - timing method ?
the answer is unknown , but most likely some other solutions do exist ; a few attempts to identify additional solutions are given in i and ii . _ simplification of @xmath26-periodicity : _ all results of this paper have been obtained for the class of @xmath26-periodic functions , which is self - consistent .
one can consider more general classes of quasi - periodic , non - periodic , or chaotic solutions .
the discussion on this topic is given in i , ii .
however , it is worth to understand the properties of @xmath26-periodic oscillations , in order not to relate these properties exclusively to more general solutions . at the same time
, the relative simplicity of calculations for the @xmath26-periodic solutions allows to obtain some advanced results which can serve as a guidance for making assumptions on the properties of more general solutions .
_ lagrangian drift vs. eulerian drift .
_ it is worth to calculate the classical ( lagrangian ) drift directly by solving ( [ exact-1a ] ) with the use of the same two - timing method .
the drift , obtained in this paper , appeared as the result of eulerian average operation ; hence , its comparison with the classical drift represents an open problem .
this topic requires a separate study which has been started in i , ii .
_ advection of vectorial admixture . _
the averaged equations for a passive vectorial admixture are presented in i , ii . these equations are closely linked to the problem of _
kinematic @xmath207-dynamo _ , see @xcite .
it is physically apparent , that for the majority of shear drift velocities @xmath208 the averaged stretching of ` material lines ' produces the linear in @xmath25 growth of a magnetic field @xmath209 . at the same time , for the averaged flows with exponential stretching of averaged ` material lines ' these examples will inevitably show the exponential growth .
another closely related research topic is the ` advection ' of an active vectorial admixture ( vorticity ) .
the vortex dynamics of oscillating flows has been studied in @xcite .
an interesting phenomenon here is the _
langmuir circulations _ , see @xcite , which has been recently analyzed from a new perspective by @xcite .
all these topics are worth to be studied by the approach presented in this paper .
the author is grateful to prof .
ilin for the checking of calculations and to prof .
morgulis for the computing example 5 .
the author wants to express special thanks to profs .
craik and h.k .
moffatt for reading this manuscript and making useful critical remarks , also many thanks to profs .
mcintyre , t.j .
pedley , m.r.e .
proctor , and d.w .
hughes for helpful discussions .
this research is supported by the grant ig / sci / doms/16/13 from sultan qaboos university , oman .
first , we list the properties of @xmath26-differentiation and tilde - integration which are intensely used in the calculations below . for @xmath26-derivatives
it is clear that @xmath210 the product of two tilde - functions @xmath211 and @xmath212 forms a general @xmath26-periodic function : @xmath213 , say . separating the tilde - part @xmath214 from @xmath215 we write @xmath216 where the introducing of braces for the tilde parts of a function is aimed to avoid the cumbersome two - level tilde notation .
as the average operation represents the integration over @xmath26 , then for products containing tilde - functions @xmath217 and their derivatives we have @xmath218 which can be seen as different versions of integration by parts .
similarly , for the commutators we have @xmath219\rangle=-\langle[\widetilde{{\boldsymbol{a}}}_\tau,\widetilde{{\boldsymbol{b}}}]\rangle=- \langle[\widetilde{{\boldsymbol{a}}}_\tau , { \boldsymbol{b}}]\rangle,\ \langle[\widetilde{{\boldsymbol{a}}},\widetilde{{\boldsymbol{b}}}^\tau]\rangle=-\langle[\widetilde{{\boldsymbol{a}}}^\tau,\widetilde{{\boldsymbol{b}}}]\rangle=- \langle[\widetilde{{\boldsymbol{a}}}^\tau , { \boldsymbol{b}}]\rangle \label{oper-15}\end{aligned}\ ] ] now , we describe the obtaining of the solution of equation for @xmath60 @xmath220 the substitution of ( [ basic-4aa ] ) into this equation produces the equations of successive approximations @xmath221 the separation of the tilde - parts @xmath222 , for @xmath223 produces the explicit recurrent expressions @xmath224 further calculations and transformation with the use of - show that the bar - parts @xmath225 satisfy the equations - .
let us present the derivations of - and - .
( [ abasic-5 ] ) is : @xmath226 the substitution of @xmath227 into ( [ a01-appr-1 ] ) gives @xmath228 .
its tilde - integration produces the unique ( inside the tilde - class ) solution @xmath229 . at the same time , ( [ a01-appr-1 ] ) does not impose any restrictions on @xmath230 , which must be determined from the next approximations .
thus the results derivable from ( [ a01-appr-1 ] ) are : @xmath231 ( [ abasic-6 ] ) is @xmath232 the use of @xmath229 and @xmath233 reduces ( [ a01-appr-3 ] ) to the equation @xmath234 .
its tilde - integration gives the unique solution @xmath235 hence , @xmath236 where @xmath237 are not defined .
( ( [ abasic-7 ] ) for @xmath60 ) is @xmath238 the use of ( [ a01-appr-1a ] ) and @xmath239 transforms ( [ 2-appr-1 ] ) into @xmath240 its bar - part is @xmath241 where we have used @xmath242 , @xmath243 , and @xmath244 .
the substitution of ( [ 01-appr-7 ] ) into ( [ 2-appr-4 ] ) produces the equation @xmath245 one may expect that the right hand side of ( [ 2-appr-5 ] ) contains both the first and the second spatial derivatives of @xmath246 , however _ all the second derivatives vanish_. in order to prove it , we introduce the commutator @xmath247= ( \widetilde{{\boldsymbol{u}}}\cdot\nabla)\widetilde{{\boldsymbol{\xi}}}-(\widetilde{{\boldsymbol{\xi}}}\cdot\nabla)\widetilde{{\boldsymbol{u}}},\label{app1 - 2}\\ & & { \boldsymbol{k}}\cdot\nabla=(\widetilde{{\boldsymbol{u}}}\cdot\nabla)(\widetilde{{\boldsymbol{\xi}}}\cdot\nabla)-(\widetilde{{\boldsymbol{\xi}}}\cdot\nabla)(\widetilde{{\boldsymbol{u}}}\cdot\nabla ) \label{app1 - 1}\end{aligned}\ ] ] the bar - part of ( [ app1 - 1 ] ) is @xmath248 at the same time , the integration by parts over @xmath26 gives @xmath249 combining ( [ app1 - 3 ] ) with ( [ app1 - 4 ] ) we obtain @xmath250 which reduces ( [ 2-appr-5 ] ) to the advection equation ( [ 4.15 ] ) with @xmath251\rangle=-\frac{1}{2}\overline{{\boldsymbol{k } } } \label{2-appr-8}\end{aligned}\ ] ] which also gives the main term in drift velocity ( [ 4.18 ] )
. the tilde - part of ( [ 2-appr-2 ] ) appears after subtracting ( [ 2-appr-4 ] ) from ( [ 2-appr-2 ] ) : @xmath252 its tilde - integration with the use of ( [ 01-appr-7 ] ) gives ( [ a4.12 ] ) : @xmath253 hence , @xmath254 can be written as @xmath255 where @xmath246 and @xmath256 are given by ( [ 4.15 ] ) , ( [ 2-appr-11 ] ) , while @xmath257 are not defined .
( ( [ abasic-7 ] ) for @xmath258 ) is : @xmath259 its bar - part is @xmath260 the substitution of ( [ 2-appr-11 ] ) into ( [ 3-appr-2 ] ) , the use of @xmath261 , and the integration by parts yield @xmath262 where @xmath263 has been already simplified in ( [ app1 - 5 ] ) . the second term in right hand side of ( [ 3-appr-3 ] )
formally contains the third , the second , and the first spatial derivatives of @xmath246 ; however _ all the third and the second derivatives vanish_. to prove it , first , we use ( [ oper-9a ] ) : @xmath264 then we use ( [ app1 - 2 ] ) , ( [ app1 - 1 ] ) to transform the sequence of operators @xmath265 and @xmath266 in each term in right hand side of ( [ app2 - 1 ] ) into their sequence in left hand side .
the result is @xmath267\label{app2 - 2}\end{aligned}\ ] ] as the result ( [ 3-appr-3 ] ) takes a form ( [ 4.16 ] ) with @xmath193 ( [ 2-appr-8 ] ) and @xmath268,\widetilde{{\boldsymbol{\xi}}}]\rangle=-\frac{1}{3}\overline{{\boldsymbol{k } } ' } \label{3-appr-4a}\end{aligned}\ ] ] which gives ( [ 4.18 ] ) .
the tilde - part of ( [ 3-appr-1 ] ) after its integration gives @xmath269 where @xmath270 is given by ( [ 2-appr-11 ] ) .
hence , @xmath271 can be written as @xmath272 where @xmath246 , @xmath273 , @xmath256 , and @xmath274 are given by ( [ 4.15 ] ) , ( [ 4.16 ] ) , ( [ 2-appr-11 ] ) , and ( [ 3-appr-5 ] ) , while @xmath275 are not defined .
( ( [ abasic-7 ] ) for @xmath276 ) is : @xmath277 its bar - part is @xmath278 the substitution of @xmath279 ( [ 3-appr-5 ] ) into ( [ 4-appr-2 ] ) , @xmath270 ( [ 2-appr-11 ] ) into @xmath279 ( [ 3-appr-5 ] ) , the integration by parts ( [ oper-9 ] ) , and the use of ( [ 2-appr-8 ] ) , ( [ 3-appr-4a ] ) yield @xmath280 the ` gothic ' shorthand operator @xmath281 ( as well as operators @xmath282 , @xmath283 , @xmath284 , @xmath285 , and @xmath286 below ) acts on @xmath246 .
right hand side of ( [ app3 - 1 ] ) formally contains the fourth , the third , the second , and the first spatial derivatives of @xmath246 ; however _ all the fourth and the third derivatives vanish_. in order to prove it we first rewrite @xmath281 as @xmath287 the use of ( [ oper-9a ] ) and ( [ app1 - 2 ] ) , ( [ app1 - 1 ] ) transforms ( [ app3 - 2 ] ) into @xmath288
let us now simplify @xmath289 and @xmath290 .
for @xmath290 we use ( [ oper-9a ] ) @xmath291 to change @xmath292 into @xmath293 in the last term we use ( [ app1 - 2 ] ) , ( [ app1 - 1 ] ) that yields : @xmath294\label{app3 - 4}\end{aligned}\ ] ] the operator @xmath289 is simplified by the version of ( [ oper-9a ] ) with four multipliers @xmath295 the multiple use of commutator ( [ app1 - 2 ] ) , ( [ app1 - 1 ] ) allows us to transform the sequence of operators @xmath265 and @xmath266 in each term in right hand side of ( [ app3 - 5 ] ) to the sequence in its lhs .
the result is @xmath296\label{app3 - 6}\end{aligned}\ ] ] now , ( [ app3 - 3 ] ) , ( [ app3 - 4 ] ) , and ( [ app3 - 6 ] ) yield @xmath297 the substitution of this expression into ( [ app3 - 1 ] ) , ( [ 4-appr-2 ] ) gives @xmath298 additional transformations of the last two operators in ( [ app3 - 8 ] ) yield @xmath299\rangle\cdot\nabla -\frac{1}{2}\langle(\widetilde{{\boldsymbol{u}}}'\cdot\nabla)\widetilde{{\boldsymbol{\xi}}}+ ( \widetilde{{\boldsymbol{\xi}}}\cdot\nabla)\widetilde{{\boldsymbol{u}}}'\rangle\cdot\nabla - \frac{1}{2}\langle \widetilde{u}'_i\widetilde{\xi}_k+\widetilde{u}'_k\widetilde{\xi}_i\rangle\frac{\partial^2}{\partial x_i\partial x_k}=\nonumber\\ & & = \frac{1}{2}\langle[\widetilde{{\boldsymbol{\xi}}},\widetilde{{\boldsymbol{\xi}}}_s]\rangle\cdot\nabla-\frac{\partial}{\partial x_k}\left(\overline{\chi}_{ik}\frac{\partial}{\partial
x_i}\right)+\frac{1}{2}\langle\widetilde{{\boldsymbol{\xi}}}{\mathrm{div}\,}\widetilde{{\boldsymbol{u}}}'+\widetilde{{\boldsymbol{u}}}'{\mathrm{div}\,}\widetilde{{\boldsymbol{\xi}}}\rangle \label{ap - trans}\\ & & \widetilde{{\boldsymbol{u}}}'\equiv\widetilde{{\boldsymbol{\xi}}}_s-[\overline{{\boldsymbol{v}}}_0,\widetilde{{\boldsymbol{\xi}}}],\quad \overline{\chi}_{ik}\equiv\frac{1}{2}\langle \widetilde{u}'_i\widetilde{\xi}_k+\widetilde{u}'_k\widetilde{\xi}_i\rangle \label{v - prime}\end{aligned}\ ] ] the substitution of ( [ ap - trans ] ) into ( [ app3 - 8 ] ) leads to the equation for @xmath300 ( [ 4.17 ] ) where the formula ( [ 4.20 ] ) for @xmath98 is obtained from ( [ v - prime ] ) by the use of definition @xmath301 .
the tilde - part of ( [ 4-appr-1 ] ) after its tilde - integration gives ( [ a4.14 ] ) @xmath302 where @xmath279 is given by ( [ 3-appr-5 ] ) .
hence , we have solved the equation for the first five approximations and obtained the correspondent required terms in ( [ basic-4aa ] ) .
simonenko , i.b .
1972 justification of averaging method for convection problem in the field of rapidly oscillating forces and for other parabolic equations . _ math .
sbornik _ ,
* 87(129 ) * , 2 , 236 - 253 ( in russian ) . | the aim of this paper is to study and classify the multiplicity of distinguished limits and asymptotic solutions for the advection equation with a general oscillating velocity field with the systematic use of the two - timing method .
our results are : \(i ) the dimensionless advection equation contains _ two independent small parameters _ , which represent the ratio of two characteristic time - scales and the spatial amplitudes of oscillations ; the scaling of the variables and parameters contains strouhal number ; \(ii ) an infinite sequence of _ distinguished limits _ has been identified ; this sequence corresponds to the successive degenerations of a _ drift velocity _ ; \(iii ) we have derived the averaged and oscillatory equations for the first _ four distinguished limits _ ; derivations are performed up to the forth orders in small parameters ; \(v ) we have shown , that _ each distinguish limit solution _ generates an infinite number of _ parametric solutions _ ; these solutions differ from each other by the slow time - scale and the amplitude of prescribed velocity ; \(vi ) we have discovered the inevitable appearance of _ pseudo - diffusion _ , which appears as a lie derivative of the averaged tensor of quadratic displacements ; we have clarified the meaning of pseudo - diffusion using a simple example ; \(vii ) our main methodological result is the introduction of a logical order into the area and classification of an infinite number of asymptotic solutions ; we hope that it can help to study the similar problems for more complex systems ; \(viii ) since in our calculations we do not employ any additional assumptions , our study can be used as a test for the validity of the two - timing hypothesis ; \(ix ) the averaged equations for five different types of oscillating velocity fields have been considered as the examples of different drifts and pseudo - diffusion . |
in the past years several theoretical and computer simulation studies have hypothesized possible scenarios to explain the thermodynamic anomalies of water in the supercooled region .
the _ critical point scenario _ ascribes the large increase of thermodynamic response functions to the presence of a liquid - liquid ( ll ) phase transition in the supercooled region between a low density liquid ( ldl ) and a high density liquid ( hdl ) .
@xcite in this picture the coexistence line between the two liquids terminates in a liquid - liquid critical point ( llcp ) . from the llcp a line of maxima of correlation length , called the widom line ,
is emanated .
@xcite the lines of extrema of several thermodynamic response functions , asymptotically merge on the widom line upon approaching the critical point from higher temperatures .
@xcite the anomalous behavior of water can alternatively be related to a _ singularity free scenario _
@xcite in which the increase in thermodynamic response functions is due to local density fluctuations that do not eventually lead to singularities .
recently another picture was proposed , the _
critical point free scenario_. @xcite in this case the hdl - ldl transition is interpreted as an order - disorder transition of the first order and no llcp occurs . experimental investigation of the deeply supercooled region of bulk water is hampered by nucleation .
@xcite indications of the existence of llcp in supercooled bulk water have been found .
@xcite in particular the llcp was hypothesized to be located at about @xmath0 k and @xmath1 mpa .
the critical point scenario was formulated by poole et al .
@xcite as a result of computer simulations on water with the st2 potential .
this interpretation was confirmed by other simulations on st2 water , @xcite until the very recent paper by liu et al .
@xcite the llcp was also found in water simulated with other potentials .
@xcite lattice models of water have indicated that upon properly tuning the parameters of the models one can switch from one scenario to the other .
@xcite simulations on spherically symmetric ramp potential particles , mimicking the properties of water , show the existence of llcp for proper choices of the parameters .
@xcite possible experimental routes for the observation of the supercooled region where the llcp is supposed to exist are therefore timely . in the natural environment water
is almost ever found as a solvent in mixtures of two or more components and water can be more easily supercooled in solutions .
@xcite indications of the possible presence of a llcp in aqueous solutions have been found experimentally , @xcite theoretically @xcite and in computer simulations .
@xcite nonetheless the existence and the position of llcp in the thermodynamic plane for an aqueous solution has not yet been determined either in experiments or in simulations . here
we present an extensive molecular dynamics ( md ) study of tip4p bulk water and of @xmath2 sodium chloride aqueous solution , nacl(aq ) , in tip4p water .
this study aims to locate the position of the llcp in the aqueous solution and to compare it with the llcp of bulk water in order to see if a measure of the llcp is experimentally feasible in the solution .
we selected this particular concentration because it was found in experiments that it is high enough to show deviation from the bulk behavior but not so high to destroy water anomalies .
@xcite the chosen salt concentration is close to that of seawater @xmath3 and it is well above the physiological concentration which is @xmath4 . the paper is organized as follows . in sec .
[ methods ] the simulation details are given .
results are presented and discussed in sec .
[ res ] an conclusions are drawn in sec .
[ conclusions ] .
the tip4p water potential has been extensively used for studying supercooled water and it has been recently shown also to successfully reproduce the phase diagram of all the stable ice phases of water .
@xcite md simulations on this potential @xcite estimated the llcp to be at @xmath5 k and @xmath6 mpa .
however the exact position has not been clearly determined so far .
ion - ion and ion - water parameters are taken from jensen and jorgensen .
@xcite the two systems are studied upon supercooling and the range of temperatures spanned goes from @xmath7 k to @xmath8 k. the range of densities studied is @xmath9 .
the total number of particles in the simulation box is @xmath10 . in the case of the solution
@xmath11 with @xmath12 and @xmath13 .
the cut - off radius is set to 9.0 and the long - range electrostatic interactions are treated by the ewald summation method . a very recent study proved the importance of the use of ewald summations @xcite in the calculation of the phase diagram especially close to the critical region .
temperature is controlled with the use of the berendsen thermostat .
@xcite the longest runs last up to 30 ns and we collected a total of 1848 state points .
the total computational time employed is about six central processing unit ( cpu ) years on a single processor .
simulations were performed using the parallel version of the dl_poly package .
@xcite the temperature of maximum density ( tmd ) lines have been calculated from the minima of the isochores .
the liquid - gas ( lg ) and the ll limit of mechanical stability ( lms ) lines have been calculated considering the points where @xmath14 .
the position of the llcp point has been estimated considering the highest temperature of convergence of the isochores in the @xmath15 plane @xcite , which is also the upper bound of the ll lms , and also considering the development of a flat region in the isotherms plane and the convergence of the lines of maxima of the isothermal compressibility @xmath16 and of the constant volume specific heat @xmath17 .
generally speaking in fact , the existence of a critical point causes large fluctuations in the region around it . as a consequence above the critical temperature
the thermodinamic response functions show loci of extrema that all converge on the widom line at the critical point @xcite .
in fig . [ fig:1 ]
we report for bulk tip4p water the @xmath15 isochores plane , the position of the llcp of the system , the lines of extrema of thermodynamic response functions , tmd points and lg lms line .
we locate the position of the llcp for tip4p bulk water to be at @xmath8 k , @xmath18 mpa and @xmath19 , being in agreement with limits previously estimated for tip4p bulk water .
@xcite in fig .
[ fig:2 ] we report the llcp , the tmd , and the lg lms for the bulk tip4p together with the experimental tmd line .
if we now compare the bulk tip4p tmd and the llcp with the experiments we note that upon shifting the tip4p phase diagram of @xmath20 k and @xmath21 mpa , the tip4p tmd exactly superposes to the experimental values and the llcp falls very close to its experimentally hypothesized position .
these shifts in pressure and temperatures of the tip4p phase diagram also agree with the shifts found for ice phases of tip4p .
@xcite we also note that the turning point of the shifted tmd line falls around @xmath22 k and @xmath23 mpa which is in the experimentally accessible region , above the melting temperature line .
@xcite experiments on stretched water have reached @xmath24 mpa .
@xcite in fig .
[ fig:3 ] , similar to the bulk , we show the isochores plane for nacl(aq ) , the estimated position of the llcp of the system , the lines of extrema of thermodynamic response functions , the tmd points and the ll and lg lms lines .
for the same set of densities we observe that the isochores for the solution globally shift to lower pressures with respect to bulk water , with a packing of the isochores close to the lg lms line .
this shift of the thermodynamic plane has been previously observed in solutions .
@xcite we locate the position of the llcp for nacl(aq ) to be at @xmath25 k , @xmath26 mpa and @xmath27 . in the solution it was possible to reach equilibrium in our md simulations for temperature low enough to determine also the two branches of the ll lms line .
the results obtained directly from simulations for bulk water and for the @xmath28 nacl(aq ) are summarized in fig .
[ fig:4 ] .
we also report in the same picture both the experimental tmd line for bulk water and a tmd point at ambient pressure extrapolated for our concentration of salt from experimental data .
@xcite both in bulk water and in nacl(aq ) the lines of maxima of the thermodynamic response functions merge in the widom line very close to the llcp .
the lms line does not move in the solution with respect to the bulk .
the tmd shifts of 10 k to lower temperatures and of circa 20 mpa to lower pressures .
the same shift in temperature is found between the experimental tmd line of bulk water and the tmd point of the solution , at ambient pressure showing that our md is able to capture not only the direction but also the magnitude of the shift in temperature found in experiments and confirming that tip4p is a very good potential for simulating water .
we therefore infer that the shift in temperature observed for the llcp between bulk tip4p water and tip4p water in solution is the same as it could be observed in experiments .
the llcp moves in the solution of 10 k toward higher temperatures and shifts in the same direction as tmd in pressure , but of a much larger amount , @xmath29 mpa .
this is the signature of the fact that ions stabilize the hdl region consistently shrinking in pressure the ldl region , in agreement with the fact the ions seem to be more favorably solvated in hdl .
@xcite this leads to an extension of the region of existence of the phase where the local structure of water needs less empty spaces .
@xcite we observed , see fig .
[ fig:2 ] , that the tip4p bulk phase diagram is shifted of @xmath30 k and @xmath31 mpa with respect to the experimental result . on the basis of the comparison with the experimental tmd
we hypothesize that this shift is preserved for the solution .
we report in fig .
[ fig:5 ] the phase diagram of nacl(aq ) as obtained directly from simulations together with the shifted phase diagram .
we observe that the experimental tmd value measured for the solution falls very close to the tmd shifted curve of the md simulation .
we predict for the experimental @xmath2
nacl(aq ) a llcp at around @xmath32 k and @xmath33 mpa . in experiments on water large negative pressures
have been obtained , in particular in nacl(aq ) with concentration @xmath34=1 mol / kg rupture occurs at @xmath35 mpa @xcite , therefore the negative critical pressure in solution @xmath33 mpa can be reached .
we report in fig .
[ fig:6 ] a schematization of the phase diagrams of both bulk and nacl(aq ) obtained from simulations results shifted to match the experimental values .
we infer that these should be the phase diagrams of water and of a @xmath2 nacl(aq ) aqueous solution as measurable in experiments .
we can also see how the ldl region in the solution shrinks with respect to the bulk . on the basis of extrapolation of the data at ambient pressure for other concentrations of nacl(aq ) @xcite we can estimate that the homogenous nucleation line for our concentration should be shifted of circa 5 k downward with respect to bulk water ( see e.g. fig . 1 in ref .
the llcp in the solution should then be within the supercooled accessible region and a measurement of the llcp appears possible .
recent experimental studies on licl(aq ) have reached the supercooled temperature of @xmath25 k @xcite , showing the existence , also in aqueous solutions , of a fragile to strong dynamical transition typical of glass formers . in water
this dynamical transition happens , both in experiments @xcite and in simulations @xcite , upon crossing the widom line and therefore appears to be related to the possible existence of a llcp .
the widom line in water , a network forming liquid , appears to play the role of switching line for hopping , favored on the side where water is less dense @xcite .
in this paper we have investigated the possibility to detect the hypothesized llcp of supercooled water in aqueous solutions of salts .
we conclude that the phase diagram of real water is reproducible by tip4p water upon applying a shift in temperature and pressure that brings the md tmd to coincide with the experimental values .
the llcp in this shifted diagram falls at @xmath36 k and @xmath37 mpa and is located very close to the value hypothesized by mishima and stanley , @xcite @xmath0 k and @xmath1 mpa , corroborating our findings . upon applying an analogous shift to the phase diagram of the aqueous solution
, we predict the llcp to be experimentally detectable at @xmath38 k @xmath39 mpa for @xmath28 nacl(aq ) .
the comparison of the phase diagrams of the aqueous solution and of bulk water shows that the ions stabilize the hdl phase consequently shrinking the ldl region .
the existence of a llcp has been connected to a double well effective potential between water molecules .
@xcite at low enough temperatures the outer and deep subwell drives the system to condense in the ldl phase while the inner subwell at higher enough pressures forces the transition into the hdl phase . if we follow this interpretation our results on the aqueous solution indicate that the effect of the ions is to shift the inner subwell to higher distances so that the hdl phase becomes stable at lower pressures with respect to bulk water . as a consequence
also the tmd line shifts closer to the llcp .
the ions instead do not affect the outer subwell since the lg lms does not change going from the bulk to the solution .
thus we propose as a valuable route for the experimentalists to solve the issue of the quest of the llcp of water to perform measurements on aqueous solution of salts .
we thank h. e. stanley for helpful discussions and for a critical reading of the manuscript .
the computational resources of the cineca for the `` progetto calcolo 891 '' , of the infn rm3-grid at the roma tre university and of the democritos national simulation center at sissa are gratefully acknowledged . | in this paper we investigate the possibility to detect the hypothesized liquid - liquid critical point of water in supercooled aqueous solutions of salts .
molecular dynamics computer simulations are conducted on bulk tip4p water and on an aqueous solution of sodium chloride in tip4p water , with concentration c = 0.67 mol / kg .
the liquid - liquid critical point is found both in the bulk and in the solution .
its position in the thermodynamic plane shifts to higher temperature and lower pressure for the solution .
comparison with available experimental data allowed us to produce the phase diagrams of both bulk water and the aqueous solution as measurable in experiments . given the position of the liquid - liquid critical point in the solution as obtained from our simulations , the experimental determination of the hypothesized liquid - liquid critical point of water in aqueous solutions of salts appears possible . |
suppose that a sorting algorithm , knowingly or unknowingly , uses element comparisons that can err .
considering sorting algorithms based solely on binary comparisons of the elements to be sorted ( algorithms such as insertion sort , selection sort , quicksort , and so on ) , what problems do we face when those comparisons are unreliable ?
for example , @xcite gives a clever @xmath0 algorithm to assure , with probability @xmath1 , that a putatively sorted sequence of length @xmath2 is truly sorted .
but knowing the structure of the ill - sorted output would likely make error checking easier .
also , in situations in which a reliable comparison is the fruit of a long process , one could chose to interupt the comparison process , thus trading reliability of comparisons ( and quality of the output ) for time . as a first step in order to understand the consequences of errors , we propose to analyze the number of inversions in the output of a sorting algorithm ( we choose quicksort @xcite ) subject to errors .
we assume throughout this paper that the elements of the sequence @xmath3 to be sorted are distinct .
we assume further that the only comparisons subject to error are those made between elements being sorted ; that is , comparisons among indices and so on are always correct .
errors in element comparisons are random events , spontaneous and independent of each other , of position , and of value , with a common probability @xmath4 , @xmath2 being the length of the list to be sorted .
the number of inversions in the output sequence @xmath5 is denoted @xmath6 we assume that the input list is presented in random order , each of the @xmath7 random orders being equiprobable .
finally we denote by @xmath8 the random number of inversions in the output sequence of quicksort subject to errors .
our result is , roughly speaking , @xmath9 when @xmath10 , meaning that @xmath11 converges to some nondegenerate probability distribution .
the surprise " , not so unexpected after the fact , is that there are phase changes in the limit law , depending on the asymptotic behaviour of @xmath12 .
the organization of this paper is as follows : the results are stated in section [ s : results ] . in section [ functequns ]
, we establish a general distributional identity for @xmath8 . in the remaining sections , we prove convergence results for @xmath8 when : * @xmath13 , @xmath14 , * @xmath4 vanishes more slowly than @xmath15 , * @xmath16 where @xmath17 is a positive constant .
the case @xmath18 is different and not treated in detail ; see remark [ r : toosmall ] . in section [ fixedpts ]
, we establish a general result of convergence using contraction methods ( cf .
@xcite ) , and we use it in section [ proofth1 ] , for the first two cases .
these methods do not apply for case @xmath19 , which requires poissonization ( see section [ proofth3 ] , where we use an embedding of quicksort in a poisson point process ) .
set @xmath20 we will always let @xmath21 denote a random variable that is uniformly distributed on @xmath22 $ ] . also , @xmath23 shall denote the set of positive integers , and @xmath24 the set of nonnegative integers .
[ p = c ] if @xmath27 , @xmath28 $ ] , then @xmath29 converges in distribution to a random variable @xmath30 whose distribution is characterized as the unique solution with finite mean of the equation @xmath31
^ 2x_c+[(2c-1)u+1-c]^2\widetilde{x}_c+t(c , u),\end{aligned}\ ] ] in which @xmath32 denotes a copy of @xmath30 , @xmath33 are independent , and @xmath34 furthermore , @xmath35}= \frac{2-c}{2(1 + 2c-2c^2)},\ ] ] and
@xmath36 as usual with laws related to quicksort , see e.g.@xcite , @xmath37 is approximately the position of the pivot of the first step of the algorithm . as in standard quicksort recurrences , the coefficients of @xmath30 and of its independent copy @xmath38 are related to the sizes of the two sublists on the left and right of the pivot , sizes respectively asymptotic to @xmath39 and @xmath40 .
the toll function @xmath41 is approximately @xmath42 times the number of inversions created in the first step : @xmath43 is approximately the number of inversions of the @xmath44 elements , smaller than the pivot but misplaced on the right of it , with the @xmath45 elements smaller than the pivot , that are placed , as they should be , on the left ; @xmath46 is the number of inversions between misplaced elements from the two sides of the pivot .
the toll function @xmath41 depends on only one of the two sources of randomness ( the randomly ordered input list , and the places of the errors ) , viz . , the first one , through @xmath21 .
the second source of randomness is killed by the law of large numbers : in the average , each of the @xmath47 misplaced numbers from the right of the pivot produces inversions with one half of the @xmath48 elements smaller than the pivot , that are placed , as they should be , on the left . as opposed to the other values of @xmath49 , the choices @xmath50 and @xmath51 lead to deterministic @xmath52 , without any surprise : for @xmath53 the output sequence is a random uniform permutation , with a number of inversions concentrated around @xmath54 ( * ? ? ?
5.1.1 ) ; for @xmath55 the output sequence is decreasing , and has @xmath56 inversions .
[ npinfini ] if @xmath59 and @xmath60 , @xmath29 converges in distribution to a random variable @xmath61 whose distribution is characterized as the unique solution with finite mean of the equation @xmath62 in , @xmath63 denotes a copy of @xmath61 and @xmath64 are independent .
furthermore , @xmath65}=1\qquad\text{and}\qquad { \textnormal{var}\left(x\right)}=\frac{1}{12}.\end{aligned}\ ] ] note that equation ( [ p = ode1 ] ) is just ( [ p = ode1 ] ) specialized to @xmath66 , but , as opposed to @xmath67 , an additional condition , @xmath68 , is needed to ensure that the law of large numbers still holds .
also , as another difference between ( [ p = ode1 ] ) and ( [ p = ode1 ] ) , for @xmath69 the errors do not change the sizes of the sublists in a significant way .
the solution @xmath61 equals half the sum of the squares of the widths of the random intervals @xmath70 $ ] defined by ( [ recurrence_sur_u ] ) below .
this is equivalent to the following statement : [ propfindprocess ] the solution @xmath61 equals half the area @xmath71 under the find limit process @xmath72 .
for both these claims , see remark [ r:1 ] .
the find process was introduced in @xcite and is pictured at figure [ findprocess ] .
assume that * @xmath74 is a poisson point process with intensity @xmath17 on @xmath75 $ ] , meaning that , for each @xmath2 , @xmath76\right)}\vert$ ] is a poisson random variable with mean @xmath17 , and the second coordinates of points of @xmath74 are uniform on @xmath22 $ ] and independent ( see @xcite for a general definition of poisson point processes ) ; * @xmath77 is an array of independent uniform random variables on @xmath22 $ ] , independent of @xmath74 ; * the random variables @xmath78 are defined recursively by @xmath79 * for @xmath80 $ ] , @xmath81 if @xmath82 , and define , for @xmath83 , ( the sum is a.s .
finite by lemma [ l : mean ] ) @xmath84 the variables @xmath85 describe a fragmentation process ( see @xcite for historical references ) : we start with @xmath86 and recursively break each interval into two at a random point ( uniformly chosen ) . in the @xmath87-th generation
we thus have a partition of @xmath86 into @xmath88 intervals @xmath89 , @xmath90 , with @xmath91 .
the interval of generation @xmath92 that contains @xmath93 is cut at step @xmath87 at the point @xmath94 .
hence @xmath95 in is the distance from @xmath93 to this cut point .
[ p=1/n ] if @xmath59 and @xmath96 , then @xmath29 converges in distribution to @xmath97 .
the family @xmath98 of random variables satisfies the distributional identity : @xmath99 in which , conditionally given that @xmath100 , @xmath101 , @xmath102 and @xmath103 are independent , @xmath101 and @xmath102 are distributed as @xmath104 and @xmath105 , respectively , and @xmath106 in which @xmath107 is a poisson random variable with mean @xmath17 , the random variables @xmath108 are uniformly distributed on @xmath109 $ ] , and @xmath107 , and the @xmath108 s are independent . furthermore , @xmath110}=1 , \qquad { \textnormal{var}\left(x({\lambda})\right)}=\frac1{12}+\frac1{3{\lambda}}.\end{aligned}\ ] ] [ remmesure ] the distributional identities , and really are equations for distributions , but it is more convenient to state them for random variables as done here . for to make sense ,
i.e. in order to insure , for instance , that @xmath101 is a random variable , it is implicitly assumed that the random variables @xmath97 depend measurably on @xmath17 .
thus a solution of is a family of probability measures @xmath111 on @xmath112 , such that there exists a family @xmath113 of random variables defined on the same probability space @xmath114 and satisfying the following properties : 1 . for @xmath83 , @xmath115 is the distribution of @xmath116 , 2 .
@xmath117 is a measurable process ( * ? ?
1 ) , meaning that the mapping @xmath118 is measurable , and such that holds for @xmath117 .
a measurable version of the stochastic process @xmath119 is defined at below ( measurability follows from ( * ? ? ?
* rem . 1.14 ) ) . for uniqueness , we need extra assumptions :
let @xmath120 denote the class of families of distributions @xmath111 satisfying ( i ) and ( ii ) above , plus the condition : 1 . for some @xmath121 , the function @xmath122}\ ] ]
is bounded on any bounded interval of @xmath123 .
let @xmath124 denote the distribution of @xmath97 .
we have [ thm : unique ] the family @xmath125 is the unique solution of in @xmath120 . we do not know whether the extra assumption ( iii ) is necessary . let us comment further on equation .
writing @xmath126 and @xmath127 , we can thus rewrite as @xmath128 where @xmath129 is either the left or right endpoint of @xmath89 ( depending on whether @xmath130 is even or odd ) .
note that , conditioned on the partitions @xmath131
, i.e.=1000 on @xmath132 , each @xmath133 is a poisson process on @xmath89 with intensity @xmath17 , with the processes @xmath133 independent .
since only the distribution of @xmath97 matters , we can by this conditioning and an obvious symmetry of the poisson processes @xmath133 just as well let @xmath129 in be the left endpoint of @xmath89 for every @xmath87 and @xmath130 .
let @xmath134 be a poisson process on @xmath135\times(0,\infty)$ ] with intensity @xmath25 , and let @xmath136 , @xmath137 .
( this is a pure jump lvy process with lvy measure @xmath138 } dt$ ] . )
let @xmath139 be independent copies of this process , independent of @xmath140 .
a scaling argument shows that can be written @xmath141 [ r:1 ] let @xmath142 .
then @xmath143 satisfies , so that @xmath143 is the limit variable @xmath61 in theorem [ npinfini ] .
( @xmath61 is a.s .
finite and has finite mean by lemma [ l : erika ] . )
moreover , the find limit process @xmath72 in @xcite is defined by @xmath144 ; hence @xmath145 .
this justifies proposition [ propfindprocess ] . moreover , by the law of large numbers , @xmath146 as @xmath147 .
it follows ( by dominated convergence using lemma [ l : erika ] ) that for the special version of @xmath97 defined at @xmath148 and hence @xmath97 converges to @xmath61 in distribution as @xmath147 . in this third case
, we have a system of equations involving an infinite family of laws , and we could not adapt the contraction method : we rather use a poissonization . the phase transition from ( [ p = ode1 ] ) to ( [ functional1 ] )
is explained easily : instead of a number of errors @xmath149 , we have now @xmath150 errors at each step , and the law of large numbers does not hold anymore for the number of inversions produced by step 1 .
actually the number @xmath107 of errors at the first step is asymptotically poisson distributed , and the @xmath107 errors are at positions @xmath151 , approximately uniformly distributed on @xmath22 $ ] .
thus , the number of inversions caused by this first step is approximately @xmath152 actually we prove a stronger theorem in each of the three cases , as we prove convergence of laws for the wasserstein @xmath153 metric @xcite .
it entails convergence of the first moment .
the convergence of higher moments is an open problem .
as we shall see in section [ proofth3 ] , the distribution tail @xmath154 decreases exponentially fast ( theorem [ tail ] ) .
[ r : toosmall ] when @xmath18 very slowly , that is @xmath155 , we conjecture that @xmath156 converges in distribution to @xmath157 , with the consequence that @xmath158 , for any positive @xmath159 . actually , the main contribution to @xmath8 comes from the `` first '' error , in some sense .
when @xmath160 , the probability that no error occurs has a positive limit : we conjecture that , conditionally given the occurence of at least one error , the situation is similar to the previous case , that is , @xmath161 converges in distribution to a random variable with values in @xmath162 . when @xmath163 , @xmath164 .
[ r : filter ] finally , we would like to stress that in the proof of convergence for one the three regimes considered in this section , we have to deal simultaneously with any sequence @xmath165 converging to @xmath166 according to this regime .
this can be observed on the key equation , for instance , in which we would like to argue , roughly speaking , that if @xmath12 is close to @xmath166 according to a given regime , then @xmath167 and @xmath168 are also close to @xmath166 according to the same regime , with a large probability : here the same probability @xmath4 is associated to three different integers , @xmath2 , @xmath169 and @xmath170 , that denote the sizes of the input list , and of the two sublists formed at the first step of quicksort , respectively .
thus @xmath4 can not be seen as a sequence indexed by @xmath2 . in order to allow such a loose relation between @xmath2 and @xmath4 , filters turn out to be more handy than sequences ( see ( * ? ? ?
convergences in the three regimes are thus understood as convergences along the three corresponding filters ( see theorem [ th de convergence ] ) .
at the first step quicksort compares all elements of the input list with the first element of the list ( usually called _ pivot _ ) .
all items less ( resp .
larger ) than the pivot are stored in a sublist on the left ( resp .
right ) of the pivot .
comparisons are not reliable , therefore @xmath171 items that should belong to the left sublist are wrongly stored in the right sublist , and @xmath172 items larger than the pivot are misplaced in the left sublist .
since its items are chosen randomly , the input list is a random permutation and the true rank of the pivot can be written @xmath173 , where @xmath21 is uniformly distributed on @xmath22 $ ] and @xmath174 is the ceiling of @xmath93 . also , conditionally given @xmath21 , @xmath171 ( resp .
@xmath172 ) is a binomial random variable with parameters ( @xmath175 , @xmath4 ) ( resp .
( @xmath176 , @xmath4 ) ) .
quicksort with error is then independently applied on the left sublist @xmath177 and on the right sublist @xmath178 and new errors occur , ultimately producing two new sublists @xmath179 and @xmath180 . set @xmath181 so that @xmath169 ( resp .
@xmath170 ) is the size of @xmath177 and @xmath182 ( resp .
@xmath178 and @xmath180 ) .
in order to enumerate the inversions of the output list , we introduce a _ purely fictitious _
error - correcting algorithm that parallels the implementation of quicksort : this fictitious error - correcting algorithm has two recursive steps , * first , the error - correcting algorithm corrects the sublists @xmath183 ( resp .
@xmath180 ) at costs @xmath184 ( resp .
@xmath185 ) ) , producing two increasing sublists @xmath186 and @xmath187 . note that @xmath188 and @xmath189 are conditionally independent , given @xmath190 .
furthermore , the two sublists @xmath177 and @xmath178 obtained at the end of step @xmath25 are in uniform random order before the second step of quicksort , so that , conditionally given @xmath190 , cost @xmath188 ( resp .
@xmath189 ) is distributed as @xmath191 ( resp .
@xmath192 ) .
* then the error - correcting algorithm corrects the errors of step @xmath25 , at a cost @xmath193 . here
@xmath194 stands for the list obtained when one puts @xmath195 , the pivot and @xmath196 side by side .
the number of inversions @xmath197 in the list @xmath194 is analyzed in detail at the end of this section .
[ 2sublists0 ] + these two steps lead to the following equation for @xmath8 : @xmath198 where @xmath199 .
we shall obtain the asymptotic distribution of @xmath197 , and as a consequence ( [ eq en i ] ) will translate , after renormalisation , into a distributional identity satisfied by the limit law of @xmath200 .
the limit law appears on both sides of the distributional identity , as expected , due to the recursive structure of quicksort , and is thus characterized as the fixed point of some transformation . at the end of the first step of the error - correcting algorithm
, we obtain two subarrays @xmath201 and @xmath196 , left and right of the pivot ( cf .
figure [ 2sublists ] ) . they are sorted in increasing order but there are @xmath172 ( red ) elements larger than the pivot just to its left and @xmath171 ( green ) elements smaller than the pivot element just to its right .
thus , the only misplaced elements that the proofreader must correct in step 2 are clustered around the pivot . [ 2sublists ] and @xmath196.,width=453 ] in order to sort the list , the red and green sublists must be exchanged .
this requires @xmath202 inversions .
we get therefore two unsorted lists @xmath203 and @xmath204 each composed of two sorted sublists . all items of @xmath205 ( resp . of @xmath204 )
are now smaller ( resp .
larger ) than the pivot , so that the length of @xmath203 ( resp . of @xmath204 ) is @xmath175 ( resp .
@xmath176 ) . it remains to sort @xmath203 and @xmath204 , at respective costs @xmath206 and @xmath207 that are conditionally independent given @xmath21 , leading to : @xmath208 let @xmath210 be the number of inversions in a list of @xmath211 elements sorted as follows : each element is painted black ( white ) with probability @xmath4 ( resp .
. then the black and white sublists are separately sorted in increasing order and the two sorted sublists are placed side by side , producing a new list @xmath213 with @xmath211 elements .
we have [ invb&w ] let @xmath214 be @xmath211 independent bernoulli random variables with the same parameter @xmath4 , and let @xmath215 .
then @xmath216 let us abbreviate @xmath217 to @xmath218 . among the @xmath219 s ,
let @xmath220 denote the @xmath218 random variables equal to @xmath25 , @xmath221 those equal to @xmath222 , with @xmath223 and @xmath224 .
now @xmath210 can be seen as the number of inversions of the list @xmath225 . in order to move the numbers @xmath226 ( @xmath227 ) to the correct position , the proofreader corrects inversions with each of the @xmath228 elements of @xmath229 that are smaller than @xmath226 and do not belong to @xmath230 .
thus @xmath231 leading to the result . with the help of proposition [ invb&w ] , we can give a useful description of the distribution of @xmath232 and @xmath233 : [ description ] conditionally given that the length of @xmath205 is @xmath234 , @xmath232 and @xmath233 are independent and distributed as @xmath235 and @xmath236 , respectively . to sum up the results of this section , renormalizing ( [ eq en i ] ) , one obtains a distributional identity satisfied by @xmath237 : @xmath238 in which @xmath239 and * @xmath21 is a uniform random variable on @xmath22 $ ] , and @xmath173 is the position of the pivot , * conditionally given @xmath240 , @xmath232 and @xmath233 are distributed as in proposition [ description ] , * @xmath241 , @xmath242 are two independent sequences with the same ( unknown ) distribution , independent of @xmath243 , and therefore of @xmath244 .
the errors having a balancing effect : @xmath199 has the same mean , @xmath245 , and a smaller variance than @xmath246 .
we prove this in the following form .
[ l:2 ] @xmath247 } & \le { \mathbb{e}\left[({\lceilnu\rceil}-1)^2+(n-{\lceilnu\rceil})^2\right ] } = \frac{(n-1)(2n-1)}{3 } \\ & \le \frac23 n^2.\end{aligned}\ ] ] the left hand side is the expected number of ordered pairs @xmath248 that end up on a common side of the pivot .
this happens if @xmath249 and @xmath130 originally are on the same side of the pivot and we either compare both correctly or make errors for both of them , or if they are on opposite sides of the pivot and we make an error for exactly one of them . hence @xmath247 } = { \bigl(p^2+(1-p)^2\bigr ) } { \mathbb{e}\left[({\lceilnu\rceil}-1)^2+(n-{\lceilnu\rceil})^2\right]}\quad&\\ { } + 2p(1-p)2 { \mathbb{e}\left[({\lceilnu\rceil}-1)(n-{\lceilnu\rceil})\right ] } & \\ = { \mathbb{e}\left[({\lceilnu\rceil}-1)^2+(n-{\lceilnu\rceil})^2\right ] } -2p(1-p ) { \mathbb{e}\left[({\lceilnu\rceil}-1-(n-{\lceilnu\rceil}))^2\right]}\qquad & \end{aligned}\ ] ] which proves the first inequality . the rest is a simple calculation .
let us say that an element @xmath250 of the list , or the comparison in which @xmath250 plays the rle of pivot , has depth @xmath87 if @xmath250 experiences @xmath92 comparisons before playing the rle of pivot .
we assume in this section that any comparison with depth @xmath251 is performed after the last comparison with depth @xmath87 .
we call step @xmath87 the set of comparisons with depth @xmath87 , and we let @xmath252 denote the number of inversions created at step @xmath87 , that is , the total number of inversions , in the output , between elements that are still in the same sublist before step @xmath87 , but are not in the same sublist after step @xmath87
. we shall need the following bound : [ l : boundk ] for every @xmath253 , @xmath254}\le \frac12 { \left(\frac23\right)}^k n^2p.\ ] ] for @xmath255 , @xmath256 , and a simple calculation yields @xmath257}&= p\frac{(n-1)(n+1)}{3}-p^2\frac{(n-1)(n-2)}{6 } \le \frac13 n^2p.\end{aligned}\ ] ] for @xmath258 we find by induction , conditioning on the partition in the first step , @xmath254 } \le { \mathbb{e}\left [ \frac12 { \left(\frac23\right)}^{k-1 } ( z_{n , p}-1)^2p + \frac12 { \left(\frac23\right)}^{k-1 } ( n - z_{n , p})^2p \right]}\ ] ] and the result follows by lemma [ l:2 ] . [ bound ] set @xmath259}$ ]
. then @xmath260 by lemma [ l : boundk ] , @xmath261 .
the proofs of the first two cases are examples of the contraction method @xcite : on one hand we have more or less explicitly defined random variables @xmath262 , @xmath263 , and @xmath264 , and we know how to prove directly that they converge to @xmath265 , @xmath266
. on the other hand , we have a family @xmath29 of random variables defined by induction : @xmath267 and a random variable @xmath61 implicitly defined by the distributional identity @xmath268 in which , in some sense , @xmath269 .
then , under additional technical conditions , the convergence of the `` coefficients '' @xmath262 , @xmath264 , entails the convergence of the `` solution '' @xmath29 .
one has to prove existence and unicity of the solutions , usually as fixed points of contracting transformations in a subspace of the space of probability measures , with a suitable metric .
in the case we are interested in , ( [ masterequ1 ] ) holds and : * @xmath270 is a fixed positive integer ; * @xmath271 is a given random vector for each @xmath272 ; * @xmath273 $ ] ; * the families @xmath274 , @xmath275 , are i.i.d . and independent of @xmath276 , and and @xmath277 . given such @xmath276 we thus define , for any distributions @xmath278 , @xmath279 when , as above , the families @xmath280 , @xmath275 , are i.i.d . and independent of @xmath276 , and further @xmath281 has the distribution @xmath282
. thus ( [ masterequ1 ] ) can be written @xmath283 for ( [ masterequ2 ] ) we similarly assume * @xmath284 is a given random vector ; * the variables @xmath285 , @xmath275 are i.i.d . and independent of @xmath286 , and @xmath287 .
given such @xmath286 we define @xmath288 when the variables @xmath285 , @xmath275 are i.i.d . with distribution @xmath289 and independent of @xmath286 .
then ( [ masterequ2 ] ) can be written @xmath290 let @xmath291 be the space of probability measures @xmath292 on @xmath293 such that @xmath294 .
the space @xmath291 is endowed with the wasserstein metric @xmath295 in which @xmath289 and @xmath296 denote the distribution functions of @xmath292 and @xmath297 , @xmath298 ( resp .
@xmath299 ) denote the generalized inverses of @xmath289 and @xmath296 and , as in previous sections , @xmath21 is a uniform random variable @xcite . since @xmath300 ( resp .
@xmath301 ) has distribution @xmath292 ( resp .
@xmath297 ) , the infimum is attained in relation ( [ metric ] ) .
the metric @xmath153 makes @xmath291 a complete metric space .
convergence of @xmath302 to @xmath303 in @xmath291 is equivalent to convergence of @xmath304 to @xmath61 in distribution _ and _ @xmath305}={\mathbb{e}\left[|x|\right]}.\ ] ] therefore convergence in @xmath291 entails @xmath306}={\mathbb{e}\left[x\right]}.\ ] ] we refer to @xcite for an extensive treatment of wasserstein metrics . in what follows , we shall improperly refer to the convergence of @xmath304 to @xmath61 in @xmath291 , meaning the convergence of their distributions .
let us take care first of relation ( [ masterequ2 ] ) : [ pointfixe ] if @xmath307}<1 $ ] and @xmath308}<\infty$ ] , then @xmath309 is a strict contraction and ( [ masterequ2 ] ) has a unique solution in @xmath291 .
let @xmath310 be a coupling of random variables , with laws @xmath292 and @xmath297 , respectively , such that @xmath311}= d_1(\mu,\nu).\ ] ] let @xmath312 be @xmath270 independent copies of @xmath310 .
furthermore , assume that @xmath286 and @xmath312 are independent
. then the probability distribution of @xmath313 is @xmath314 ( resp .
@xmath315 ) and @xmath316}\\ & \le&d_1(\mu,\nu)\ \sum_{i=1}^i{\mathbb{e}\left[\big|a^{(i)}\big|\right]}.\end{aligned}\ ] ] thus @xmath309 is a contraction with contraction constant smaller than 1 .
since @xmath291 is a complete metric space , this implies that @xmath309 has a unique fixed point in @xmath291 , by banach s fixed point theorem .
we prove now a theorem which is a variant of those used by the previously cited authors : the difference is not deep , but here we deal with family of laws , not sequences , as we have two parameters , @xmath2 and @xmath4 . as a consequence , to cover theorems [ p = c ] and [ npinfini ] , it will be convenient in their proofs to consider convergence with respect to a _
_ @xmath317 on @xmath318 $ ] , see ( * ? ? ?
* chap . 1 ) .
the collection of sets @xmath319}\cap\left[0,1\right ] ) , \hspace{1 cm } n\ge 0,\ \varepsilon>0,\ ] ] is a basis for the filter @xmath320 corresponding to theorem [ p = c ] , while @xmath321 is a basis for the filter @xmath322 corresponding to theorem [ npinfini ] . [ th de convergence ] suppose that holds for @xmath323 and @xmath324 ; i.e.@xmath325 for @xmath323 and @xmath326 , where @xmath327 . if * @xmath328}\right)}_{n , p}$ ] is bounded , * @xmath307}<1,$ ] * @xmath329@xmath330 * @xmath331 } = 0,\qquad\forall v\in\mathcal f,$ ] then @xmath29 converges in distribution to @xmath289 , the unique solution of the equation @xmath332 in @xmath291 .
more precisely , @xmath333 along @xmath317 .
we need a lemma before proving theorem [ th de convergence ] .
[ lemme rosler ] assume that three families of nonnegative numbers @xmath334 , @xmath335 , and @xmath336 satisfy the inequalities : @xmath337 let @xmath317 be a filter . under the following assumptions : * @xmath338 is nonnegative and bounded , * for some @xmath339 and some @xmath340 , @xmath341 , * @xmath342 , * @xmath343 we have @xmath344 the proof is a variant of the proof of ( * ? ? ?
* proposition 3.3 ) .
let @xmath345 be a bound for @xmath338 , and let @xmath346 for any @xmath347 , let @xmath348 be such that for @xmath349 , @xmath350 then for @xmath351 we have @xmath352 taking lim sups , we obtain that for any @xmath347 , @xmath353 thus @xmath354 , and so @xmath355 .
we can choose @xmath285 and the family @xmath356 in such a way that @xmath357}=d_1(g_{k , p},f),\ ] ] and we can also choose the families @xmath358 to be i.i.d . then @xmath359}\\ & \leq\sum_{k=0}^{n-1 } { \mathbb{e}\left[\sum_{i=1}^{i}\big|a_{n , p}^{(i)}\bbone_{z_{n , p}^{(i)}=k}\big|\right ] } { \mathbb{e}\left[\big|x_{k , p}^{(i)}-x^{(i)}\big|\right ] } + b_{n , p}\\ & \leq\sum_{k=0}^{n-1}\gamma_{k , n , p}\,d_1(g_{k , p},f)+b_{n , p}\end{aligned}\ ] ] with @xmath360}+{\mathbb{e}\left[{\left|t_{n , p}-t\right|}\right]},\\ \gamma_{k , n , p}&=&\sum_{i=1}^{i}{\mathbb{e}\left[\big|a_{n , p}^{(i)}\big| \bbone_{z_{n , p}^{(i)}=k}\right]}.\end{aligned}\ ] ] let @xmath345 be a bound for @xmath328}\right)}_{n , p}$ ] , and set @xmath361 let us check the assumptions of lemma [ lemme rosler ] : @xmath362}+{\mathbb{e}\left[x\right]}\le m + { \mathbb{e}\left[x\right]}\ ; \ ] ] for the second assumption of lemma [ lemme rosler ] , @xmath363 } = \sum_{i=1}^{i}{\mathbb{e}\left[\big|a^{(i)}\big|\right]}<1\ ; \ ] ] @xmath364 by assumption iii ) , as @xmath365}&=&{\mathbb{e}\left[x\right]}\sum_{i=1}^{i}{\mathbb{e}\left[\big|a^{(i)}_{n , p}-a^{(i)}\big|\right]}\ ; \end{aligned}\ ] ] finally @xmath366}.\ ] ] therefore @xmath367 vanishes along @xmath317 and the proof of the theorem is now complete .
the following theorem is folklore .
it gives the means and variances in theorems [ p = c ] and [ npinfini ] , after some computations .
[ usefulthmaboutvariance ] suppose that holds , where @xmath368}<1 $ ] and @xmath369}<\infty$ ] ; in other words , @xmath370 , where @xmath289 is the unique solution in @xmath291 to @xmath332 . then @xmath371 } = \frac{{\mathbb{e}\left[t\right]}}{1-\sum_i { \mathbb{e}\left[a^{(i)}\right]}}.\ ] ] moreover , if further @xmath372}<1 $ ] and @xmath373}<\infty$ ] , then @xmath374}<\infty$ ] and @xmath375}+ 2\,{\mathbb{e}\left[x\right ] } { \mathbb{e}\left[t\sum_i a^{(i)}\right ] } + { \mathbb{e}\left[{\left(\sum_i a^{(i)}\right)}^2 - 1\right]}({\mathbb{e}\left[x\right]})^2 } { 1-\sum_i { \mathbb{e}\left[a^{(i)\,2}\right]}}.\ ] ] taking expectations in ( [ masterequ2 ] ) we obtain @xmath376}= \sum_i { \mathbb{e}\left[a^{(i)}\right ] } { \mathbb{e}\left[x\right ] } + { \mathbb{e}\left[t\right]}$ ] , which yields ( [ noel ] ) .
for the second part , let @xmath377},\ , \int x^2\,d\mu(x)<\infty\}$ ] .
it is easy to see that now @xmath309 is a strict contraction in @xmath378 with the @xmath379 metric ; hence @xmath309 has a unique fixed point in @xmath378 . since @xmath380 , this fixed point
must be @xmath289 , which shows that @xmath374 } < \infty$ ] .
if we square and take the expectation , we obtain @xmath381 } & = { \mathbb{e}\left[\sum_{i=1}^i a^{(i)\,2}\right ] } { \mathbb{e}\left[x^2\right ] } + \sum_{1\le i\neq j\le i } { \mathbb{e}\left[a^{(i)}a^{(j)}\right ] } ( { \mathbb{e}\left[x\right]})^2\\ & \hspace{3.5cm}+ 2 \sum_{i=1}^i { \mathbb{e}\left[a^{(i)}t\right ] } { \mathbb{e}\left[x\right ] } + { \mathbb{e}\left [ t^2\right]},\end{aligned}\ ] ] which yields .
we apply theorem [ th de convergence ] to the distributional identity ( [ eq en x ] ) , with @xmath382 , @xmath383 and @xmath384 here the distribution of @xmath385 does not depend on @xmath249 .
we verify the assumptions ii)iv ) of theorem [ th de convergence ] for theorems [ p = c ] and [ npinfini ] together ; for the second theorem take @xmath66 .
the first assumption holds true by proposition [ bound ] .
we have @xmath386 ^ 2,\\ a^{(2)}&=b=[(2c-1)u+1-c]^2,\end{aligned}\ ] ] and @xmath387 $ ] .
easy computations give @xmath388 ^ 2\right]}+{\mathbb{e}\left[[(2c-1)u+1-c]^2\right]}=\frac23(1-c+c^2)\le \frac 23.\ ] ] we must prove the convergence of @xmath389 , @xmath390 and @xmath264 to @xmath391 , @xmath392 and @xmath41 , in @xmath393 . recall . from proposition [ description ] we know that , conditioned on @xmath21 , @xmath394 and thus @xmath395 hence , taking the expectation , @xmath396 and thus @xmath397 consequently , @xmath398 and , similarly but more sharply , @xmath399 similarly , @xmath400 and @xmath401 from , and follows @xmath402 it follows easily from cauchy
schwarz s inequality that multiplication is a continuous bilinear map @xmath403 .
hence yields @xmath404 verifying the first assertion .
similarly implies @xmath405 too . for @xmath264
we first observe that , similarly , from and , @xmath406 moreover , since @xmath407 , and imply @xmath408 and @xmath409 . for the terms @xmath206 and
@xmath410 we use proposition [ invb&w ] .
we have @xmath411 and thus , uniformly for @xmath412 , @xmath413 which , using cauchy
schwarz again , yields @xmath414 moreover , let @xmath415 . then @xmath416 and @xmath417 and thus @xmath418 proposition [ invb&w ] now yields , by and , uniformly for @xmath419 , @xmath420 consequently , using proposition [ description ] , @xmath421 collecting the various terms above , we find @xmath422 .
as already noticed at the beginning of the section , the distribution of @xmath423 does not depend on @xmath424 , so in order to prove the two theorems , we only have to check that the fourth assumption holds for @xmath425 , for an arbitrary set in each of the two filters : @xmath426 } = 0 , \qquad \forall v\in\mathcal f_i , \forall i\in\{1,2\ } ; \ ] ] also , the expectation on the left hand side of ( [ equation_filter ] ) is decreasing in @xmath427 , so we need only to check ( [ equation_filter ] ) for typical elements of the filters basis . but for @xmath428 ( resp . for @xmath429 ) , @xmath430 } & \le { \left(\frac{n-1}n\right)}^2,\\ { \mathbb{e}\left[\big|a_{n , p}\big|\textrm { ; } ( z_{n , p},p)\notin \widetilde v_{n,\varepsilon}\right ] } & \le { \left(\frac{n}{np}\right)}^2.\end{aligned}\ ] ] the proof of these theorems is done in four steps : 1 .
we prove that @xmath97 defined at ( [ def de i ] ) is almost surely finite , and has exponentially decreasing distribution tail .
thus it has moments of all orders .
2 . with the help of a poisson point process representation of quicksort
, we prove the convergence of certain copies of @xmath29 to a copy of @xmath97 for the norm @xmath431 .
this entails the weak convergence .
we prove that @xmath97 satisfies the functional equation ( [ functional1 ] ) , and that ( [ functional1 ] ) has a unique solution under the extra assumptions in theorem [ thm : unique ] .
4 . we compute the first and second moments of @xmath97 , as required for the proof of theorem [ p=1/n ] , and we also give an induction formula for moments of larger order . in this section ,
we prove some properties of the family of random variables @xmath432 defined by .
recall that the increasing sequence @xmath433 , defined by the recurrence relation ( [ recurrence_sur_u ] ) , splits @xmath22 $ ] in @xmath88 intervals , obtained recursively by breaking each of the @xmath434 intervals of the previous step into two random pieces . for
@xmath435 and @xmath436 , let @xmath437 we begin with a simple estimate ( see also @xcite ) : [ l : erika ] @xmath438}= 3^{-k}$ ] .
the length @xmath439 is the product of @xmath87 independent random variables , each uniform on @xmath22 $ ]
. hence @xmath438}={\left({\mathbb{e}\left[u^2\right]}\right)}^k=3^{-k}$ ] .
[ bigg ] for @xmath440 , @xmath441 is a @xmath442-martingale , and @xmath443}=1 $ ] .
clearly @xmath444}=1 $ ] . also :
@xmath445 } & = { \left(\frac{1+\alpha}2\right)}^{k+1}\sum_{i=1}^{2^k}{\mathbb{e}\left[w^{\alpha}_{k+1,2i-1 } + w^{\alpha}_{k+1,2i}|\mathcal{f}_k\right]}\\ & = { \left(\frac{1+\alpha}2\right)}^{k+1}\sum_{i=1}^{2^k}w^{\alpha}_{k , i } { \mathbb{e}\left[{u}_{k , i}^{\alpha}+(1-{u}_{k , i})^{\alpha}\right]}\\ & = { \left(\frac{1+\alpha}2\right)}^{k}\sum_{i=1}^{2^k}w^{\alpha}_{k , i}.\end{aligned}\ ] ] let @xmath446 denote the larger real solution of the equation @xmath447 .
lemma [ bigg ] entails that [ l : moment ] @xmath448}\le \rho^{k}$ ] . clearly , @xmath449 thus , for @xmath450 , @xmath451}\le\left({\mathbb{e}\left[m_k^{\alpha}\right]}\right)^{1/\alpha}\leq { \left(\frac2{1+\alpha}\right)}^{k/\alpha}\left({\mathbb{e}\left[f_{k,\alpha}\right]}\right)^{1/\alpha } = { \left(\frac2{1+\alpha}\right)}^{k/\alpha}.\ ] ] the rate @xmath452 reaches its minimum for @xmath453 , a constant that is an old friend of quicksort and binary search trees @xcite .
this leads to the desired value for @xmath454 . a weaker form of this inequality ( for @xmath455 ) , actually sufficient for our purposes ,
is given in @xcite .
the sequence @xmath441 is a specialization of martingales that are of a great use for the study of general branching random walks , see for instance @xcite , of which binary search trees are a special case @xcite .
[ l : mean ] @xmath456}=1 $ ] . set @xmath457 . inspecting
, we see that @xmath458}=\frac1 2\ \sum_{k\ge 1}{\left(\frac23\right)}^{k}f_{k,2},\ ] ] because , conditionally given @xmath459 , the expected number of points of @xmath133 is @xmath460 and each of them has an expected contribution @xmath461 to @xmath97 . as a consequence of lemma [ l :
moment ] , we have [ tail ] for each fixed @xmath462 , the distribution tail @xmath154 decreases exponentially fast .
equivalently , we prove this result for @xmath463 . since @xmath464 we have @xmath465 where @xmath466 is a poisson random variable with mean @xmath17 .
we split the tail of this bound on @xmath467 as follows : @xmath468 in which @xmath469 we have , by the standard chernoff bound for the poisson distribution , @xmath470 the last inequality holding only for @xmath471 .
also @xmath472 using a markov first moment inequality to bound both terms . for any @xmath473 in @xmath162 , the choice @xmath474 leads to an exponential decrease of the tail .
we assume that the input list for quicksort contains the integers @xmath475 in random order .
we model our error - prone quicksort as follows using the variables @xmath476 and @xmath74 in section [ s : results ] , but with the intensity @xmath17 of @xmath74 replaced by @xmath477 : in the first step , we use the pivot @xmath478 and let for each @xmath249 ( except the pivot ) there be an error in the comparison of @xmath249 and the pivot if @xmath479 \neq\emptyset$ ] .
( recall that @xmath126 . )
note that our choice of @xmath480 yields the right error probability @xmath4 .
let @xmath481 be the position of the pivot after the first step .
( this position was earlier denoted @xmath190 ; it may differ from @xmath482 because of errors . )
the items of the left sublist will thus be placed in positions @xmath483 and those in the right sublist in positions @xmath484 .
let @xmath485 and @xmath486 .
when the @xmath87-th step begins , we have a set of @xmath434 sublists @xmath487 , the elements of @xmath488 being in positions @xmath489 , @xmath490 ( with the convention that the sublist is empty when @xmath491 ) . in each nonempty such sublist we choose as pivot the item with rank @xmath492 in this sublist , so that its position in the final output will be exactly @xmath493 in case no errors occurs while processing the sublist .
we assume an error is made when comparing the element at position @xmath249 with the pivot @xmath494 if @xmath495 \neq\emptyset$ ]
. let @xmath496 .
let @xmath497 be the position of the pivot @xmath494 after the comparisons ( as in the first step , @xmath497 may differ from @xmath494 because of errors ) ; let @xmath498 if the sublist was empty .
set @xmath499 we expect @xmath500 and @xmath501 to converge to @xmath85 as @xmath502 .
this procedure ( stopped when there are no more nonempty sublists ) is an exact simulation of the erratic quicksort , so we may assume that @xmath8 is the number of inversions created by it .
as in section [ functequns ] , let @xmath252 be the number of inversions created at step @xmath87 , so @xmath503 we will prove that , using the notation of , @xmath504 for each @xmath87 . since also , by lemmas [ l : boundk ] and [ l : erika ] , @xmath505 } + \frac1{{\lambda}(n , p ) } { \mathbb{e}\left[\sum_{j=1}^{2^k } \sum_{x\in\pi_{k , j}}|x - x_{k , j}|\right]}\\ & = \frac 1{n^2p } { \mathbb{e}\left[i^{(k)}(n , p ) \right ] } + { \mathbb{e}\left [ \frac12 \sum_{j=1}^{2^k}w_{k , j}^2\right ] } \le { \left(\frac23\right)}^k,\end{aligned}\ ] ] it follows by dominated convergence that , using , @xmath506 moreover , @xmath507 , and it follows easily from that @xmath508 .
hence we have @xmath509 , which proves the convergence .
it remains to verify .
set @xmath510 relation is equivalent to @xmath511 for simplicity , we write in the sequel @xmath17 instead of @xmath480 .
we begin with a lemma .
[ ll ] for each @xmath87 and @xmath130 , @xmath512 recall that @xmath513 , so translates to @xmath514 we use induction on @xmath87 . comparing the definitions of @xmath85 and @xmath501
, we see that it suffices to consider an odd @xmath515 , and in that case there are three sources of a difference : 1 . the differences between @xmath516 and @xmath517 and between @xmath518 and @xmath519 .
by the induction hypothesis , this contributes at most @xmath520 .
2 . the @xmath521 inside ( and the rounding by ) the ceiling function .
this contributes at most @xmath15 .
3 . the shift of the pivot , from @xmath494 to @xmath497 , caused by the erroneous comparisons .
the shift is bounded by the total number of errors at step @xmath87 , so its mean is less than @xmath17 , and the contribution is less than @xmath522 .
we return to proving . for @xmath255
, @xmath523 is just @xmath197 studied in section [ functequns ] , and yields @xmath524 let @xmath525 be the set of items @xmath249 such that an error was made in the comparison with @xmath482 .
relation entails that @xmath526 we shall denote this last sum @xmath527 .
thus , we have @xmath528 furthermore @xmath529}&=(m-1)p(1-p)+((m-1)p)^2\le np+n^2p^2 . \label{boundagain}\end{aligned}\ ] ] hence , @xmath530 moreover , @xmath531 differs from @xmath532 in in four ways only ( recall that @xmath533 ) : 1 .
@xmath534 differs from @xmath93 by at most @xmath15 .
since the expected number of terms is not larger than @xmath17 , this gives a contribution @xmath535 .
2 . @xmath536 , which by lemma [ ll ] has expectation @xmath535 .
thus this too gives a contribution @xmath535 .
if there are two or more points in @xmath479 $ ] for some @xmath249 , @xmath537 contains more terms than @xmath538 .
it is easily seen that the expected number of such extra points in each interval @xmath539 $ ] is less than @xmath540 , and each point contributes at most 1 to @xmath537 .
each point in @xmath541 $ ] contributes for an extra term in @xmath537 again .
the expected number of such extra points is @xmath522 and each of these terms contributes at most 1 to @xmath537 .
this verifies for @xmath255 . for @xmath542
we argue similarly .
we can approximate @xmath252 by the sum of the distances between the errors and the respective pivots , @xmath543 as follows : let @xmath544 be the set of items @xmath545 subject to error when compared with @xmath546 , and let @xmath547 be the @xmath548-algebra generated by @xmath549 and @xmath550 .
as for @xmath255 , using relation , we obtain the following bound : @xmath551 } & \le \sum_{j\le 2^{k-1}}{\left(p^2{\left(\#\ell_{k-1,j}\right)}^2+p\#\ell_{k-1,j}\right)}\\ & \le 2^{k-1}(n^2p^2+np ) = { \mathcal o{\left(1\right)}},\end{aligned}\ ] ] and as a consequence , @xmath552 now , @xmath553 differs from @xmath554 in in the same four ways as for @xmath255 , plus an extra fifth way : 1 .
see the case @xmath255 .
2 . @xmath555 , which by lemma [ ll ] has expectation @xmath535 .
thus this too gives a contribution @xmath535 .
3 . two or more points in @xmath495 $ ] for some @xmath249 , see the case @xmath255 .
each point in @xmath556 $ ] contributes for an extra term in @xmath557 .
the expected number of such extra points is @xmath558 and each of these terms contributes at most 1 to @xmath557 .
there is a new source of error in this approximation , because some points @xmath93 in @xmath559 and the corresponding positions @xmath560 belong to subintervals that do not correspond to each other , because the endpoints @xmath561 differ somewhat from @xmath562 . by lemma [ ll ] ,
the expected number of such cases is @xmath535 , so again we get a contribution of order @xmath535 only .
this verifies and thus the convergence of @xmath29 to @xmath97 .
we check that @xmath97 satisfies the distributional identity and some side conditions needed for the computations of moments .
[ ilexiste1 ] @xmath432 is a solution of .
moreover , @xmath563}<\infty$ ] and @xmath564}\to0 $ ] as @xmath565 , for @xmath323 .
all moments are finite by theorem [ tail ] .
moreover , @xmath566}\to0 $ ] as @xmath565 by and dominated convergence . for @xmath567 ,
let @xmath568 be a poisson point process of intensity @xmath17 on @xmath569 $ ] , and let @xmath570 be independent uniform random variables as in section [ s : results ] , and further independent of @xmath568 .
define @xmath571 and @xmath572 as in section [ s : results ] , with the slight modification @xmath573 and set @xmath574 note that @xmath575 . shifting and rescaling @xmath568 , we obtain @xmath576
let us split @xmath97 : we have @xmath577 we see , using general properties of poisson point processes and the recursive construction of @xmath571 , that @xmath578 in the sense that , conditionally given that @xmath579 , @xmath580 , @xmath581 and @xmath582 are independent and distributed as @xmath583 , @xmath584 , @xmath585 , respectively .
also @xmath586 is uniformly distributed on @xmath22 $ ] .
let @xmath111 and @xmath587 be two solutions of in @xmath120 .
let @xmath113 and @xmath588 denote two measurable processes representing respectively @xmath292 and @xmath589 , in the sense of remark [ remmesure ] ( i ) . without loss of generality , we can assume that @xmath117 and @xmath72 share the same underlying probabilistic space , and the same exponent @xmath473 .
then , by definition of @xmath120 , for @xmath590 , @xmath591}\ ] ] is finite .
let @xmath592 denote the infimum of @xmath593 over all couples of representations @xmath594 of @xmath292 and @xmath589 , lying on the same probabilistic space , and assume that @xmath595 .
let @xmath596 be such a couple of representations , satisfying furthermore @xmath597 consider a probabilistic space on which are defined three _
independent _ random variables @xmath598 , @xmath599 and @xmath21 , @xmath598 and @xmath599 being two copies of @xmath596 , @xmath21 being uniform on @xmath162 . finally , for every @xmath462 , set @xmath600 then @xmath601 and @xmath602 are representations of @xmath292 ( resp .
@xmath589 ) and satisfy remark [ remmesure ] ( ii ) .
moreover , we have , for @xmath603 , @xmath604 } & = \mathbb{e}\left[\lambda^\alpha\left|u^2 y_1({\lambda}u ) + ( 1-u)^2 y_2({\lambda}(1-u))\right.\right.\\ & \hspace{2 cm } \left.\left.- u^2 z_1({\lambda}u ) - ( 1-u)^2 z_2({\lambda}(1-u))\right|\right ] \\ & \leq 2{\mathbb{e}\left[\lambda^\alpha u^2\left|y_1({\lambda}u)-z_1({\lambda}u)\right|\right ] } \\ & \leq 2\int_0 ^ 1 u^{2-\alpha}\ { \mathbb{e}\left[(\lambda u)^\alpha \left|y_1({\lambda}u)-z_1({\lambda}u)\right|\right ] } du \\ &
< \delta,\end{aligned}\ ] ] leading to a contradiction .
the aim of this section is the computation of moments of @xmath97 , completing the proof of theorem [ p=1/n ] .
if one uses directly , the computations of moments by induction are hardly tractable because all three terms on the right of depend on @xmath21 . to circumvent this problem , we consider a new distributional identity @xmath605 in which * @xmath606 is as in section [ s : results ] ; equivalently , @xmath607 * @xmath606 and @xmath608 are independent ; * conditionally , given @xmath100 , @xmath609 and @xmath610 are independent and distributed as @xmath611 and @xmath612 , respectively .
the next propositions establish relations between @xmath97 and solutions of , eventually providing an algorithm for the computation of moments of @xmath97 ( see and ) .
[ 5to34 ] the family @xmath613 , in which @xmath606 and @xmath97 are assumed independent , is a solution of .
[ unique2 ] the @xmath2-th moment of @xmath116 is a polynomial of degree @xmath2 in the variable @xmath17 . before proving propositions [ 5to34 ] and [ unique2 ]
, we need a lemma .
[ factsaboutw ] the @xmath2-th moment @xmath614}$ ] is a polynomial of degree @xmath2 with nonnegative coefficients and for @xmath615 , @xmath616 .
owing to campbell s theorem @xcite , we have @xmath617 } = \exp { \left ( { \lambda}{\left ( { \mathbb{e}\left[e^{su}\right ] } -1\right ) } \right ) } = \exp { \left ( { \lambda}{\left ( \frac s{2!}+\frac { s^2}{3!}+\dots\right ) } \right)}.\ ] ] expanding the last expression gives the lemma . to show , it is enough to show @xmath618 where , as usual , conditioned on @xmath100 , the terms on the right hand side are independent with the right distributions .
this follows immediately from , since @xmath619 consider the sequence of integral equations @xmath620 in which @xmath621 where @xmath622 is the @xmath178-th moment of @xmath623 .
proposition [ unique2 ] is a consequence of the next lemma . the induction formula and the initial condition @xmath624 defines a unique sequence of polynomials , @xmath625 .
furthermore , @xmath626 has degree @xmath2 , and vanishes at @xmath222 . for @xmath627 , the @xmath2-th moment @xmath628}$ ]
is equal to @xmath629 .
consider @xmath627 and assume that the properties in the lemma hold for @xmath630 .
then , for @xmath87 and @xmath177 smaller than @xmath2 , and @xmath631 , the expression @xmath632 is a polynomial with degree @xmath2 and , due to lemma [ factsaboutw ] , vanishes at 0 .
thus , in this case , @xmath633 is a polynomial with degree @xmath2 , vanishing at 0 .
it is now easy to check that a polynomial @xmath629 satisfies if and only if , for @xmath634 , @xmath635}p_n=\frac{n+k+1}{n+k-1}\ { \left[{\lambda}^k\right]}\psi_n.\ ] ] also , by the induction assumptions , @xmath636 } & = { \mathbb{e}\left[\big(\xi({\lambda})+uy(u{\lambda})+(1-u)\widetilde{y}((1-u){\lambda})\big)^n\right]}\\ & = \sum_{r+k+\ell = n } \binom{n}{r , k,\ell}g_r({\lambda } ) { \mathbb{e}\left[u^k(1-u)^{\ell}y(u{\lambda})^k\widetilde{y}((1-u){\lambda})^{\ell}\right]}\\ & = 2{\mathbb{e}\left[u^ny({\lambda}u)^n\right]}+\psi_n({\lambda}).\end{aligned}\ ] ] note that @xmath637 for @xmath638 . by remark
[ remmesure ] , @xmath639}$ ] is nonnegative and measurable .
thus , for @xmath83 , we can rewrite the previous equation : @xmath640 since @xmath641 is assumed to be finite and @xmath637 , the integral on the right hand side is convergent , and thus it is a continuous function of @xmath17 . as a consequence @xmath642 belongs to @xmath643 , and is a solution on @xmath123 of the following differential equation : @xmath644 by proposition [ ilexiste1 ] and lemma [ factsaboutw ] , @xmath645 as @xmath565 , but the general solution of the differential equation is @xmath646 .
thus @xmath647 on @xmath123 . as a consequence of these results
, we deduce that : [ polynomialprop ] the function @xmath648}$ ] is a polynomial of degree @xmath2 that vanishes at @xmath222 . since @xmath649 , with independent summands ,
we obtain @xmath650 } = \sum_{0\le k\le m}{m\choose k}{\mathbb{e}\left[\xi({\lambda})^{m - k}\right]}{\lambda}^k{\mathbb{e}\left [ x({\lambda})^k\right]}.\ ] ] the result follows by induction . the moments of @xmath116 , and thus of @xmath97 , can be computed up to arbitrary order with the help of and . for the first two moments , the calculations run as follows . expanding @xmath651 in the proof of lemma [ factsaboutw ]
we obtain [ l : xi ] @xmath652}=\frac12{\lambda}\qquad\text{and}\qquad g_2({\lambda})={\mathbb{e}\left[\xi({\lambda})^2\right]}=\frac13{\lambda}+\frac14{\lambda}^2.\ ] ] @xmath653}={\mathbb{e}\left[\xi({\lambda})\right]}={\lambda}\qquad\text{and}\qquad { \lambda}^2{\textnormal{var}\left(x({\lambda})\right)}={\textnormal{var}\left(\xi({\lambda})\right)}=\frac13{\lambda}+\frac{1}{12}{\lambda}^2.\ ] ] taking @xmath654 in and , we find , using lemma [ l : xi ] , @xmath655 taking @xmath656 , we similarly find @xmath657 since @xmath649 , with independent summands , @xmath658}={\mathbb{e}\left[\xi({\lambda})\right]}+{\lambda}{\mathbb{e}\left [ x({\lambda})\right]},\ ] ] which by lemma [ l : xi ] yields @xmath659}={\lambda}$ ] . similarly , @xmath660 } = p_2({\lambda})-{\mathbb{e}\left[\xi({\lambda})^2\right]}-2{\mathbb{e}\left[\xi({\lambda})\right]}{\mathbb{e}\left[{\lambda}x({\lambda})\right ] } = \tfrac13{\lambda}+\tfrac{13}{12}{\lambda}^2,\end{aligned}\ ] ] which yields the variance formula . the formulas for mean and variance of @xmath97 can also be obtained directly from and lemma [ l : xi ] ; we leave this as an exercise .
we have presented a probabilistic analysis of quicksort when some comparisons can err .
analysing other sorting algorithms such as merge sort , insertion sort or selection is even more intricate .
they do not fit into the model presented in this paper and further more involved probabilistic models / arguments are required .
we conjecture that the same normalization holds for the number of inversions in the _ output of merge sort _ for @xmath661 , @xmath662 , and that the limit law @xmath663 satisfies @xmath664}=\sum_{k\ge 0}\ \frac{2^k}{(2^k+2)(2^k+3)}= 0.454674373\dots<{\mathbb{e}\left[x({\lambda})\right]}.\ ] ]
a more rigorous formulation and a shorter proof of theorem [ thm : unique ] are born from discussions with uwe rsler .
also , we thank two anonymous referees , whose careful reading led to substantial improvements . | we provide a probabilistic analysis of the output of quicksort when comparisons can err . |
first - order temporal logic ( @xmath0 ) has been shown to be a powerful formalism for expressing sophisticated dynamic properties .
unfortunately , this power also leads to strong intractability .
recently , however , a fragment of @xmath0 , called _ monodic _ @xmath0 , has been investigated , both in terms of its theoretical @xcite and practical @xcite properties .
essentially , monodicity allows for one free variable in every temporal formula .
although clearly restrictive , this fragment has been shown to be useful in expressive description logics , infinite - state verification , and spatio - temporal logics @xcite .
we here develop a new temporal logic , combining decidable fragments of monodic @xmath0 @xcite with recent developments in xor temporal logics @xcite , and apply this to the verification of parameterised systems .
we use a communicating finite state machine model of computation , and can specify not only basic synchronous , parameterised systems with instantaneous broadcast communication @xcite , but the powerful temporal language allows us also to specify asynchronously executing machines and more sophisticated communication properties , such as delayed delivery of messages .
in addition , and in contrast to many other approaches @xcite , not only safety , but also liveness and fairness properties , can be verified through automatic deductive verification .
finally , in contrast to work on regular model checking @xcite and constraint based verification using counting abstraction @xcite , the logical approach is both complete and decidable .
the verification of concurrent systems often comes down to the analysis of multiple finite - state automata , for example of the following form . in describing such automata , both automata - theoretic and logical approaches may be used . while _ temporal logic
_ @xcite provides a clear , concise and intuitive description of the system , automate - theoretic techniques such as _ model checking _
@xcite have been shown to be more useful in practice .
recently , however , a propositional , linear - time temporal logic with improved deductive properties has been introduced @xcite , providing the possibility of practical deductive verification in the future .
the essence of this approach is to provide an xor constraint between key propositions .
these constraints state that exactly one proposition from a xor set can be true at any moment in time .
thus , the automaton above can be described by the following clauses which are implicitly in the scope of a ` @xmath1 ' ( ` always in the future ' ) operator .
@xmath2 here ` @xmath3 ' is a temporal operator denoting ` at the next moment ' and ` @xmath4 ' is a temporal operator which holds only at the initial moment in time .
the inherent assumption that at any moment in time exactly one of @xmath5 , @xmath6 , @xmath7 or @xmath8 holds , is denoted by the following .
@xmath9 with the complexity of the decision problem ( regarding @xmath5 , @xmath6 , etc ) being polynomial , then the properties of any finite collection of such automata can be tractably verified using this _ propositional _ xor temporal logic . however , one might argue that this deductive approach , although elegant and concise , is still no better than a model checking approach , since it targets just _ finite _ collections of ( finite ) state machines .
thus , this naturally leads to the question of whether the xor temporal approach can be extended to _ first - order temporal logics _ and , if so , whether a form of tractability still applies . in such an approach
, we can consider _ infinite _ numbers of finite - state automata ( initially , all of the same structure ) .
previously , we have shown that @xmath0 can be used to elegantly specify such a system , simply by assuming the argument to each predicate represents a particular automaton @xcite .
thus , in the following @xmath10 is true if automaton @xmath11 is in state @xmath5 : @xmath12 thus , @xmath0 can be used to specify and verify broadcast protocols between synchronous components @xcite . in this paper
we define a logic , , which allows us to not only to specify and verify systems of the above form , but also to specify and verify more sophisticated asynchronous systems , and to carry out verification with a reasonable complexity . [ [ section ] ]
first - order ( discrete , linear time ) temporal logic , @xmath0 , is an extension of classical first - order logic with operators that deal with a discrete and linear model of time ( isomorphic to the natural numbers , @xmath13 ) . [ [ syntax . ] ] syntax .
+ + + + + + + the symbols used in @xmath0 are * _ predicate symbols : _ @xmath14 each of which is of a fixed arity ( null - ary predicate symbols are _ propositions _ ) ; * _ variables : _ @xmath15 ; * _ constants : _ @xmath16 ; * _ boolean operators : _ @xmath17 , @xmath18 , @xmath19 , @xmath20 , @xmath21 , @xmath22 ( ` true ' ) , @xmath23 ( ` false ' ) ; * _ first - order quantifiers : _ @xmath24 ( ` for all ' ) and @xmath25 ( ` there exists ' ) ; and * _ temporal operators : _ @xmath1 ( ` always in the future ' ) , @xmath26 ( ` sometime in the future ' ) , @xmath3 ( ` at the next moment ' ) , @xmath27 ( until ) , @xmath28 ( weak until ) , and @xmath4 ( at the first moment in time ) .
although the language contains constants , neither equality nor function symbols are allowed .
the set of well - formed @xmath0-formulae is defined in the standard way @xcite : * booleans @xmath22 and @xmath23 are atomic @xmath0-formulae ; * if @xmath29 is an @xmath30-ary predicate symbol and @xmath31 , @xmath32 , are variables or constants , then @xmath33 is an atomic @xmath0-formula ; * if @xmath34 and @xmath35 are @xmath0-formulae , so are @xmath36 , @xmath37 , @xmath38 , @xmath39 , and @xmath40 ; * if @xmath34 is an @xmath0-formula and @xmath41 is a variable , then @xmath42 and @xmath43 are @xmath0-formulae ; * if @xmath34 and @xmath35 are @xmath0-formulae , then so are @xmath44 , @xmath45 , @xmath46 , @xmath47 , @xmath48 , and @xmath4 . a _ literal _ is an atomic @xmath0-formula or its negation .
[ [ semantics ] ] semantics , + + + + + + + + + + intuitively , @xmath0formulae are interpreted in _ first - order temporal structures _ which are sequences @xmath49 of _ worlds _ , @xmath50 with truth values in different worlds being connected via temporal operators .
more formally , for every moment of time @xmath51 , there is a corresponding _ first - order _ structure , @xmath52 , where every @xmath53 is a non - empty set such that whenever @xmath54 , @xmath55 , and @xmath56 is an interpretation of predicate and constant symbols over @xmath53 .
we require that the interpretation of constants is _
rigid_. thus , for every constant @xmath57 and all moments of time @xmath58 , we have @xmath59 .
a _ ( variable ) assignment _ @xmath60 is a function from the set of individual variables to @xmath61 .
we denote the set of all assignments by @xmath62 .
the set of variable assignments @xmath63 corresponding to @xmath64 is a subset of the set of all assignments , @xmath65 ; clearly , @xmath66 if @xmath54
. the _ truth _ relation @xmath67 in a structure @xmath49 , is defined inductively on the construction of @xmath34 _ only for those assignments @xmath60 that satisfy the condition @xmath68_. see fig . [
fig : sem ] for details .
@xmath69 @xmath49 is a _ model _ for a formula @xmath34 ( or @xmath34 is _ true _ in @xmath49 ) if , and only if , there exists an assignment @xmath60 in @xmath70 such that @xmath71 .
a formula is _ satisfiable _ if , and only if , it has a model .
a formula is _ valid _ if , and only if , it is true in any temporal structure @xmath49 under any assignment @xmath60 in @xmath70 .
the models introduced above are known as _ models with expanding domains _ since @xmath72 .
another important class of models consists of _ models with constant domains _ in which the class of first - order temporal structures , where @xmath0 formulae are interpreted , is restricted to structures @xmath73 , @xmath74 , such that @xmath75 for all @xmath76 .
the notions of truth and validity are defined similarly to the expanding domain case .
it is known @xcite that satisfiability over expanding domains can be reduced to satisfiability over constant domains with only a polynomial increase in the size of formulae .
the set of valid formulae of @xmath0 is not recursively enumerable .
furthermore , it is known that even `` small '' fragments of @xmath0 , such as the _ two - variable monadic _ fragment ( where all predicates are unary ) , are not recursively enumerable @xcite .
however , the set of valid _ monodic _ formulae is known to be finitely axiomatisable @xcite .
an @xmath0-formula @xmath34 is called _ monodic _ if , and only if , any subformula of the form @xmath77 , where @xmath78 is one of @xmath3 , @xmath1 , @xmath26 ( or @xmath79 , where @xmath78 is one of @xmath27 , @xmath28 ) , contains at most one free variable .
we note that the addition of either equality or function symbols to the monodic fragment generally leads to the loss of recursive enumerability @xcite .
thus , monodic @xmath0 is expressive , yet even small extensions lead to serious problems .
further , even with its recursive enumerability , monodic @xmath0 is generally undecidable .
to recover decidability , the easiest route is to restrict the first order part to some decidable fragment of first - order logic , such as the guarded , two - variable or monadic fragments .
we here choose the latter , since monadic predicates fit well with our intended application to parameterised systems .
recall that monadicity requires that all predicates have arity of at most ` 1 ' .
thus , we use monadic , monodic @xmath0 @xcite .
a practical approach to proving monodic temporal formulae is to use _ fine - grained temporal resolution _
@xcite , which has been implemented in the theorem prover temp @xcite . in the past
, temp has been successfully applied to problems from several domains @xcite , in particular , to examples specified in the temporal logics of knowledge ( the fusion of propositional linear - time temporal logic with multi - modal s5 ) @xcite . from this work
it is clear that monodic first - order temporal logic is an important tool for specifying complex systems .
however , it is also clear that the complexity , even of _ monadic _ monodic first - order temporal logic , makes this approach difficult to use for larger applications @xcite .
an additional restriction we make to the above logic involves implicit xor constraints over predicates .
such restrictions were introduced into temporal logics in @xcite , where the correspondence with bchi automata was described , and generalised in @xcite . in both cases , the decision problem is of much better ( generally , polynomial ) complexity than that for the standard , unconstrained , logic . however , in these papers only _ propositional _ temporal logic was considered .
we now add such an xor constraint to .
the set of predicate symbols @xmath80 , is now partitioned into a set of xor - sets , @xmath81 , @xmath82 , @xmath83 , @xmath84 , with one _ non - xor _ set @xmath85 such that 1 .
all @xmath86 are disjoint with each other , 2 .
@xmath85 is disjoint with every @xmath86 , 3 .
@xmath87 @xmath88 , and 4 . for each @xmath86 ,
exactly _ one _ predicate within @xmath86 is satisfied ( for any element of the domain ) at any moment in time .
consider the formula @xmath89 where @xmath90 and @xmath91 .
the above formula states that , for any element of the domain , a , then one of @xmath92 or @xmath93 must be satisfied and one of @xmath94 , @xmath95 or @xmath96 must be satisfied . to simplify our description
, we will define a _
normal form _ into which formulae can be translated . in the following : *
@xmath97 denotes a conjunction of negated xor predicates from the set @xmath86 ; * @xmath98 denotes a disjunction of ( positive ) xor predicates from the set @xmath86 ; * @xmath99 denotes a conjunction of non - xor literals ; * @xmath100 denotes a disjunction of non - xor literals .
a _ step _ clause is defined as follows : @xmath101 a _ monodic temporal problem in divided separated normal form ( dsnf ) _
@xcite is a quadruple @xmath102 , where : 1 . the universal part , @xmath103 , is a finite set of arbitrary closed first - order formulae ; 2 .
the initial part , @xmath104 , is , again , a finite set of arbitrary closed first - order formulae ; 3 .
the step part , @xmath105 , is a finite set of step clauses ; and 4 .
the eventuality part , @xmath106 , is a finite set of eventuality clauses of the form @xmath107 , where @xmath108 is a unary literal . in what follows
, we will not distinguish between a finite set of formulae @xmath109 and the conjunction @xmath110 of formulae within the set . with each monodic temporal problem , we associate the formula @xmath111 now , when we talk about particular properties of a temporal problem ( e.g. , satisfiability , validity , logical consequences etc ) we mean properties of the associated formula .
every monodic formula can be translated to the normal form in satisfiability preserving way using a renaming and unwinding technique which substitutes non - atomic subformulae and replaces temporal operators by their fixed point definitions as described , for example , in @xcite .
a step in this transformation is the following : we recursively rename each innermost open subformula @xmath112 , whose main connective is a temporal operator , by @xmath113 , where @xmath114 is a new unary predicate , and rename each innermost closed subformula @xmath115 , whose main connective is a temporal operator , by @xmath116 , where @xmath116 is a new propositional variable . while renaming introduces new , non - xor predicates and propositions , practical problems stemming from verification are nearly in the normal form , see section [ sec : model ] .
first - order temporal logics are notorious for being of a high complexity .
even decidable sub - fragments of monodic first - order temporal logic can be too complex for practical use . for example , satisfiability of monodic monadic @xmath0 logic is known to be @xmath117-complete @xcite . however , imposing xor restrictions we obtain better complexity bounds .
[ th : complexity ] satisfiability of monodic monadic formulae ( in the normal form ) can be decided in @xmath118 time , where @xmath119 , , @xmath120 are cardinalities of the sets of xor predicates , and @xmath121 is the cardinality of the set of non - xor predicates . before we sketch the proof of this result , we show how the xor restrictions influence the complexity of the satisfiability problem for monadic first - order ( non - temporal ) logic .
[ lemma : fom ] satisfiability of monadic first - order formulae can be decided in @xmath122 , where @xmath30 is the length of the formula , and @xmath119 , , @xmath120 , @xmath121 are as in theorem [ th : complexity ] . as in @xcite , proposition 6.2.9 , the non - deterministic decision procedure first guesses a structure and then verifies that the structure is a model for the given formula .
it was shown , @xcite , proposition 6.2.1 , exercise 6.2.3 , that if a monadic first - order formula has a model , it also has a model , whose domain is the set of all _ predicate colours_. a predicate colour , @xmath123 , is a set of unary literals such that for every predicate @xmath124 from the set of all predicates @xmath125 , either @xmath124 or @xmath126 belongs to @xmath123 .
notice that under the conditions of the lemma , there are at most @xmath127 different predicate colours .
hence , the structure to guess is of @xmath128 size
. it should be clear that one can evaluate a monadic formula of the size @xmath30 in a structure of the size @xmath129 in deterministic @xmath130 time .
therefore , the overall complexity of the decision procedure is @xmath122 . for simplicity of presentation
, we assume the formula contains no propositions .
satisfiability of a monodic @xmath0 formula is equivalent to a property of the _ behaviour graph _ for the formula , checkable in time polynomial in the product of the number of different predicate colours and the size of the graph , see @xcite , theorem 5.15 . for unrestricted @xmath0 formulae ,
the size of the behaviour graph is double exponential in the number of predicates .
we estimate now the size of the behaviour graph and time needed for its construction for formulae .
let @xmath131 be a set of predicate colours and @xmath132 be a map from the set of constants , @xmath133 , to @xmath131 .
a couple @xmath134 is called a _
colour scheme_. nodes of the behaviour graph are colour schemes . clearly , there are no more than @xmath135 different colour schemes .
however , not every colour scheme is a node of the behaviour graph : a colour scheme @xmath136 is a node if , and only if , a monadic formula of first - order ( non - temporal ) logic , constructed from the given formula and the colour scheme itself , is satisfiable ( for details see @xcite ) .
a similar first - order monadic condition determines which nodes are connected with edges .
it can be seen that the size of the formula is polynomial in both cases . by lemma [ lemma : fom ] , satisfiability of monadic first - order formulae can be decided in deterministic @xmath137 time .
overall , the behaviour graph , representing all possible models , for an formula can be constructed in @xmath138 time .
in previous work , notably @xcite a parameterised finite state machine based model , suitable for the specification and verification of protocols over arbitrary numbers of processes was defined .
essentially , this uses a family of identical , and synchronously executing , finite state automata with a rudimentary form of communication : if one automaton makes a transition ( an action ) @xmath139 , then it is required that _ all _ other automata simultaneously make a complementary transition ( reaction ) @xmath140 .
in @xcite we translated this automata model into monodic @xmath0 and used automated theorem proving in that logic to verify parameterised cache coherence protocols @xcite .
the model assumed not only synchronous behaviour of the communicating automata , but instantaneous broadcast .
here we present a more general model suitable for specification of both synchronous and asynchronous systems ( protocols ) with ( possibly ) delayed broadcast and give its faithful translation into .
this not only exhibits the power of the logic but , with the improved complexity results of the previous section , provides a route towards the practical verification of temporal properties of infinite state systems .
we begin with a description of both the asynchronous model , and the delayed broadcast approach .
[ def : protocol - simple ] a protocol , p is a tuple @xmath141 , where * @xmath142 is a finite set of states ; * @xmath143 is a set of initial states ; * @xmath144 , where * * @xmath145 is a finite set of local actions ; * * @xmath146 is a finite set of broadcast actions , + i.e. `` send a message '' ; * * @xmath147 is the set of broadcast reactions , i.e. `` receive a message '' ; * @xmath148 is a transition relation that satisfies the following property @xmath149 i.e. , `` readiness to receive a message in any state '' .
further , we define a notion of global machine , which is a set of @xmath30 finite automata , where @xmath30 is a parameter , each following the protocol and able to communicate with others via ( possibly delayed ) broadcast .
to model asynchrony , we introduce a special automaton action , @xmath150 , meaning the automaton is not active and so its state does not change . at any moment
an arbitrary group of automata may be idle and all non - idle automata perform their actions in accordance with the transition function @xmath151 ; different automata may perform different actions .
[ def : glob_mach1 ] given a protocol , @xmath152 , the global machine @xmath153 of dimension _
@xmath30 _ is the tuple @xmath154 , where * @xmath155 * @xmath156 * @xmath157 is a transition relation that satisfies the following property @xmath158\ , .
\end{array}\ ] ] * @xmath159 is a communication environment , that is a set of possible sets of messages in transition .
an element @xmath160 is said to be a global configuration of the machine .
a run of a global machine @xmath153 is a possibly infinite sequence @xmath161 of global configurations of @xmath153 satisfying the properties ( 1)(6 ) listed below . in this formulation
we assume @xmath162 and @xmath163 .
1 . @xmath164 + ( `` initially all automata are in initial states '' ) ; 2 .
@xmath165 + ( `` initially there are no messages in transition '' ) ; 3 . @xmath166 + ( `` an arbitrary part of the automata can fire '' ; 4 .
@xmath167 + ( `` delivery to all participants is guaranteed '' ) ; 5 .
@xmath168 $ ] ( `` one can receive only messages kept by the environment , or sent at the same moment of time '' ) in order to formulate further requirements we introduce the following notation : @xmath169 @xmath170 @xmath171 then , the last requirement the run should satisfy is 1 .
@xmath172 [ [ example - asynchronous - floodset - protocol . ] ] example : asynchronous floodset protocol .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we illustrate the use of the above model by presenting the specification of an asynchronous floodset protocol in our model .
this is a variant of the _ floodset algorithm with alternative decision rule _ ( in terms of @xcite , p.105 ) designed for solution of the consensus problem .
the setting is as follows .
there are @xmath30 processes , each having an _ input bit _ and an _
output bit_. the processes work asynchronously , run the same algorithm and use _ broadcast _ for communication .
the broadcasted messages are guaranteed to be delivered , though possibly with arbitrary delays .
( the process is described graphically in fig .
[ fig : flood ] . ) , scaledwidth=40.0% ] the goal of the algorithm is to eventually reach an agreement , i.e. to produce an output bit , which would be the same for all processes .
it is required also that if all processes have the same input bit , that bit should be produced as an output bit .
the asynchronous floodset protocol we consider here is adapted from @xcite .
main differences with original protocol are : * the original protocol was synchronous , while our variant is asynchronous ; * the original protocol assumed instantaneous message delivery , while we allow arbitrary delays in delivery ; and * although the original protocol was designed to work in the presence of crash ( or fail - stop ) failures , we assume , for simplicity , that there are no failures . because of the absence of failures the protocol is very simple and unlike the original one does not require `` retransmission '' of any value
. we will show later ( in section [ sec : var ] ) how to include the case of crash failures in the specification ( and verification ) .
thus , the asynchronous floodset protocol is defined , informally , as follows .
* at the first round of computations , every process broadcasts its input bit .
* at every round the ( tentative ) output bit is set to the minimum value ever seen so far .
the correctness criterion for this protocol is that , eventually , the output bits of all processes will be the same .
now we can specify the asynchronous floodset as a protocol @xmath173 , where @xmath174 ; @xmath175 ; @xmath176 with @xmath177 , @xmath178 , @xmath179 . the transition relation @xmath180 @xmath181 @xmath182 @xmath183 @xmath184 @xmath185 .
given a protocol @xmath186 , we define its translation to as follows . for each @xmath187 , introduce a monadic predicate symbol @xmath188 and for each @xmath189 introduce a monadic predicate symbol @xmath190 .
for each @xmath191 we introduce also a propositional symbol @xmath192 . intuitively , elements of the domain in the temporal representation will represent exemplars of finite automata , and the formula @xmath193 is intended to represent `` automaton x is in state @xmath194 '' .
the formula @xmath195 is going to represent `` automaton @xmath41 performs action @xmath196 '' .
proposition @xmath192 will denote the fact `` message @xmath196 is in transition '' ( i.e. it has been sent but not all participants have received it . ) because of intended meaning we define two xor - sets : @xmath197 and @xmath198 .
all other predicates belong to the set of non - xor predicates .
we define the temporal translation of @xmath199 , called @xmath200 , as a conjunction of the formulae in fig .
[ fig : trans ] .
note that , in order to define the temporal translation of requirement ( 6 ) above , ( on the dynamics of environment updates ) we introduce the unary predicate symbol @xmath201 for every @xmath202 .
we now consider the correctness of the temporal translation .
this translation of protocol @xmath203 is faithful in the following sense .
[ prop : trans ] given a protocol , @xmath199 , and a global machine , @xmath153 , of dimension @xmath30 , then any temporal model @xmath204 of @xmath200 with the finite domain @xmath205 of size @xmath30 represents some run @xmath161 of @xmath153 as follows : @xmath206 is @xmath207-th configuration of the run iff @xmath208 , @xmath209 and @xmath210 dually , for any run of @xmath153 there is a temporal model of @xmath200 with a domain of size @xmath30 representing this run . by routine inspection of the definitions of runs , temporal models and the translation .
the above model allows various modifications and corresponding version of proposition [ prop : trans ] still holds .
[ [ determinism . ] ] determinism .
+ + + + + + + + + + + + the basic model allows non - deterministic actions . to specify the case of deterministic actions only
, one should replace the `` action effects '' axiom in fig .
[ fig : trans ] by the following variant : @xmath211\ ] ] for all @xmath212 [ [ explicit - bounds - on - delivery . ] ] explicit bounds on delivery .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + in the basic mode , no explicit bounds on delivery time are given .
to introduce bounds one has to replace the `` guarantee of delivery '' axiom with the following one : @xmath213\ ] ] for all @xmath214 and some @xmath30 ( representing the maximal delay ) .
[ [ finite - bounds - on - delivery . ] ] finite bounds on delivery . + + + + + + + + + + + + + + + + + + + + + + + + + + one may replace the `` guarantee of delivery '' axiom with the following one @xmath215\ ] ] for all @xmath214 .
[ [ crashes . ] ] crashes .
+ + + + + + + + one may replace the `` guarantee of delivery '' axiom by an axiom stating that only the messages sent by normal ( non - crashed ) participants will be delivered to all participants .
( see @xcite for examples of such specifications in a @xmath0 context . )
[ [ guarded - actions . ] ] guarded actions .
+ + + + + + + + + + + + + + + + one can also extend the model with guarded actions , where action can be performed depending on global conditions in global configurations . returning to the floodset protocol
, one may consider a variation of the asynchronous protocol suitable for resolving the consensus problem in the presence of _ crash failures_. we can modify the above setting as follows .
now , processes may fail and , from that point onward , such processes send no further messages .
note , however , that the messages sent by a process _ in the moment of failure _ may be delivered to _ an arbitrary subset _ of the non - faulty processes .
the goal of the algorithm also has to be modified , so only _ non - faulty _ processes are required to eventually reach an agreement .
thus , the floodset protocol considered above is modified by adding the following rule : * at every round ( later than the first ) , a process broadcasts any value _ the first time it sees it_. now , in order to specify this protocol the variation of the model with crashes should be used .
the above rule can be easily encoded in the model and we leave it as an exercise for the reader .
an interesting point here is that the protocol is actually correct under the assumption that _ only finitely many processes may fail .
_ this assumption is automatically satisfied in our automata model , but not in its temporal translation .
instead , one may use the above _ finite bounds on delivery _ axiom to prove the correctness of this variation of the algorithm .
now we have all the ingredients to perform the verification of parameterised protocols . given a protocol @xmath203 , we can translate it into a temporal formula @xmath200 .
for the temporal representation , @xmath216 of a required correctness condition , we then check whether @xmath217 is valid temporal formula .
if it is valid , then the protocol is correct for all possible values of the parameter ( sizes ) .
correctness conditions can , of course , be described using any legal formula .
for example , for the above floodset protocol(s ) we have a liveness condition to verify : @xmath218 or , alternatively @xmath219\ ] ] in the case of a protocol working in presence of processor crashes .
while space precludes describing many further conditions , we just note that , in @xcite , we have demonstrated how this approach can be used to verify safety properties , i.e with @xmath220 . since we have the power of , but with decidability results
, we can also automatically verify fairness formulae of the form @xmath221 .
in the propositional case , the incorporation of xor constraints within temporal logics has been shown to be advantageous , not only because of the reduced complexity of the decision procedure ( essentially , polynomial rather than exponential ; @xcite ) , but also because of the strong fit between the scenarios to be modelled ( for example , finite - state verification ) and the xor logic @xcite ) .
the xor constraints essentially allow us to select a set of names / propositions that must occur exclusively . in the case of verification for finite state automata
, we typically consider the automaton states , or the input symbols , as being represented by such sets .
modelling a scenario thus becomes a problem of engineering suitable ( combinations of ) xor sets . in this paper
, we have developed an xor version of @xmath0 , providing : its syntax and semantics ; conditions for decidability ; and detailed complexity of the decision procedure .
as well as being an extension and combination of the work reported in both @xcite and @xcite , this work forms the basis for tractable temporal reasoning over infinite state problems . in order to motivate this further
, we considered a general model concerning the verification of infinite numbers of identical processes .
we provide an extension of the work in @xcite and @xcite , tackling liveness properties of infinite - state systems , verification of asynchronous infinite - state systems , and varieties of communication within infinite - state systems .
in particular , we are able to capture some of the more complex aspects of _ asynchrony _ and _ communication _ , together with the verification of more sophisticated _ liveness
_ and _ fairness _ properties .
the work in @xcite on basic temporal specification such as the above have indeed shown that deductive verification can here be attempted but is expensive the incorporation of xor provides significant improvements in complexity .
the properties of first - order temporal logics have been studied , for example , in @xcite .
proof methods for the monodic fragment of first order - temporal logics , based on resolution or tableaux have been proposed in @xcite .
model checking for parameterised and infinite state - systems is considered in @xcite .
formulae are translated into to a bchi transducer with regular accepting states .
techniques from regular model checking are then used to search for models .
this approach has been applied to several algorithms verifying safety properties and some liveness properties .
* have theoretically non - primitive recursive upper bounds for decision procedures ( although they work well for small , interesting , examples ) in our case the upper bounds are definitely primitive - recursive ; * are not suitable ( or , have not been used ) for asynchronous systems with delayed broadcast it is not clear how to adapt these methods for such systems ; and * typically lead to undecidable problems if applied to liveness properties .
future work involves exploring further the framework described in this paper in particular the development of an implementation to prove properties of protocols in practice .
further , we would like to see if we can extend the range of systems we can tackle beyond the monodic fragment .
we also note that some of the variations we might desire to include in section [ sec : var ] can lead to undecidable fragments .
however , for some of these variations , we have correct although ( inevitably ) incomplete methods , see @xcite .
we wish to explore these boundaries further .
10 p. a. abdulla , b. jonsson , m. nilsson , j. dorso , and m. saksena . .
in _ proc .
16th international conference on computer aided verification ( cav ) _ , volume 3114 of _ lncs _ , pages 348360 .
springer , 2004 .
a. artale , e. franconi , f. wolter , and m. zakharyaschev . a temporal description logic for reasoning over conceptual schemas and queries . in _ proc .
european conference on logics in artificial intelligence ( jelia ) _ , volume 2424 of _ lncs _ , pages 98110 .
springer , 2002 .
j. brotherston , a. degtyarev , m. fisher , and a. lisitsa . .
in _ proc . international conference on logic for programming , artificial intelligence , and reasoning ( lpar )
_ , volume 2514 of _ lncs _ , pages 86101 .
springer verlag , 2002 .
d. gabelaia , r. kontchakov , a. kurucz , f. wolter , and m. zakharyaschev . on the computational complexity of spatio - temporal logics . in _ proc .
16th international florida artificial intelligence research society conference ( flairs ) _ , pages 460464 . aaai press , 2003 .
i. hodkinson , r. kontchakov , a. kurucz , f. wolter , and m. zakharyaschev . on the computational complexity of decidable fragments of first - order linear temporal logics . in _ proc .
international symposium on temporal representation and reasoning ( time ) _ , pages 9198 .
ieee cs press , 2003 . | in this paper we consider the specification and verification of infinite - state systems using temporal logic .
in particular , we describe parameterised systems using a new variety of first - order temporal logic that is both powerful enough for this form of specification and tractable enough for practical deductive verification .
importantly , the power of the temporal language allows us to describe ( and verify ) asynchronous systems , communication delays and more complex properties such as liveness and fairness properties .
these aspects appear difficult for many other approaches to infinite - state verification . |
for understanding the basic physics of high-@xmath4 superconductivity , the two - dimensional @xmath0-@xmath1 model has been extensively studied as one of the most promising and simplified models which describe charge and spin dynamics in the cuo@xmath5 plane.@xcite mean - field theories@xcite and numerical calculations@xcite have shown that the d@xmath6 wave superconductivity ( sc ) takes place in a reasonable parameter region in the phase diagram ( @xmath7 and the doping rate @xmath8 ) .
therefore , as far as the d - wave sc is concerned , the experimentally observed phase diagram of high-@xmath4 superconductors and the results in the @xmath0-@xmath1 model agree quite well .
recently , however , there are some experiments which indicate the importance of the interplay between the antiferromagnetism and d - wave sc .
the most interesting phenomenon is a realization of stripe state in some materials , which has charge ordering as well as incommensurate antiferromagnetic ( af ) spin ordering.@xcite other potential problems are the af state induced around nonmagnetic impurities and a possibility of af vortex cores.@xcite in order to study these problems , it is necessary to develop a theory in which the d - wave sc and the af correlation are treated in a reliable way .
usual mean - field theories generally overestimate the af long - range order , so that they give unphysical results even in the uniform case . in this paper
we develop a new type of analytic theory which gives a reliable estimate of the af correlation as well as the d - wave sc .
our scheme is an extension of the gutzwiller approximation.@xcite in the @xmath0-@xmath1 model the double occupancy of up- and down - spin electrons at any site is prohibited . to study its ground state ,
it is natural to use a projected mean - field wave function in which the double occupancies are excluded.@xcite as the mean - field wave functions , a sdw mean - field solution at half filling,@xcite bcs wave functions,@xcite and a coexistent state of the af and d - wave sc@xcite have been used .
variational energies are calculated in the variational monte carlo ( vmc ) simulations , which treat exactly the constraints of no double occupancies . among the above trial states , the coexistent state has the lowerest variational energy in the doping region @xmath9.@xcite on the other hand , the analytic theories in which the constraint is treated approximately give large discrepancies in the estimation of the af correlation .
for example , in the slave - boson mean - field theory which takes account of the constraint as an average , the af order is too overestimated and it extends up to unphysical doping rates ( @xmath10).@xcite thus in the slave - boson theory , we are unable to discuss the stripe state , for example , because it is stabilized near @xmath11 doping where the slave - boson theory gives the af state even for the bulk .
the gutzwiller approximation is more advanced approximation than the simple mean - field theories in treating the constraint .
renormalizations of expectation values are introduced by comparing the statistical weighting factors in the wave functions with and without projection.@xcite as a result , the parameters @xmath0 and @xmath1 are renormalized to @xmath12 and @xmath13 .
it has been shown that the gutzwiller approximation gives a fairly reliable estimation of the variational energy for the pure d - wave sc state when it is compared with the vmc results.@xcite however it was shown that there is no region in the phase diagram where the af state is stabilized.@xcite this contradicts with the vmc results . in order to clarify the origin of these discrepancies , we investigated the vmc data@xcite and found that the gutzwiller approximation has to be modified when the af correlations are present . based on these observations , we extend the gutzwiller approximation in this paper and derive an analytic formalism for the renormalization factors which reproduces the vmc results . we show that it is important to take account of the longer - range correlations for the weighting factors , in contrast to the previous approximation where only the site - diagonal expectation values such as @xmath14 are considered . as a result ,
the renormalization factors for @xmath0 and @xmath1 have nonlinear dependences on the expectation values of cooper pairs and af moment .
we think that this is the essence of the strong correlation in the sense that the renormalization appears solely from the projection operators .
the merit of the present scheme is that it can be easily applied to the inhomogeneous systems , where the vmc simulations are not so feasible .
thus a reliable analytic formulation can be given to the problems , such as the stripe state , vortex cores and states around impurities , where the interplay between the af and d - wave sc plays an important role .
preliminary results of these problems have been described elsewhere.@xcite the outline of this paper is as follows : in sec .
ii we briefly review the formulation of the original gutzwiller approximation in order to make the present paper self - contained .
we also show our final results for the extended gutzwiller approximation before going into the details of the derivation . in sec .
iii , the original gutzwiller approximation is extended in order to include longer - range correlations .
we divide the whole system into cells and introduce the configurations in each cell for evaluating the weighting factors . using the general formulation developed in sec .
iii , we obtain the renormalization factors for the half - filled case in sec .
we approximate the weighting factors in a perturbative way with respect to the nearest - neighbor correlations . in iv.d , the physical origin of the enhancement of the renormalization factor for the @xmath2-component of exchange interaction
is discussed in a viewpoint of statistical weights of real - space spin configurations . in sec .
v , the results at half filling are generalized to the less - than - half - filled case .
the present method is applied to the projected variational states in sec .
the self - consistency equations are derived and solved numerically to show that they give reasonable estimate of af long - range order near half filling .
finally section vii is devoted to a summary and discussions on related problems .
we consider the gutzwiller approximation for the two - dimensional @xmath0-@xmath1 model on a square lattice : = -t_ijp_g(c_i^ c_j+h.c.)p_g+j_ij _ i_j , where @xmath15 represents the sum over the nearest - neighbor sites and @xmath16 .
the gutzwiller s projection operator @xmath17 is defined as @xmath18 . for this hamiltonian , we assume the projected bcs - sdw mean - field wave function@xcite [ vwf ] |=p_g|_0(_d^v,_af^v , ^v ) , where @xmath19 , @xmath20 and @xmath21 are the variational parameters relating to d - wave sc , af and chemical potential , respectively , and @xmath22 is a hartree - fock type wave function with d - wave sc and af orders . the wave function @xmath23 is a natural generalization of the rvb state proposed by anderson.@xcite it has been shown that the coexistent state of af and d - wave sc has the best variational energy in the vmc simulations for the doping rate @xmath24.@xcite in evaluating the variational energy , the projection operator makes difficulties for an analytic approach .
the gutzwiller approximation was developed for this purpose . in this method
, the effect of the projection is taken into account by renormalizations of expectation values as follows : c_i^ c_j= g_tc_i^ c_j_0 , _
i_j= g_s_i_j_0 , [ defgs ] where @xmath25 represents the expectation value in terms of @xmath26 represents the normalized expectation value in @xmath27 ; =
. the coefficients @xmath28 and @xmath29 are the renormalization factors of the expectation values , which we call as gutzwiller factors in the following . using these notations ,
the variational energy @xmath30 is rewritten as e_var = _ eff _ 0 , [ evar ] where the parameters @xmath0 and @xmath1 in @xmath31 are replaced with t_eff = g_t t , j_eff = g_s j. a systematic estimation of @xmath28 and @xmath29 was developed by ogawa _
et al._@xcite whose method is briefly reviewed in ii.a .
for the case with af order parameter , it can be shown that g_t= , g_s= , [ origg ] where @xmath32 is the hole concentration , @xmath33 , and @xmath34 is the expectation value of af order parameter in terms of @xmath35 defined as m= ( _ j_0 - _ j _ 0 ) .
these gutzwiller factors , however , do not reproduce the vmc results , as mentioned in i. we find that this discrepancy is due to the fact that only the site - diagonal expectation values are taken into account in obtaining eq .
( [ origg ] ) . in this paper
we extend the method by ogawa _
et al . _ systematically to include the longer range correlations .
we show that it is important to include the effects of the nearest - neighbor expectation values , such as & & = c_i^c_j _ 0 , + & & = c_i^c_j^_0 , [ defxx ] in the estimation of @xmath28 and @xmath29 .
this corresponds to taking account of the effect of surroundings of the corresponding bond .
furthermore we find that the gutzwiller factors @xmath29 for the exchange interaction have different values for the @xmath2 component ( denoted as @xmath36 ) and @xmath3 component ( @xmath37 ) ; i.e. , s^z_i s^z_j & & = g_s^z s^z_i s^z_j _ 0 , + s^+_i s^-_j & & = g_s^xy s^+_i s^-_j _ 0 , instead of ( [ defgs ] ) in the presence of @xmath34 .
actually from the vmc calculations,@xcite we have estimated the behaviors of @xmath36 and @xmath38 as a function of @xmath34 , using the relations [ vmcgf ] g_s^z= , g_s^xy= , where the numerators @xmath39 and @xmath40 are evaluated numerically in vmc simulations .
it was found that @xmath36 is enhanced compared with @xmath38 and that this enhancement is essential for the stabilization of the af order . in the present extended gutzwiller approximation , this result is reproduced .
physically the weighting factor of @xmath36 for a specific configuration of a bond is enhanced due to the effect of the surrounding af correlations . before going into details of the derivation
, we show our final results .
they are summarized as follows : g_s^xy = ( ) ^2 a^-7 , and g_s^z = g_s^xy , [ gszfinal ] where a & & = 1 + , + x & & = 2 ^ 2(^2-^2)+8m^2(^2+^2)+4(^2+^2)^2 , + x_2 & & = 2(^2+^2 ) .
[ xx2 ] the above expressions sufficiently reproduce the results obtained in the vmc simulations
. however we find that the slight difference can be improved by using x = 2 ^ 2(^2-^2)+8m^2(^2+^2)+2(^2+^2)^2 , instead of eq .
( [ xx2 ] ) . a typical @xmath34-dependence of @xmath36 and @xmath38 are shown in fig .
1 for the half - filled case , @xmath41 , and with @xmath42 being fixed at the two values , @xmath43 and @xmath44 .
the behaviors in fig .
1 agree with those obtained in the vmc simulations using the relation in eq .
( [ vmcgf ] ) .
the meanings of the rather complicated gutzwiller factors become apparent in the following sections . the gutzwiller factor for the hopping term
is given by g_t= a. [ gtfinal ] figure 2 shows the @xmath34-dependences of @xmath45 and @xmath28 for @xmath46 , which can be compared with the vmc results in ref .
note here that the slave - boson mean - field theory simply gives @xmath47 and @xmath48 .
the nonlinear dependences of @xmath45 and @xmath28 on @xmath49 and @xmath34 are beyond the slave boson theory . in the following subsection we review the original formulation of the gutzwiller approximation which is useful for the generalization in the later sections .
let us start with the formulation of the gutzwiller approximation developed by ogawa _
et al _ @xcite when it is applied to the @xmath0-@xmath1 model .
in this subsection we do not consider the af order .
the ultimate aim is to evaluate the variational energy e_var = = , using @xmath50 .
first , ogawa _ et al _ rewrote the denominator as ( @xmath51 ) | & & = _ 0 | p_g p_g | _ 0 + & & = _ 0 | _ j ( 1-n_j n_j ) | _ 0 + & & = _ 0 | _ j \ { n_j(1-n_j ) + n_j(1-n_j ) + & & + ( 1-n_j)(1-n_j ) } | _ 0 .
[ deno ] the expansion of the product in ( [ deno ] ) leads to various real - space configurations .
therefore @xmath52 can be rewritten as _ config . _ 0 & & | _ j n_j(1-n_j ) _ j
n_j(1-n_j ) + & & _ j ( 1-n_j ) ( 1-n_j ) | _ 0 , [ deno2 ] where @xmath53 ( @xmath54 ) is a subset of the lattice sites which is singly occupied by an up - spin ( down - spin ) electron and @xmath55 is a subset which is vacant .
the summation is over all the possible configurations , i.e. , all the possible combinations of @xmath56 . in principle , eq .
( [ deno2 ] ) can be calculated using wick s theorem since @xmath35 is a mean - field wave function , although it is very complicated .
the simplest gutzwiller approximation is to estimate ( [ deno2 ] ) by using only the site - diagonal expectation values , which gives _ config .
& & _ j n_j(1-n_j ) _ 0 _ j n_j(1-n_j ) _ 0 + & & _ j ( 1-n_j)(1-n_j )
_ 0 , [ deno3 ] with @xmath25 meaning the expectation value in @xmath35 .
for example , the contribution from the site @xmath57 is evaluated as _ a n_j(1-n_j ) _ 0 = ( 1- ) , where @xmath58 is the average electron density , @xmath59 , with @xmath60 ( @xmath61 ) being the total number of sites ( total electron number ) .
we call @xmath62 as the weight for @xmath63 site belonging to the subset @xmath64 .
since the hole density is @xmath65 , we obtain _ a & & = _ b = ( 1- ) = , + _ e & & = ( 1- ) ^2 = .
then the expectation value ( [ deno3 ] ) is rewritten as [ deno0 ] _ config .
_ a^n_a _ b^n_b _ e^n_e = _ a^n_a _ b^n_b _ e^n_e , where @xmath66 is the number of sites in the subset @xmath53 , and so on , ( @xmath67 ) . in this simple case ,
@xmath66 is equal to the number of up - spin electrons , so that n_a = n_b= , n_e = n - n_e .
a similar procedure can be carried out for the estimation of the numerator in @xmath68 .
for the exchange term , we have _ 0&&| p_g s_^+ s_m^- p_g | _ 0 + & & = _ 0 | s_^+ s_m^- _ j , m \ { n_j(1-n_j )
+ n_j(1-n_j ) + & & + ( 1-n_j)(1-n_j ) } | _ 0 + & & = _ config . _ 0 | s_^+ s_m^- _ j n_j(1-n_j ) _ j
n_j(1-n_j ) + & & _ j ( 1-n_j ) ( 1-n_j ) | _ 0 .
[ nume ] here the summation is over all the possible configurations @xmath69 in which the two sites @xmath70 and @xmath34 are excluded .
the evaluation of ( [ nume ] ) with the site - diagonal expectation values for @xmath71 leads to _ a^n_a _ b^n_b _ e^n_e s_^+ s_m^- _ 0 .
[ gaexp ] since the sites @xmath70 and @xmath34 contain one up - spin electron and one down - spin electron , we have @xmath72 , @xmath73 , and @xmath74 . by combining the estimation of the denominator in eq .
( [ deno0 ] ) , we obtain the original gutzwiller approximation s_^+ s_m^- & & = + & & = _
a^-1_b^-1 s_^+ s_m^- _ 0 + & & = s_^+ s_m^- _ 0 .
this gives @xmath75 .
ogawa _ et al_@xcite also considered the above formalism in the case with af long - range order for the hubbard model .
the application to the @xmath0-@xmath1 model was carried out by zhang _ et al_.@xcite in this subsection we follow their methods to obtain the gutzwiller approximation in the @xmath0-@xmath1 model . in the af case ,
the sublattices 1 and 2 are distinguished and their magnetizations are defined as ( n_j _ 0 - n_j _ 0 ) = m , for the sublattice 1 and @xmath76 for the sublattice 2 . thus we denote n_j _ 0 & & = + m r , + n_j _ 0 & & = -m
w , for the sublattice 1 , where @xmath77 ( @xmath78 ) means the average electron density with the right ( wrong ) spin direction on the sublattice . @xmath77 and @xmath78 are exchanged for the sublattice 2 . using these notations ,
the denominator of @xmath68 is evaluated as _ 0 | p_g p_g | _ 0 = & & _ config .
_ a1^n_a1 _ b1^n_b1 _ e1^n_e1 _ a2^n_a2 _ b2^n_b2 _ e2^n_e2 + = & & _ n_a1 , n_b1 + & & _ a1^n_a1 _ b1^n_b1 _ e1^n_e1 _ a2^n_a2 _ b2^n_b2 _ e2^n_e2 , [ denoaf ] where the configurations in the sublattice 1 are specified by @xmath79 , and so on .
contrary to the previous subsection , eq .
( [ denoaf ] ) has a summation over possible values of @xmath80 and @xmath81 because there are only four constraints between six numbers , @xmath82 :
n_a1+n_b1+n_e1&&= , n_a2+n_b2+n_e2= , + n_a1+n_a2&&= , n_b1+n_b2=. + the weights in the site - diagonal expectation values are _ a1 & & = n_j _ 0 1-n_j _ 0 = r(1-w ) , + _ b1 & & = n_j _ 0 1-n_j _ 0 = w(1-r ) , + _ e1 & & = 1-n_j _ 0 1-n_j _ 0 = ( 1-r)(1-w ) , for the sublattice 1 , and _
a2&&=_b1 , + _ b2&&=_a1 , + _ e2&&=_e1 , for the sublattice 2 . since @xmath83 and @xmath60 are large numbers , eq .
( [ denoaf ] ) is further approximated by the largest term in the summation . by taking the partial derivative of the logarithm of each term in ( [ denoaf ] ) with respect to @xmath80 , we obtain - + ( - ) + _ a1 - _ a2=0 , [ maximize ] with @xmath84 being the value of @xmath80 which gives the largest term in the summation . by solving ( [ maximize ] ) , @xmath84 is given by = = .
[ maximizesol ] similarly we obtain = .
the summation in ( [ denoaf ] ) is approximated by the term with @xmath84 and @xmath85 .
again the numerator such as @xmath86 is approximated in the similar way . as a result , the gutzwiller approximation in the presence of af order , @xmath34 , becomes s_^+ s_m^- & & = + & & = _
a1 ^ -1_b1 ^ -1 s_^+ s_m^- _ 0 + & & = s_^+ s_m^- _ 0 + & & = s_^+ s_m^- _ 0 , [ gs0 ] where a useful relation = has been used
. equation ( [ gs0 ] ) gives g_s^xy= , which is simply eq .
( [ origg ] ) .
it is straightforward to see g_s^z = g_s^xy , and g_t = .
the approximation in the preceding section is restricted to the site - diagonal ( on - site ) expectation values .
it was shown that this approximation is not good enough to reproduce the af order obtained in the vmc simulation.@xcite a straightforward extension of the previous gutzwiller approximation is to consider longer - range expectation values .
the nearest - neighbor expectation was considered before.@xcite in this section , we develop a general formalism to take account of the longer - range effects systematically . in evaluating the denominator of @xmath68 , i.e. , @xmath87 ,
various configurations have been classified depending on the state ( @xmath88 or a hole ) on each site as @xmath56 in the previous section . instead of this
we divide the whole system into _ cells _ consisting of @xmath89 sites .
( later we take @xmath89 as a large enough number . ) for each cell , there are @xmath90 configurations because each site has @xmath88 or a hole .
we denote these configurations as _ states _ of a cell . using these states ,
the denominator of @xmath68 can be approximated as _ 0 | p_g p_g | _ 0 & & = _ all the possible states _
i^n_i + & & = _ \ { n_i } _ i=1^k _ i^n_i , [ denogen ] with @xmath91 . here
@xmath92 is the number of cells in the @xmath93-th state , and @xmath94 is the weight of the @xmath93-th state cell in the analogy with eq .
( [ denoaf ] ) .
explicitly the weight @xmath94 is defined as _
i = & & _ 0 | _
j n_j(1-n_j ) _
j n_j(1-n_j ) + & & _ j
( 1-n_j)(1-n_j ) | _ 0 , with @xmath95 and @xmath96 being the sites in the cell . to calculate
@xmath94 is the most important part of the present theory which we carry out later .
the last summation in ( [ denogen ] ) is over all the possible values of @xmath92 under the following constraints : & & _ i=1^k n_i n_i = , _ i=1^k n_i n_i = , + & & _
i=1^k n_i = n- n_e , [ constr ] where @xmath97 ( @xmath98 ) represents the number of up - spins ( down - spins ) in the @xmath93-th state cell , and @xmath99 represents the number of holes , respectively .
since a cell has @xmath89 sites , a relation n_i + n_i + = n_c , [ constr2 ] holds . under the above constraints , we look for the largest term in eq .
( [ denogen ] ) as was done in the original gutzwiller approximation ( see eq .
( [ maximize ] ) ) .
since the constraints ( [ constr ] ) are rather complicated , it is useful to introduce the lagrange multipliers ( @xmath100 ) . taking the partial derivative of & & - _ ( _ i=1^k n_i n_i - ) + & & - _ ( _ i=1^k n_i n_i - ) - ( _ i=1^k n_i - n + n_e )
, we obtain = _ i exp ( -_n_i -_n_i - ) . from symmetry
we find @xmath101 , and using the relation ( [ constr2 ] ) we get = _
i e^-n_c e^(- ) .
furthermore we introduce new variables @xmath102 and @xmath103 as ^-n_c & & , + e^- & & p . using these variables ,
@xmath104 is rewritten as = p^. the variables @xmath102 and @xmath103 are to be determined from the constraints ( [ constr ] ) which are equivalent to _
i=1^k = , _
i=1^k = n- n_e .
these conditions become _
i=1^k p^&&= 1 , + _
i=1^k p^&&=n_c .
it is convenient to classify all the possible states of a cell into subgroups which contain @xmath63 holes ( @xmath105 ) .
( we call these subgroups as @xmath63-hole sectors . )
then the quantity w_j = _ i with j holes _
i , [ wj ] represents the total weight of the states in this subgroup . using @xmath106 , we can rewrite the constraints as _ j=0^n_c p^j & & = 1 , + _
j=0^n_c j p^j & & = n_c . [ constr3 ] the first equality is rewritten as @xmath107 so that @xmath102 represents the total weight . in the following sections , we evaluate @xmath106 and then determine @xmath102 and @xmath103 from eq .
( [ constr3 ] ) .
the numerator in @xmath68 is evaluated in a similar way .
let us consider the expectation value of an operator @xmath108 ( such as @xmath109 ) which operates on a certain part inside a cell of the @xmath110-th state .
( we call this cell as the central cell . )
then all the configurations in the other cells are classified by @xmath111 which is the number of cells of the @xmath93-th state .
then the expectation value becomes _
0 | p_g p_g | _ 0 = _
i_0 _ \ { n_i } _ i=1^k _ i^n_i _ i_0 , [ oexp ] in the analogy with eq . ( [ gaexp ] ) . here
@xmath112 indicates the expectation value of @xmath108 together with terms such as @xmath113 inside the central cell of the @xmath110-th state .
determining the largest terms on the right - hand side and taking the ratio to the denominator , we finally obtain & & = _
i_0 n_i_0 _
i_0 ^ -1 _ i_0 + & & = _ i_0 _
[ numegen ] the evaluation of the largest term and the derivation of the above result is shown in appendix a. if we consider only the site - diagonal expectation values to estimate @xmath94 , eq .
( [ numegen ] ) reproduces the original gutzwiller approximation shown in the previous section .
this is summarized in appendix b.
to obtain the gutzwiller factors formulated in the last section , it is necessary to evaluate the weights @xmath94 for each state . although these weights are generally complicated , we estimate them assuming that the corrections to the original gutzwiller approximation ( reproduced in appendix b ) are small . to be more specific , we evaluate @xmath106 in the lowest orders with respect to the intersite correlations @xmath114 and @xmath115 defined in eq .
( [ defxx ] ) .
typical values for @xmath114 and @xmath115 are less than @xmath116 , so that the perturbation with respect to them will be justified . in this section
we calculate the gutzwiller approximation for the half filling , since it is less complicated than the doped case .
all the cells do not contain holes so that only the weight , @xmath117 , is to be calculated .
we call this subset of states as 0-hole sector . after the half - filled case , it is rather straightforward to extend the results to the doped case .
first we calculate the weight @xmath94 of the @xmath93-th state which has @xmath97 up - spin electrons , @xmath98 down - spin electrons : _
i = _ j n_j(1-n_j ) _ j
n_j(1-n_j ) _ 0 . in the lowest order of @xmath114 and @xmath115 or in the zeroth order
, we have to take only the site - diagonal expectation values .
this gives _
i^(0)= [ r(1-w)]^n_right [ w(1-r)]^n_wrong , where @xmath118 ( @xmath119 ) means the number of sites on which the right ( wrong ) spices of spin direction is located depending on sublattice 1 and 2 .
the summation over all the possible configurations gives w_0^(0 ) = [ r(1-w)+w(1-r)]^n_c = ( n-2rw)^n_c . in the next order with respect to @xmath114 and @xmath115
, we have to consider contributions from expectation values of bonds in the cell as shown in fig .
3 . for example , consider a bond connecting the sites @xmath93 and @xmath63 ( see fig .
3(a ) ) and a state in which two up - spin electrons occupy both @xmath93 and @xmath63 sites .
the expectation value in this bond gives p _ & & = ( 1-n_i ) n_i n_j ( 1-n_j ) _ 0 + & & = rw(1-r)(1-w)-rw^2 -(1-r)(1-w)^2 + & & -r(1-r)^2 - w(1-w)^2 + ( ^2+^2)^2 , where we have used the wick s theorem . generally @xmath114 and @xmath115 can be complex numbers and @xmath120 in the above expression mean @xmath121 implicitly . in the similar way we calculate @xmath122 .
writing that the left - spin in the subscript indicates the spin on the sublattice 1 , they are calculated as p _ & & = p _ , + p _ & & = r^2(1-w)^2 + 2r(1-w)^2 + & & + r^2 ^ 2 + ( 1-w)^2 ^ 2 + ( ^2+^2)^2 , + p _ & & = w^2(1-r)^2 + 2w(1-r)^2 + & & + w^2 ^ 2 + ( 1-r)^2 ^ 2 + ( ^2+^2)^2 . in the estimation of @xmath123 we need their summation p_+p _
+ p_+p _ = ( n-2rw)^2 + x , [ totalp ] with @xmath124 defined in eq .
( [ xx2 ] ) , where we have used the relations @xmath125 and @xmath126 .
the first term @xmath127 on the r.h.s . is the zeroth order contribution which has been included in @xmath128 . in this sense
, @xmath124 is a kind of connected ( or cummulant ) contribution coming from the real - space diagram in fig .
the actual values of @xmath124 is roughly @xmath129 so that our perturbation scheme can be justified .
denoting the number of bonds in a cell as @xmath130 , we have a contribution to @xmath117 as w_0^(1 ) = n_b x ( n-2rw)^n_c-2 , [ w00 ] where the factor @xmath131 comes from the contributions of the other @xmath132 sites in the cell except for @xmath93 and @xmath63 . in the similar way
, we evaluate the higher order contributions to @xmath117 . by considering two bonds in the cell
we approximate their contribution as w_0^(2 ) = _
n_bc_2 x^2 ( n-2rw)^n_c-4 , [ secondc ] where @xmath133 is the number of choices for the positions of the two bonds in a cell .
the summation of these series leads to w_0 & & = _
j=0^n_b _ n_bc_j x^j ( n-2rw)^n_c-2j + & & = ( n-2rw)^n_c ( 1+)^n_b . [ w0 ] let us discuss the neglected contributions to @xmath117 . figures 3(b ) and 3(c ) show real - space diagrams whose connected expectation values can contribute to @xmath117 .
although their contributions are not included , it can be shown that they are smaller than the terms in ( [ w0 ] ) .
for example , the contribution from fig .
3(b ) is in the order of @xmath134 and @xmath135 etc .
, so that it is smaller than @xmath123 .
the contribution from fig .
3(c ) is in the order of @xmath136 which is smaller than ( [ secondc ] ) by a factor @xmath137 .
there is another effect neglected in @xmath117 : the higher order terms in ( [ w0 ] ) become less and less correct because the number of the bonds giving @xmath124 is not exactly @xmath138 due to their overlapping .
however the error is again in the order smaller than @xmath138 . in this sense ,
( [ w0 ] ) is a kind of a summation of most dominant terms in the order of @xmath139 . in order to obtain the gutzwiller factor for @xmath109 at half filling we need to calculate s_^+
i_0 s_^+ s_m^- _ i_0 , according to the general formulation ( [ numegen ] ) .
note that @xmath140 and the central cell ( which includes sites @xmath70 and @xmath34 ) does not contain holes , i.e. , @xmath141 .
the average in a central cell of the @xmath110-th state , @xmath142 , is evaluated in a similar perturbation scheme as in @xmath117 . in the lowest order of @xmath114 and @xmath115 , we have ( fig . 4(a ) )
s_^+ s_m^- _ 0 ( n-2rw)^n_c-2 , [ gsxy0 ] where the factor @xmath131 comes from the contributions from the sites in the central cell other than the sites @xmath70 and @xmath34 .
~#1#1 # 1#1 # 1#1 in the next order of @xmath114 and @xmath115 , there are contributions of @xmath124 from the bonds in the cell .
this leads to _b x s_^+ s_m^- _ 0 ( n-2rw)^n_c-4 , [ gsxy1 ] where @xmath143 is the number of bonds which are not connected to the sites @xmath70 and @xmath34 directly ( fig .
in general @xmath143 depends on the shape of the cell . however , if we choose the large enough cell , we have _b = n_b-7 , [ nb7 ] as is evident from fig .
4(b ) . although the dotted bonds in fig .
4(b ) do not give @xmath124 , they may still give different contributions to @xmath142 , such as s_^+ s_m^- n_m ( 1-n_m ) _ c , [ gsxynn ] where @xmath144 is a nearest neighbor site of @xmath34 and @xmath145 means a connected expectation value excluding the disconnected terms such as @xmath146 ( the disconnected expectation values have been already taken into account in eq .
( [ gsxy0 ] ) . ) by calculating ( [ gsxynn ] ) , however , we find that there are no connected conributions : actually the two electron operators in @xmath147 have to make contractions with the electron operators on the site @xmath34 in the wick s expansion , which gives just the disconnected expectation value . by summing up the series of corrections such as ( [ gsxy1 ] ) we obtain & &
_ i_0 s_^+ s_m^- _ i_0 + & & = s_^+ s_m^- _ 0 \ { ( n-2rw)^n_c-2 + & & + _b x ( n-2rw)^n_c-4 + } + & & = s_^+ s_m^- _ 0 ( n-2rw)^n_c-2 ( 1+)^_b . combining with @xmath117 , the final result for the extended gutzwiller approximation for @xmath109 becomes s_^+ s_m^- = ( 1+)^-(n_b - _b ) .
since the gutzwiller factor is defined as the ratio between @xmath148 and @xmath149 , we have g_s^xy & & = ( 1+)^-(n_b - _b ) + & & = , where we have put a 1 + , which appears frequently in the following discussion .
if we choose a large enough cell like @xmath150 , we obtain g_s^xy= , [ gxyhf ] according to eq .
( [ nb7 ] ) .
physically the factor @xmath151 with @xmath152 represents the exclusion effect of the bonds as shown in fig .
4(b ) .
the gutzwiller factor for the @xmath2-component of @xmath153 is calculated in the similar way . the lowest orders with respect to @xmath114 and @xmath115 give the similar results to ( [ gsxy0 ] ) and ( [ gsxy1 ] ) , in which @xmath149 is replaced with @xmath154 .
however , for @xmath155 , an additional contribution appears from the diagram in fig .
5(a ) which did not give contributions in @xmath142 .
this is the reason why the anisotropy for the gutzwiller factor between @xmath3 component and @xmath2 component appears , which is the most important feature of our extended gutzwiller approximation . assuming that the site @xmath70 and @xmath144 in fig .
5(a ) are in the sublattice 1 , we have a contribution & & s_^z s_m^z \ { n_m(1-n_m ) + n_m(1-n_m ) } _ c + & & = s_^z _ 0 s_m^z n_m(1-n_m ) + s_m^z n_m(1-n_m ) _ c + & & = ( p_-p _ + p_-p _ ) + & & = -m^2 x_2 , with @xmath156 defined in eq .
( [ xx2 ] ) .
since the number of bonds connected to the site @xmath34 is n_2 , the contribution to @xmath157 turns out to be -2n_2
m^2 x_2 ( n-2rw)^n_c-3 , where the contributions from the bonds connected to the site @xmath70 are also included .
similarly the diagram in fig .
5(b ) gives a contribution & & s_^z n_(1-n_ ) + s_^z n_(1-n_ ) _ c + & & s_m^z n_m(1-n_m ) + s_m^z n_m(1-n_m ) _ c + & & = -(p_-p _ + p_-p_)^2 + & & = -m^2 x_2 ^ 2 . by counting the number of possible combination of the bonds , we have -n_2 ^ 2 m^2 x_2 ^ 2 ( n-2rw)^n_c-4 .
the higher order terms of @xmath158 are calculated as before , which become s_^z s_m^z = \ { & & s_^z s_m^z _ 0 ( n-2rw)^n_c-2 a^_b + -&&2n_2 m^2 x_2 ( n-2rw)^n_c-3 a^ n_b + -&&n_2 ^ 2 m^2 x_2 ^ 2 ( n-2rw)^n_c-4 a^
n_b } , with @xmath159 and @xmath160 being the numbers of the bonds which are not connected directly to the diagrams in figs .
5(a ) and 5(b ) , respectively .
they are shown in figs .
6(a ) and ( b ) . in the large enough cell
, we have n_b & & = n_b -10 , + n_b & & = n_b -13 . combining with @xmath117
, we obtain s_^z s_m^z = & & \
{ s_^z s_m^z _ 0 + & & - m^2 x_2 a^-(_b- n_b ) + & & - m^2 x_2 ^ 2 a^-(_b- n_b ) } .
+ & & [ szsz ] in this case , @xmath161 is not proportional to @xmath154 . since the gutzwiller factor is the ratio between the two , we have some nontrivial contribution from the second and third terms on the r.h.s of ( [ szsz ] ) . using s_^z s_m^z _ 0
= -m^2 - , the final expression for the gutzwiller factor turns out to be g_s^z & & = g_s^xy + & & , [ gsz ] where we have assumed _b - n_b = 2(_b - n_b ) . for a large enough cell
we have @xmath162 and @xmath163 , so that g_s^z = g_s^xy .
the above formula is one of the most important results in this paper .
typical @xmath34-dependences have been shown in fig . 1 .
let us here check some limiting cases .
when @xmath164 , we have g_s^xy = g_s^z = 4 a^-7 , which reproduces the original gutzwiller approximation . for small @xmath34 , we obtain g_s^z = g_s^xy \ { 1 + 4m^2 ( a^-3 + a^-6 ) } .
it is apparent that @xmath36 has an enhancement as a function of @xmath34 , which gives the reasonable af long - range order at half filling . in the usual interpretation of the gutzwiller approximation
, a comparison is made between the probabilities of spin configurations in the wave functions with and without the projection operators.@xcite when we consider only two sites @xmath70 and @xmath34 for @xmath153 , the summation of the probabilities of spin configurations @xmath165 in the wave function _ without _ the projection is given by @xmath166 which was calculated as @xmath167 in eq .
( [ totalp ] ) . on the other hand , in the wave function _ with _ the projection
, we have always one of the above four states at half filling , and thus the probability is equal to 1 .
the ratio of these probabilities gives the gutzwiller factor g_s^xy = g_s^z = .
[ 2sitegs ] however this gutzwiller approximation does not give a reasonable answer compared with the vmc results as mentioned before .
for the physical understanding of the results in the previous subsections , it is necessary to go beyond the two sites , @xmath70 and @xmath34 , because the enhancement of @xmath36 comes from the diagrams in fig .
5 . for this purpose
we consider the spin configurations on three sites as shown in fig .
7 . to calculate the probabilities of these configurations
is straightforward .
for example for fig .
7(c ) we have = 2 r(1-w ) p _ - 2r^3(1-w)^3 , where the second term on the r.h.s .
appears to avoid the double counting of the zeroth order with respect to @xmath114 and @xmath115 .
we can see that the presence of the third site @xmath144 enhances the probability of this configuration , @xmath168 , because @xmath169 in the presence of af order . as a result
, the weight of the configuration @xmath170 on @xmath70 and @xmath34 sites is increased in the wave function .
this causes the enhancement of @xmath36 . to understand this effect more quantitatively , we calculate @xmath161 directly from the configurations in fig . 7 as follows : & & 4 s_^z s_m^z + & & = . the denominator becomes ( n-2rw)\ { ( n-2rw)^2 + 2x } , while the numerator becomes & & ( n-2rw)(p_-p _ -p_+p _ ) + & & + 2m(p_-p _ + p_-p _ ) + 4m^2(n-2rw ) + & & = 4(n-2rw ) s_^z s_m^z _ 0 -4m^2x_2 . although the denominator is larger than the two - site case in eq .
( [ 2sitegs ] ) , the second term in the numerator enhances @xmath171 , which is mainly from the contribution of the configuration 7(c ) . using @xmath172 and taking the ratio ,
we obtain & & = + & & \
{ x_2 + 4m^2 ( 1 + ) } .
this is essentially the result obtained in the previous subsection for @xmath36 .
since we have considered only the site @xmath144 as the third term , @xmath173 in eq .
( [ gsz ] ) is replaced with @xmath174 in this case .
the first factor @xmath175 corresponds the exclusion effect which was represented by @xmath176 in @xmath38 . from this analysis
it is apparent that the enhancement of @xmath36 as a function of @xmath34 is due to the increase of the probability of @xmath177 in the presence of the af circumstances .
in the presence of holes , the gutzwiller factors are calculated similarly as in the half - filled case . in this subsection
we obtain the total weight , @xmath102 .
since the weight for the zero - hole sector @xmath117 has been calculated , we have to evaluate w_1 = _ i with 1 hole _ i , and @xmath178 and so on . in the lowest order of @xmath114 and @xmath115
, we have w_1^(0 ) = n_c ( 1-r)(1-w ) ( n-2rw)^n_c-1 , where the factor @xmath89 comes from the possible position of the hole in a cell and the factor @xmath179 comes from the expectation value of the hole position , @xmath180 . in the next order , by counting the number of bonds which contribute @xmath124 , we obtain n_c ( 1-r)(1-w ) n_1b x ( n-2rw)^n_c-3 , where @xmath181 means the number of bonds which are not connected the hole site as shown in fig .
8(a ) . in the large enough cell
we have n_1b= n_b -4 . although the dotted bonds in fig .
8(a ) do not give @xmath124 , they give alternative contributions to @xmath182 in the same order .
this contribution is calculated as & & ( 1-n_i)(1-n_i ) \ { n_j(1-n_j ) + n_j(1-n_j ) } _ 0 + & & = ( n-2rw)(1-r)(1-w)+y , with y=(1+)(^2-^2 ) -4m^2(^2+^2 ) -2 ( ^2+^2)^2 . here
@xmath93 represents the site of the hole and @xmath63 is one of the nearest neighbor sites . since the number of the dotted bonds in fig .
8(a ) is @xmath183 , the additional contribution to @xmath182 becomes n_c ( n_b - n_1b ) y ( n-2rw)^n_c-2 . the higher order terms with respect to @xmath124 can be taken into account in the similar way as before and we obtain w_1&&= n_c ( 1-r)(1-w ) ( n-2rw)^n_c-1 a^n_1b + & & + n_c ( n_b - n_1b ) y ( n-2rw)^n_c-2 a^_b , where @xmath143 appears because the diagram in fig .
8(a ) excludes the same number of bonds as in fig .
we assume that _b = n_1b-3 , which leads to w_1&&= n_c \ { ( 1-r)(1-w)+ ya^-3 } + & & ( n-2rw)^n_c-1a^n_1b . strictly speaking ,
when the hole is located on the boundary of the cell , the weight should be different .
however we make an approximation of large enough cell . continuing the similar arguments
, we obtain w_2&&=_n_cc_2 \ { ( 1-r)(1-w)+ ya^-3 } ^2 + & & ( n-2rw)^n_c-2a^n_2b , [ w2lthf ] for the two - hole sector , where @xmath184 is the number of bonds which are not connected to the two sites of holes ( fig .
we have neglected the case in which two holes come to the nearest neighbor sites . from eq .
( [ w2lthf ] ) it is apparent that the @xmath63-hole sector , @xmath106 , can be generally approximated as w_j = _
n_cc_j z^j ( n-2rw)^n_c - j a^n_j , b , with z=(1-r)(1-w)+ ya^-3 . finally approximating that @xmath185 , the constraints which determine @xmath102 and @xmath103 ( eq . ( [ constr3 ] ) )
p^j & & = ( n-2rw+pza^-4)^n_ca^n_b=1 , + _
j=0^n_c jp^j & & = pza^-4(n-2rw+pza^-4)^n_c-1a^n_b + & & = n_c . from these constraints
we obtain p & & = a^4 , + w & & = ( ) ^n_c a^n_b , [ pandw ] for less - than - half - filling . in the similar way we estimate @xmath38 . in the case
when there are @xmath63 holes in the central cell , we have & & _
i_0 with j holes s_^+ s_m^- _ i_0 + & & = _ n_c-2c_j z^j ( n-2rw)^n_c-2-j a^_b-4j s_^+ s_m^- _ 0 . by summing up all the sectors with different number of holes ,
we have ( according to the general formula ( [ numegen ] ) ) , s_^+ s_m^- & & = _ j=0^n_c-2 _ i_0 with j holes s_^+ s_m^- _
i_0 + & & = ( n-2rw+pza^-4)^n_c-2a^_b s_^+ s_m^- _ 0 + & & = ( ) ^2 a^-(n_b - _b ) s_^+ s_m^- _ 0 . from this
we obtain the gutzwiller factor g_s^xy=()^2 a^-7 .
this is a simple generalization of the results at half filling obtained in eq .
( [ gxyhf ] ) .
although the estimation of @xmath36 is essentially the same as in the half - filled case , there are some additional diagrams to be taken into account which are shown in fig .
the detailed calculations are summarized in appendix c. the final result is g_s^z & & = g_s^xy + & & .
the quantity @xmath103 is given in eq .
( [ pandw ] ) . for @xmath36
, we use the lowest order approximation p ~ ~2 , since @xmath186 and @xmath187 in the lowest order with respect to @xmath188 and @xmath115 .
this gives our final expression for @xmath36 given in eq .
( [ gszfinal ] ) . as for the gutzwiller factor for the hopping term , @xmath28
, we have to calculate ( 1-n_i ) c_i^ c_j(1-n_j ) _ i_0 .
since the hopping term needs at least one hole , the lowest sector is the one - hole sector .
for this sector , we have & & _
i_0 with 1 hole ( 1-n_i ) c_i^ c_j(1-n_j ) _ i_0 + & & = ( 1-n_i ) c_i^ c_j(1-n_j ) _
0 a^_b ( n-2rw)^n_c-2 + & & = t a^_b ( n-2rw)^n_c-2 , where the expectation value of the correlated hopping , @xmath189 , is calculated as t & & ( 1-n_i ) c_i^ c_j(1-n_j ) _ 0 + & & = ( 1-r)(1-w)- ^3 -^2 + & & = \ { ( 1-r)(1-w ) - } .
it is easy to show that for the @xmath190-hole sector _
n_c-2c_j t a^_j , b ( n-2rw)^n_c-2-j z^j . therefore summing up all the sectors , we obtain c_i^c_j & & = _ j=0^n_c-2 _
n_c-2c_j t a^_j , b ( n-2rw)^n_c-2-j z^j p^j+1 + & & = ( ) ^2 t a^-(n_b - _b ) p. since @xmath191 , the gutzwiller factor @xmath192 becomes g_t & & = ( ) ^2 \ { ( 1-r)(1-w ) - } a^-(n_b - _b )
p + & & = a^-(n_b - _b)+4 , + & & where we have substituted the value of @xmath103 obtained in eq .
( [ pandw ] ) .
the diagrams in fig . 5
give contributions of the order of @xmath193 or @xmath194 so that they are neglected . finally comparing the definitions of @xmath124 and @xmath195
, we use an approximation y~- , for the small @xmath32 case . using this approximation ,
the quantity @xmath2 can be rewritten as z & & = ( 1-r)(1-w ) \ { 1 + ya^-3 } + & & = ( 1-r)(1-w ) \ { 1 - 4(n_b - n_1b)x } + & & = ( 1-r)(1-w ) a^-(n_b - n_1b ) + & & = ( 1-r)(1-w ) a^-4 .
here we are considering @xmath196 and @xmath114 as small quantities . substituting this approximate @xmath2 into @xmath28 we have g_t & & = a^-(n_b - _b)+8 + & & = a.
this is our final expression for @xmath28 given in eq .
( [ gtfinal ] ) .
in this section we calculate the energy of the variational state using the gutzwiller factors obtained in the previous sections .
we will show that the variational energies and the magnitudes of order parameters agree with the results in vmc simulations . in the variational state @xmath197 in eq .
( [ vwf ] ) , @xmath22 is a hartree - fock type wave function with the @xmath198-wave sc and af orders .
it is expressed as@xcite @xmath199|0\rangle,\end{aligned}\ ] ] where @xmath200 @xmath201 , @xmath202 and @xmath203 .
the annihilation operators @xmath204 are related to the electron operators through the following unitary transformation , @xmath205 with @xmath206 here @xmath207 are annihilation operators of an electron on the a(b)-sublattice and @xmath208 represent @xmath209(+1 ) and @xmath210(-1 ) .
the wave vector @xmath211 is limited to half of the brillouin zone where @xmath212 .
we can confirm that @xmath213 is a vacuum of the annihilation operators which diagonalize @xmath214 .
\label{mfh}\end{aligned}\ ] ] in order to clarify the correspondence to the mean - field theory , let us consider the effective hamiltonian , @xmath215 in eq .
( [ evar ] ) . in @xmath215
the parameter @xmath0 in @xmath216 is replaced with t_eff = g_t t , and the exchange term is replaced with j_i _ j = g_s^xy j ( s^x_i s^x_j + s^y_i s^y_j ) + g_s^z j s^z_i s^z_j , where @xmath28 , @xmath38 and @xmath36 are the gutzwiller factors obtained in the previous sections .
when we apply the mean - field theory to @xmath215 , we obtain the similar hamiltonian as in eq .
( [ mfh ] ) but with the replacements t & & t_eff + j_eff ^v , + _ d^v & & j_eff ^v , + _ af^v & & 2j_eff^z m^v , with j_eff = g_s^xy j + g_s^z j , j_eff^z = g_s^z j. note that in the usual mean - field theory the self - consistency equations give @xmath217 and @xmath218 , where @xmath219 and @xmath34 are the expectation values @xmath220 , @xmath221 , and @xmath222 used in the previous sections .
we will see shortly that @xmath223 and @xmath114 etc . are slightly different due to the dependence of the gutzwiller factors on @xmath219 and @xmath34 .
using the wave function @xmath35 , the expectation values become @xmath224 with f_k & = & j_eff^v _ k , + e_k & = & , where @xmath60 is the number of sites and the summation over @xmath211 is limited to half of the brillouin zone . using these expectation values
we obtain e_var & = & _ eff _ 0 + & = & -8n t_eff _ -4n j_eff ( ^2 + ^2 ) + & & -2n j_eff^z m^2 - ( nn - n_e ) . by taking the derivatives with respect to the variational parameters , @xmath225 , and @xmath226 , we find the following self - consistency equations : ^v & = & - _ 0 , + ^v & = & - _ 0 , + m^v & = & m - _ 0 , + ^v & = & - _ 0 , [ sceq ] where the partial derivative of @xmath227 is applied to the gutzwiller factors , @xmath45 and @xmath28 .
figure 10 shows the self - consistent parameters @xmath49 and @xmath34 satisfying ( [ sceq ] ) as a function of the doping @xmath33 for @xmath7 .
note that these are the expectation values in the wave function @xmath213 without the projection . the dashed line in fig .
10 represents the results when the af order is suppressed , i.e. , @xmath34 is fixed to zero .
it is apparent that the presence of the af order has a small effect on the expectation values @xmath115 and @xmath114 .
the coexistent state between the d - wave sc and af order parameters is stabilize up to the doping rate @xmath228 .
this is consistent with the vmc simulations .
the actual expectation values in the wave function @xmath229 with projection are different from @xmath49 and @xmath34 .
for example , the expectation value of @xmath230 is _ exp c_i^c_j
= g_t c_i^c_j_0 = g_t . in the similar way
, the actual expectation values in @xmath229 are obtained using corresponding gutzwiller factors .
for @xmath231 we repeat the similar arguments as for @xmath36 to obtain & & m_exp n_i - n_i = g_m m , + & & g_m = a^-4 . + & & for the expectation value of @xmath232 , we have to take care of the fact that the number of holes changes in the cell . here
we use the average number of holes as an approximation . in the same level of approximations as for @xmath28
we obtain _ exp & & c_i^c_j^= g _ ,
+ g_&&= a^-(n_b - _b)+8 + & & = a. figure 11 shows the actual expectation values , @xmath233 and @xmath231 as a function of the doping @xmath33 for @xmath7 .
in this paper we have developed a new approach for studying the effect of strong correlation or the gutzwiller s projection in the two - dimensional @xmath0-@xmath1 model .
it is based on the extended gutzwiller approximation , in which the effects of longer - range correlations are taken into account .
these correlations play important roles for the interplay between the af and d - wave sc .
let us summarize our main results and discuss related problems .
\(1 ) generally the expectation values with respect to the projected wave function are strongly renormalized due to the exclusion of double occupancies . in the slave - boson mean - field theory ,
this effect is taken into account by assuming the replacement @xmath234 and @xmath235 .
however we have shown that the renormalization is not so simple .
first of all the renormalization factor for the exchange interaction is anisotropic , i.e. , @xmath236 in the presence of af moment .
furthermore @xmath237 and @xmath36 have nonlinear dependences on the expectation values @xmath49 and @xmath34 . in this sense
the extended gutzwiller approximation is beyond the simple slave - boson mean - field theory .
we think that essence of the strong correlation is contained in these gutzwiller factors , because they stem solely from the projection .
the physical meanings of the gutzwiller factors are clarified : \(i ) the factor @xmath151 appearing in @xmath38 and @xmath36 represents the exclusion effect shown in fig .
when we calculate @xmath238 , the surrounding six bonds can not make contribution of nearest neighbor correlation @xmath124 . as is evident in fig . 1
, the factor @xmath151 reduces the value of @xmath239 even for the case @xmath164 .
\(ii ) the enhancement of @xmath36 as a function of @xmath34 is the most important feature .
this was observed in the vmc simulations.@xcite here we have identified its origin as the effect of surrounding af correlations by discussing the probabilities of spin configurations in fig .
the enhancement of @xmath36 is due to the increase of @xmath170 configuration caused by the af circumstances .
\(2 ) we have studied the projected variational states as an application of the new gutzwiller approximation .
it is shown that our scheme reproduces the results in the vmc simulations.@xcite the enhancement of @xmath36 as a function of @xmath34 is essential for reproducing the coexistent state of the af and d - wave sc orders found in vmc.@xcite \(3 ) at half filling ( @xmath41 ) , there are several interesting features . in the original gutzwiller approximation ,
the af state at half filling was not obtained@xcite which was unphysical .
however in the present gutzwiller approximation , the af state with d - wave sc order parameter becomes the most optimized state as obtained in vmc .
note that the expectation value of the sc order parameter , @xmath233 , is zero because of the gutzwiller factor @xmath240 at half filling , although @xmath115 and the variational parameter @xmath241 are nonzero .
the self - consistent values which we obtain are ( fig .
10 ) ^v & & = 0 , + ^v & & = ^v = 0.080 , + & & = = 0.163 , + m^v & & = 0.017 , +
m & & = 0.147 . [ scval ]
the actual expectation value with the projection , @xmath231 , is ( see fig .
11 ) m_exp= g_m m = 0.373 .
this is a reasonable magnitude and close to the monte carlo result for the heisenberg model ( @xmath242).@xcite the relations @xmath243 and @xmath244 are the manifestation of the su(2 ) symmetry at half filling.@xcite actually the self - consistency equations for @xmath114 and @xmath115 become the same at @xmath41 .
it is worth while noting here that , due to the su(2 ) symmetry , the coexistent state of the af and d - wave sc is equivalent to the @xmath245-flux state with af long range order discussed by hsu.@xcite the self - consistent value , @xmath246 in eq .
( [ scval ] ) , is smaller than the result @xmath247 in the simple mean - field theory .
instead it is slightly larger than the value at which the gutzwiller factor @xmath36 has a maximum , as shown in fig . 1 .
if @xmath34 is increased from the self - consistent value , @xmath36 decreases .
this indicates that @xmath34 is determined so as to minimize the exchange energy @xmath248 by optimizing the energy gain due to the long - range order and the energy loss due to the reduction of @xmath36 .
this situation is , in some sense , similar to the quantum fluctuations of the heisenberg spin system discussed in the spin - wave theory : the spin fluctuation , which reduces @xmath34 from the mean - field value @xmath249 , leads to the enhancement of the gutzwiller s renormalization factors to gain the energy .
\(4 ) for less - than - half - filled case , the difference between the variational parameter @xmath241 and the expectation value @xmath233 is remarkable : @xmath241 is finite and moreover it increases as the doping rate @xmath32 decreases , while @xmath233 is proportional to @xmath32 near half filling .
this is because the gutzwiller factor , @xmath250 , is proportional to @xmath32 ( see section vi ) .
we interpret that @xmath241 is the bcs - type energy gap observed in scanning tunnel spectroscopy@xcite or in break junctions@xcite , because it is the parameter embedded in the wave function even at half filling .
the excitation spectra will have a large energy gap corresponding to @xmath241 , while the true long - range order , @xmath233 , is reduced due to the projection or the strong correlation .
the increase of @xmath241 in decreasing @xmath32 is consistent with the dependence observed experimentally .
\(5 ) we discuss here the relation to the so(5 ) theory.@xcite in our formulation , the combination @xmath251 appears frequently .
if the so(5 ) symmetry is exact in the @xmath0-@xmath1 model , the free energy will have a systematic dependence such as f(^2+m^2 ) f(_exp^2+m_exp^2 ) .
for example , the numerator in @xmath252 has a combination 16(m^2+^2+^2 ) .
also the factor giving the enhancement of @xmath36 contains 4m^2+x_2 = 4m^2 + 2 ^ 2 + 2 ^ 2 .
although these combinations remind us of the so(5 ) symmetry , our gutzwiller approximation does not show exact symmetry .
nevertheless a tendency similar to the so(5 ) prediction can be seen from the gutzwiller factors in fig .
1 . comparing @xmath38 and @xmath36 for the cases with @xmath253 and @xmath44
, we find that @xmath38 and @xmath36 are larger for smaller @xmath115 .
this is mainly from the exclusion effect of @xmath252 , because even at @xmath164 this effect is observed .
therefore if we consider a situation where the d - wave sc order parameter is suppressed , then the gutzwiller factor is enhanced causing the increase of af moment .
this gives the similar phenomena predicted in the so(5 ) theory.@xcite \(6 ) one advantage of the present theory is that it is easily applied to the inhomogeneous cases , such as the stripe state , vortex cores , and magnetic states around nonmagnetic or magnetic impurities , where the interplay between the af and d - wave sc correlations plays an important role .
since our gutzwiller approximation gives a reasonable estimate of the variational energies in the presence of af correlations , a reliable analytic formulation can be given .
the simplest way of applying the present scheme to inhomogeneous problems is to assume that @xmath45 and @xmath28 for each bond @xmath254 are determined locally from the expectation values @xmath255 and @xmath256 for the bond . in this case , if the d - wave order parameter is reduced around vortex cores , impurities or stripes , then the gutzwiller factors @xmath38 and @xmath36 are enhanced locally as expected from fig . 1 .
this effect causes the local development of af correlations , which can be observed experimentally . from these viewpoint ,
preliminary calculations for the vortex cores@xcite and stripe states@xcite have been published elsewhere .
of course the vmc simulations give more accurate evaluation of the variational energies , if they are used for the inhomogeneous systems like stripe states .
however there are some difficulties in applying the vmc .
firstly we have to treat a fairly large unit cell to study the slowly varying order parameters especially near half filling.@xcite for example , the incommensurability in the stripe state is close to @xmath257 so that the period of the stripe pattern becomes fairly long . in this case
the vmc simulations become difficult . furthermore the choice of the functional form of the trial state in vmc is restricted , since only the small number of variational parameters can be used practically . on the contrary , our scheme based on the mean - field - type gutzwiller approximation can be used in a fairly large system sizes .
moreover the order parameters on all the bonds , @xmath255 and @xmath258 can be optimized in the similar sense to unrestricted hartree - fock theory .
therefore we can search for the microscopically optimized variational states in our scheme.@xcite the authors wish to thank t. m. rice , h. fukuyama , y. tanaka , h. tsuchiura and m. sigrist for useful discussions .
they also thank c .- m . ho for critical reading of the manuscript .
this work is supported in part by a grant - in - aid of of the ministry of education , science , sports and culture .
the evaluation of eq .
( [ oexp ] ) is slightly complicated .
the constraints for @xmath259 are _
n_i n_i & & = -n_i_0 , _
i=1^k n_i n_i = -n_i_0 , + _ i=1^k n_i & & = n-
n_e -n_hi_0 , instead of ( [ constr ] ) because @xmath260 represents the number of cells of the @xmath93-th state except for the central cell . in the same way as in the denominator , the largest term is given by = ( p)^ , with slightly different values of @xmath261 and @xmath262 because of the difference of the constraints .
from the constraints ( a1 ) , we can see that @xmath263 satisfies the relations _
i=1^k = -1 , _
i=1^k = n- n_e - n_hi_0 .
therefore if we define the difference @xmath264 , we have important relations _ i=1^k = -1 , _
i=1^k = -n_hi_0 .
by use of these , the ratio between the numerator and the denominator in eq .
( [ numegen ] ) is calculated as follows : & & = + & & = _
i_0 _ i=1^k _
i_0 + & & = _ i_0 _ i=1^k ( ) ^- _ i_0 + & & = _ i_0 _ i=1^k ( ) ^- _ i_0 + & & = _ i_0 ( ) ^- p^- _ i_0 + & & = _ i_0 _
this gives eq .
( [ numegen ] ) .
in this appendix we show that the original gutzwiller approximation described in ii is reproduced if we assume only the site - diagonal expectation values in the generalized formulation in section iii .
first we calculate the weight @xmath94 of the @xmath93-th state which has @xmath97 up - spin electrons , @xmath98 down - spin electrons and @xmath99 holes .
if we take only the site - diagonal expectation values , we have _
i= [ r(1-w)]^n_right [ w(1-r)]^n_wrong [ ( 1-r)(1-w)]^ , where @xmath118 ( @xmath119 ) means the number of sites where the right ( wrong ) spices of spin direction is located depending on the sublattice 1 and 2 . then the total weight in the subgroup with @xmath63 holes ( eq . ( [ wj ] ) ) can be calculated exactly as w_j & & = _
i with j holes _
i + & & = _ n_c c_j [ r(1-w)+w(1-r)]^n_c - j ^j + & & = _ n_c c_j ( n-2rw)^n_c - j ( 1-r)^j ( 1-w)^j , where @xmath265 is the number of choices of the positions of @xmath63 holes . using these @xmath106 , the constraints ( [ constr3 ] ) have simple forms as & & _ j=0^n_c p^j = \ { n-2rw + ( 1-r)(1-w)p}^n_c= 1 , + & & _ j=0^n_c j p^j = ( 1-r)(1-w)p + & & \ { n-2rw + ( 1-r)(1-w)p}^n_c-1= n_c .
these can be solved easily to give p= , + w= ( ) ^n_c .
finally the expectation value for @xmath109 , for example , becomes & & + & & = _
i_0 s_^+ s_m^- _
i_0 + & & = _ j=0^n_c-2 _ n_c-2c_j ( n-2rw)^n_c-2-j ( 1-r)^j ( 1-w)^j s_^+ s_m^- _ 0 + & & = \ { n-2rw + ( 1-r)(1-w)p}^n_c-2 s_^+ s_m^- _ 0 + & & = s_^+ s_m^- _ 0 , which is exactly the same as the results in the original gutzwiller approximation in section ii.b , eq .
( [ gs0 ] ) .
in the similar way to @xmath38 , we evaluate @xmath36 . in the zero - hole sector , we have _
i_0 with 0 holes s_^z s_m^z _ i_0 & & = s_^z s_m^z _ 0 ( n-2rw)^n_c-2 a^_b + -&&2n_2 m^2 x_2 ( n-2rw)^n_c-3 a^_b-3 + -&&n_2 ^ 2 m^2 x_2 ^ 2 ( n-2rw)^n_c-4 a^_b-6
. + & & for the one - hole sector , there are five contributions from the diagrams in fig .
the first three diagrams are simple extensions of the zero - hole sector .
however figs .
9(d ) and 9(e ) are new - type contributions due to the presence of a hole .
for the diagrams in figs .
9(d ) and 9(e ) we calculate & & = s_^z s_m^z ( 1-n_m)(1-n_m ) _ c + & & = m^2x_2 , + & & = \ { n_(1-n_)+ n_(1-n_ ) } s_^z s_m^z + & & ( 1-n_m)(1-n_m ) _ c + & & = m^2x_2 ^ 2 . using these expectation values ,
we obtain & & _
i_0 with 1 holes s_^z s_m^z _
i_0 + & & = _
n_c-2c_1 s_^z s_m^z _ 0 z ( n-2rw)^n_c-3 a^_1b + & & -_n_c-3c_1 2n_2 m^2 x_2 z ( n-2rw)^n_c-4 a^_1b-3 + & & -_n_c-4c_1 n_2 ^ 2 m^2 x_2 ^ 2 z ( n-2rw)^n_c-5 a^_1b-6 + & & + n_2 m^2 x_2 ( n-2rw)^n_c-3 a^_b-3 + & & + n_2 ^ 2 m^2 x_2 ^ 2 ( n-2rw)^n_c-4 a^_b-6 . in the two hole sector , we need to calculate a diagram in fig .
9(f ) which is @xmath266 . taking account of these diagrams and counting the higher order terms with respect to @xmath124 , we approximate as & & _ i_0 with 2 holes s_^z s_m^z _
i_0 + & & = _
n_c-2c_2 s_^z s_m^z _ 0 z^2 ( n-2rw)^n_c-4 a^_2b + & & - _
n_c-3c_2 2n_2 m^2 x_2 z^2 ( n-2rw)^n_c-5 a^_2b-3 + & & - _ n_c-4c_2 n_2 ^ 2 m^2 x_2 ^ 2 z^2 ( n-2rw)^n_c-6 a^_2b-6 + & & + _ n_c-3c_1 n_2 m^2 x_2 z ( n-2rw)^n_c-4 a^_1b-3 + & & + _
n_c-4c_1 n_2 ^ 2 m^2
x_2 ^ 2 z ( n-2rw)^n_c-5 a^_1b-6 + & & - m^2 x_2 ^ 2 ( n-2rw)^n_c-4 a^_b-6 .
finally generalization to higher order contributions leads to & & s_^z s_m^z = _
j=0^n_c p^j _ i_0 with j holes s_^z s_m^z _ i_0 + & & = _ j=0^n_c-2 p^j
_ n_c-2c_j s_^z s_m^z _ 0 z^j(n-2rw)^n_c-2-j a^_j , b + & & - _ j=0^n_c-3 p^j _
n_c-3c_j 2n_2 m^2 x_2 z^j ( n-2rw)^n_c-3-j a^_j , b-3 + & & - _
j=0^n_c-4 p^j _
n_c-4c_j n_2 ^ 2 m^2 x_2 ^ 2 z^j ( n-2rw)^n_c-4-j a^_j , b-6 + & & + _
j=0^n_c-3 p^j+1
_ n_c-3c_j n_2 m^2 x_2 z^j ( n-2rw)^n_c-3-j a^_j , b-3 + & & + _
j=0^n_c-4 p^j+1
_ n_c-4c_j n_2 ^ 2 m^2 x_2 ^ 2 z^j ( n-2rw)^n_c-4-j a^_j , b-6 + & & - _
j=0^n_c-4 p^j+2 _
n_c-4c_j m^2 x_2 ^ 2 z^j ( n-2rw)^n_c-4-j a^_j , b-6 + & & = ( ) ^2 s_^z s_m^z _ 0 a^-(n_b - _b ) + & & - ( ) ^3 2n_2 m^2 x_2
( 1- ) a^-(n_b- n_b ) + & & - ( ) ^4 n_2 ^ 2 m^2 x_2 ^ 2 ( 1-)^2 a^-(n_b- n_b ) + & & = ( ) ^2 a^-(n_b - _b ) + & & .
thus the gutzwiller factor is g_s^z = g_s^xy . | a new type of analytic estimation of the effect of strong correlation is developed for the two - dimensional @xmath0-@xmath1 model .
it is based on the gutzwiller approximation which gives the renormalization of parameters , @xmath0 and @xmath1 , due to the gutzwiller s projection operator excluding the double occupancies .
the finite - range correlations are taken into account compared with the conventional gutzwiller approximation where only the on - site expectation values are considered .
it is shown that the essential point of the renormalization is its nonlinear dependence on the expectation values of cooper pairs and antiferromagnetic moment .
in particular the renormalization factor of @xmath1 becomes anisotropic in the presence of antiferromagnetic moment , i.e. , that for the @xmath2-component is enhanced compared with that for the @xmath3-component .
the physical origin of this enhancement is identified as the surrounding antiferromagnetic correlations of a bond .
the self - consistency equations for the uniform variational wave functions are derived and solved numerically .
our result gives a reasonable estimate of antiferromagnetic order parameters near half filling , in contrast to the conventional slave - boson mean - field theory and the original gutzwiller approximation .
it is also found that , at half filling , the renormalization of @xmath1 represents some of the quantum fluctuations of the heisenberg spin system in a different manner from the spin - wave theory . for finite doping ,
our results have some similarity to so(5 ) symmetric theory .
applications to the inhomogeneous systems such as the vortex state , around nonmagnetic impurities , and stripe state are discussed . |
interactions of atoms with the electromagnetic field near micro and nano - scale dielectric structures is currently a topic of great interest @xcite .
on the one hand , atoms in the vicinity of microscopic resonators can interact strongly with photons in the resonator leading to quantum information applications such as single photon switching @xcite . on the other hand ,
recent advances have been made in trapping arrays of atoms near the surface of nanofibers using guided modes @xcite for which large optical densities are realizable a feature which is also conducive to quantum information as well as more general quantum optics applications .
trapping of single atoms near to nanostructure based cavities has also seen recent advances @xcite .
( a ) depiction of the trapping scheme showing a nanofiber mounted on a grating illuminated from below ( i.e. from @xmath0 towards @xmath1 ) by a focused laser beam .
this conceptual diagram omits the grating substrate for clarity .
scattering of the diffracted beam by the fiber produces an array of traps near the fiber surface .
( b ) schematic diagram showing the regions of importance for the theoretical derivation of the trapping field .
a fiber of radius @xmath2 is mounted on a grating of period @xmath3 , for integer @xmath4 , depth @xmath5 and slat width @xmath6 , where @xmath7 is the grating duty cycle .
a @xmath8-polarized plane wave is incident on the grating from the substrate side at an angle @xmath9 .
the grating diffracts the incident wave into three orders with angles @xmath10 and @xmath11 with respect to the fiber axis .
these plane waves are then scattered by the fiber giving rise to a trapping field @xmath12 . ] nonetheless , a number of challenges remain in this area .
one problem is the trapping of atoms in optimum positions for coupling to nanowaveguide - based cavities @xcite . in standard ,
freespace cavity qed experiments , this problem has already been solved with applications to atomic self organization @xcite and atomic spin squeezing @xcite , but these methods do not translate well to the nanowaveguide case .
another challenge is the vector light shift imparted on the atoms when they are trapped using the guided mode of a nanowaveguide @xcite . as an additional matter , the observation of collective excitation effects such as superradiance using nanofibers @xcite would be greatly simplified if the spacing of a trapping array could be decoupled from the wavelength of light used to create it .
here we describe a scheme to trap atoms in arrays near to the surface of a composite nanostructure which can potentially solve these problems as well as provide new features compared to the aforementioned studies .
the device we consider here is a nanostructure comprised of a nanofiber mounted on a grating .
similar devices have recently been experimentally investigated for the purposes of enhancing the coupling of a quantum emitter to the guided modes of an optical nanofiber @xcite . here
, we will show that by illuminating such a device , it is possible to create periodic arrays of trapping potentials near to the nanofiber surface as depicted in fig .
[ fig : setup](a ) .
the basic principle of the device is that a longer period grating ( referred to here as a _ trapping grating _ ) between two bragg gratings on a nanostructured silica substrate can serve the dual purposes of 1 ) enhancing emission into the integrated nanofiber by acting as a cavity and 2 ) acting as a first order grating for trapping light which illuminates the device .
the principle advantages of the proposed device are as follows : a ) an array of traps can be formed along the nanofiber surface , with the number of traps and their period determined by the design of the nano - structured substrate , rather than the wavelength of the trapping light . a major application of this ,
as we will show later , is trapping devices where the trapping sites are _ geometrically guaranteed _ to have their centers aligned to antinodes of a cavity mode of the device .
b ) unlike trapping schemes which use the evanescent tails of the guided modes , in our scheme no trapping light is present in the guided mode , which should lead to improved signal - to - noise ratios for detection of photons emitted into the guided mode .
c ) unlike trapping arrays formed by the guided modes of the nanofiber , the traps formed by this technique have linear rather than elliptical polarization .
this means the potentially decohering effect of vector light shifts @xcite is not present for our device .
d ) the composite nature of the device has a number of benefits as given in @xcite .
of particular importance in this instance is the separation of the nanowaveguide function and the index modulation function into separate devices .
this allows for much greater diffraction efficiencies into the @xmath13st order compared to what could be achieved with structures patterned directly onto the waveguide , leading to relatively larger trapping depths along the fiber axis .
the paper proceeds in the following way .
first we consider the trapping potential formed when a plane wave is incident on the system with an infinitely long trapping grating and no bragg - grating region , as depicted in fig .
[ fig : setup](b ) .
this system may be analysed both numerically and using an analytical model which allows a crosscheck of the numerically calculated potential .
having established the validity of our numerics , we then go on to simulate the full device , including the bragg mirror regions , using fdtd simulations .
the device is depicted in a to - scale conceptual image in fig . [
fig : setup](a ) and in schematic form in fig .
[ fig : setup](b ) .
( note that the grating substrate is omitted in fig .
[ fig : setup](a ) for clarity ) .
an optical nanofiber of radius @xmath2 is mounted on a nano - structured substrate and illuminated from the grating side ( single - illumination configuration ) or from both sides ( dual - illumination configuration ) by a red - detuned laser beam with a vacuum wavelength @xmath14 .
the corresponding wave number of the trapping laser is @xmath15 and the trapping laser frequency is @xmath16 , where @xmath17 is the speed of light in the vacuum .
the nano - structured substrate is divided into two regions . to either side of the center ,
a bragg grating region is present .
the two bragg gratings , which have period @xmath18 act as mirrors with respect to the fundamental mode of the nanofiber @xcite .
the central region is the trapping grating ( shown in fig .
[ fig : setup](b ) ) with period @xmath19 for some real number @xmath4 , i.e. , the trapping grating is a first order grating relative to the incident trapping light .
we characterize the length of the trapping region by counting the number of regions @xmath20 of width @xmath3 . in this paper , @xmath20 is always an odd number .
we note that with respect to the bragg mirrors , the central trapping grating region may be considered to constitute a cavity of total width @xmath21 .
this cavity region acts to enhance coupling of spontaneous emission to the fiber guided modes . in all regions of the nano - structured substrate , the grating slats have depth @xmath22 m and width @xmath23 nm , where @xmath7 is the bragg grating duty cycle . as shown in fig .
[ fig : setup](b ) , light incident on the device at angle @xmath9 is diffracted into the @xmath24th and @xmath13st orders at angles @xmath25 and @xmath26 respectively . in the remainder of the paper , we will take @xmath27 implying that the light is normally incident on the grating substrate .
we define our axes so that @xmath28 is always in the middle of the trapping grating region and @xmath29 corresponds to the fiber center as shown in fig .
[ fig : setup](b ) . the unit vectors
@xmath30 , @xmath31 and @xmath32 are defined to lie along the @xmath33 , @xmath8 and @xmath34 axes respectively . in the analytical treatment which follows below , we will only consider the case of normally incident , @xmath8-polarized trapping light which is red - detuned from the atomic resonance .
the effect of changing the incident angle , polarization or detuning will be discussed in the discussion section .
by applying the theory of a planewave scattered from a dielectric grating @xcite together with scattering theory for a sub - wavelength dimension cylinder @xcite the field outside the device for the incident @xmath8-polarized field can be found .
a detailed derivation is given in the appendix .
to summarize the theoretical approach : we make the approximation that evanescent orders of the grating may be ignored , thereby truncating the series which gives the output field .
we also assume that the field around the nanofiber can be found by taking the field at the output of the grating as the input field to the scattering problem involving the nanofiber . to find the field around the fiber we need to find the scattered field due to the @xmath24th and @xmath13st order plane waves .
to apply the standard scattering formalism , we first evaluate the angles @xmath35 where @xmath36 as shown in fig .
[ fig : setup](b ) . in the present paper
, we will consider only the case where the trapping beam is at normal incidence to the grating , giving @xmath37 .
the other angles are given by @xmath38 , where @xmath39 is the kronecker delta function , @xmath40 is the wavenumber of the @xmath33 component of the diffracted field and @xmath41 is the wavenumber of the @xmath34 component of the diffracted field . the field for @xmath42
is then given by @xmath43,\end{aligned}\ ] ] where @xmath44 is the field due to the grating in the absence of the nanofiber ( see appendix ) , @xmath45 is the transmission coefficient of the @xmath46th diffraction order , @xmath47 and @xmath48 are @xmath49th order cylindrical harmonics whose coefficients @xmath50 and @xmath51 respectively depend on the incident angles of the transmitted orders of the grating .
these coefficients are defined fully in the appendix , where a brief review of scattering theory for a dielectric cylinder is given for the case considered here . in the dual illumination scheme , a second laser illuminates the device from the fiber side . to derive the trapping field in this case
, we will assume that the only fields to be considered are those of the second trapping laser itself , and the scattered field caused by the second trapping laser being incident on the nanofiber .
that is , we ignore any reflections from the grating or , indeed , any effects due to the grating at all . because of the low reflectivity of the grating at the wavelength considered here , this approximation is not a bad one . in particular , for the space @xmath42 , the assumption turns out to produce good agreement with fdtd simulation results .
the incident field is given by @xmath52 where @xmath53 is the relative phase between trapping laser 1 and trapping laser 2 which we will assume is adjustable .
the scattered field is then @xmath54 where the sign of @xmath33 in the argument of the spherical harmonics @xmath47 has been flipped to account for the fact that the incident wave is coming from the negative @xmath33 direction , and the terms in @xmath48 are zero due to the normal incidence of the second trapping beam . finally , we may write the field due to illumination by both trapping beams 1 and 2 as @xmath55 note that in what follows , we will use @xmath12 to denote either the trapping potential for single or dual illumination .
the meaning of @xmath12 will always be clear from the context in which it is used . to calculate the trapping potential experienced by @xmath56cs atoms , we follow closely the formalism of ref .
the optical trapping potential may be calculated as @xmath57 where @xmath58 the index @xmath59 runs over different transitions of cesium that contribute to the trapping potential . for our purposes
, we will include the four dominant lines of the atom @xcite corresponding to wavelengths @xmath60 nm , @xmath61 nm , @xmath62 nm and @xmath63 nm , where @xmath64 .
the transition strengths for these lines are @xmath65s@xmath66 , @xmath67s@xmath66 , @xmath68s@xmath66 and @xmath69s@xmath66 .
the statistical weights of each transition are @xmath70 , @xmath71 , @xmath72 , @xmath73 and the ground state has weight @xmath74 .
finally , we define the linewidths @xmath75 , where @xmath76 denotes a lower level @xmath77 .
however for the detuning of the trapping light considered here , the contribution of the terms @xmath78 to @xmath7 are negligible . for convenience ,
all of our analytical and simulation results assume an electric field amplitude of 1 .
we scale the potential to provide trap depth predictions for a given optical power as follows . for simplicity
, we use the waist region of a circular gaussian beam as a reference .
we assume a power @xmath79 and a beam waist radius of @xmath80 .
then the peak intensity in the beam is given by @xmath81 .
the scaling factor for the optical part of the trapping potential is then @xmath82 , where @xmath83 is the vacuum permittivity .
next , we consider the contribution of the van der waals potential at the fiber surface to the trapping potential . here
we will use the simple flat - surface , bulk - medium van der waals potential experienced by a cs atom close to a silica surface which is given in mk by @xmath84 where @xmath85 is measured in @xmath86 m .
this flat potential approaches the van der waals potential due to a silica fiber when the distance from the fiber surface approaches zero @xcite .
the van der waals potential due to the grating is neglected as the trapping potential minima always lie between two grating slats , i.e. , @xmath87 nm away from the nearest slat for the parameters used in this paper .
this is sufficiently far that the contribution from the grating slats to the van der waals potential is negligible .
the total trapping potential is the sum of the optical and van der waals potentials : @xmath88 where it is understood that the factor @xmath89 is only necessary to scale the optical potential if the analytical results and/or simulations have used incident fields of unity amplitude .
comparison of intensities @xmath90 as predicted by theory and fdtd simulations .
( a ) and ( b ) show @xmath91 in the @xmath92 plane for @xmath93 as given by theory and fdtd numerics respectively . in ( c ) and ( d ) , @xmath91 from theory and simulations respectively is shown in the @xmath94 plane for @xmath95 m .
the fiber radius was @xmath96 m and @xmath97 m with @xmath98 . ] trapping potential produced near the nanofiber surface for ( a ) @xmath99 , ( b ) @xmath95 m , @xmath28 and ( c ) @xmath100 m , @xmath101 respectively , corresponding to figs .
[ fig : intens](a - c ) . in all cases ,
solid blue lines show the theoretical prediction while red lines show the predictions from fdtd simulations .
( d ) shows how the calculated position of the trapping minimum ( blue line ) and trap depth ( red line ) vary as a function of the fiber radius .
the blue crosses show the trap position and the red circles show the trap depth given by the fdtd simulation in each case .
the optical power was set to 250 mw . ]
we simulated the electromagnetic field in the system shown in fig .
[ fig : setup](b ) for an incident plane wave using the finite difference time domain ( fdtd ) method ( lumerical inc . ) . here ,
and throughout the remainder of this paper , we assume an optical power of 250 mw per beam , a beam spot size of 10 @xmath86 m and a trapping light wavelength of 937 nm ( the red detuned magic wavelength for @xmath56cs ) in order to calculate the trapping potential generated by the intensity distribution . figure [ fig : intens ] shows a comparison between squared electric field amplitudes as calculated from eq .
[ eq : etrapi ] and by using fdtd numerical simulations .
we see good agreement between the squared field profiles in the @xmath92 plane ( figs .
[ fig : intens](a ) and ( b ) ) and the @xmath94 plane ( figs . [
fig : intens](c ) and ( d ) ) both in the qualitative shape of the field pattern and the intensity .
in particular , figs . [
fig : intens](a ) and ( b ) show the generation of a periodic array of intensity maxima close to the nanofiber surface .
[ fig : intens ] ( c ) and ( d ) show that along the axis perpendicular to the fiber , for @xmath33 set to the position of the intensity maximum in the @xmath92 plane , a strong intensity maximum coincides with the center of the fiber . away from the center
is an area of low intensity , and for larger @xmath8 , the diffraction pattern due to the grating is restored since the field scattered by the fiber tends to zero in this region .
figures [ fig : pot](a - c ) show the corresponding trapping potential as a function of @xmath33 , @xmath34 and @xmath8 respectively , as calculated from eq .
[ eq : trappot ] and from the fdtd data in all cases assuming a power of @xmath102 mw .
we see that potential depths of mk order are achievable in all directions for these conditions , although the effect of the van der waals potential significantly lowers the potential depth along the @xmath33-axis .
[ fig : pot](d ) shows how the trapping depth @xmath103 and the trap minimum position @xmath104 on the @xmath33-axis vary as the fiber radius @xmath2 is varied .
it may be seen that increasing the nanofiber radius leads to the trapping potential moving closer to the nanofiber surface , with the trapping potential depth @xmath105 decreasing due to the effect of the van der waals potential .
it may be seen that for trapping minima positions less that @xmath106 nm the potential depth becomes vanishing .
.[tab : singleillum ] theoretical and fdtd calculated trapping frequencies for single illumination . [ cols="<,<,<",options="header " , ] table [ tab : singledual ] compares the trapping frequencies for fdtd calculated trapping potentials in the cases of single and dual illumination as shown in figs .
[ fig : gaussillumsingle](e - g ) and [ fig : gaussillumdouble](d - f ) respectively . as expected , we find essentially no change along the @xmath8-axis , and the change along the @xmath34-axis is small
. however the @xmath33 axis value is more than quadrupled due to the tighter trap created by the standing wave as well as the closer proximity of the trap to the fiber surface .
we now consider aspects of our system not discussed in the preceding sections .
first the question arises as to our choice of parameters for the trapping light . regarding blue detuned trapping schemes ,
because the focusing effect of the nanofiber determines the distance of the trapping potential minimum from the fiber surface and the focus is always an intensity maximum , trapping sites tend to be further from the nanofiber in the blue detuned case .
we also calculated the potential at the blue - detuned magic wavelength of 686 nm , but we found that red - detuned trapping light produced traps closer to the nanofiber surface .
as for the polarization of the trapping beam , in this work we used exclusively @xmath8 polarized light .
this polarization is parallel to the grating slats , so that the output polarization of the grating is also @xmath107polarized . on the other hand ,
@xmath107 polarized illumination of the nanofiber produces a field which includes localized @xmath108polarized components at azimuthal angles of @xmath109 on the fiber surface .
nonetheless , we found that the polarization of the trapping field is also @xmath8-polarized near the trapping minima to a good approximation , with less than @xmath110 @xmath108polarization at the trap center , and slightly less than @xmath111 @xmath108 polarized at 300 nm away from the trap center in the @xmath107direction ( corresponding to a temperature of @xmath112 mk ) .
additionally , we found that illuminating the device with @xmath34-polarized light did not produce trapping potentials due to the mixture of polarizations resulting after diffraction by the grating .
next , we note that in the case of non - normal incidence , a diffraction pattern ( and thus a trap ) is only produced if the incident angle @xmath113 .
this is because the gratings considered here are strictly first order gratings .
this means that the device must be precisely aligned in order to produce a trapping potential .
additionally , here we have principally considered the trapping sites closest to the fiber surface .
this is justified by the exponential scaling of the coupling between a dipole emitter and the fiber guided mode , atoms trapped in more distant trapping sites have negligible coupling to the nanofiber @xcite .
another issue to consider in the present scheme is the possibility that large light intensities might affect the integrity of the very thin silica slats on the grating .
however , the slats are at intensity nodes of the trapping field for the cases considered here . additionally , the trapping intensity produced by the interference of diffracted orders of the grating is only of order twice the input intensity . for these reasons the thin grating slats are not in general exposed to large field intensities .
lastly , we consider issues related to the tuning of the cavity .
illumination of the device may cause thermal expansion and thus drifting of the cavity mode from its design wavelength .
we note , however , that much of the optical power for circular beam illumination is unused , due to the thinness of the nanofiber .
if , for example , we used an elliptical beam with a @xmath34-axis radius of 10 @xmath86 m as before but an @xmath8-axis radius of only 1 @xmath86 m , the optical power required to achieve the same intensity at the device surface would be only @xmath111 of that necessary in the case of a circular beam ( i.e. @xmath114 mw ) . in this way heating of the silica substrate would be drastically reduced .
if tuning of the cavity resonance is required , it can be achieved by slightly rotating the nanofiber relative to the grating slats ( increasing the resonance wavelength ) or by lifting the fiber slightly off the grating substrate ( reducing the resonance wavelength ) . in conclusion , we have analytically and numerically investigated trapping atoms close to the surface of a composite nanooptical device consisting of an optical nanofiber mounted on a grating .
illumination of the device from the grating side produces an array of traps with separations from the nanofiber surface down to 200 nm . adding illumination from the fiber side
can produce trapping potentials within 100 nm of the nanofiber surface .
bragg gratings to either side of the trapping grating create a cavity enhancing the coupling of spontaneous emission into the nanofiber guided modes , and by appropriately designing the grating structure , arrays of atoms may be trapped at positions which correspond to antinodes of the cavity mode .
because optical nanofibers offer automatic coupling to standard fiber optical networks , and due to the advantages of the present scheme over existing nanofiber trapping schemes , we expect that it may be useful in a variety of situations .
these include the realization of collectively enhanced scattering into the nanofiber guided modes from an array of atoms , through to quantum information applications where distant atoms must be coupled to the same cavity mode .
ms acknowledges support from a grant - in - aid for scientific research ( grant no .
15k13544 ) from the japan society for the promotion of science ( jsps ) .
kpn acknowledges support from a grant - in - aid for scientific research ( grant no .
15h05462 ) from the japan society for the promotion of science ( jsps ) .
the authors thank k. hakuta for his scientific input at the early stages of this research and for his support .
we also thank f. le kien for discussions regarding the van der waals potential , m. morinaga for advice regarding the scattering problem , and r. yalla for discussions related to the numerical simulations .
for comparison with theoretical results , we performed fdtd simulations using plane wave sources in our simulations . for scaling the potential ,
we choose a beam waist radius of @xmath115 m , an optical power of @xmath116 mw , and a wavelength of @xmath117 nm . in the dual illumination case , we assumed beam recycling was possible so that the power in both forward and backward propagating beams was set to @xmath79 . in fdtd simulations of periodic systems using plane waves
, care must be taken in interpreting the results since artifacts of the fdtd method can form regular periodic intensity maxima of a similar nature to the true trapping maxima we wish to find .
we used periodic boundary conditions for these simulations to ensure that no edge - effects due to the fdtd boundary conditions were present .
we also checked that removing all structures in the simulation gave rise to a flat intensity profile , and that removing the grating structure led to the disappearance of all @xmath34-periodic structures in the output intensity pattern .
once these checks had been passed , we could be confident that any local intensity maxima in the simulations corresponded to the physical nature of the scattered field rather than numerical artifacts .
comparison with the analytical results given in the previous sections also functioned as a further cross - check of our numerical results .
we note that in the case of dual illumination , the relative phase @xmath53 is just the phase of the second beam as set in our fdtd simulations .
it is neccesary to match the phase between the theoretical trapping potential and the fdtd calculated trapping potential for a single data set before the fdtd and theory results can be compared for arbitrary phases . in the simulations for the case of realistic experimental structures
, we illuminated the device with a gaussian beam .
we used absorbing rather than periodic boundary conditions , and it was necessary to make the fdtd simulation region large enough that the gaussian beam tails had decayed effectively to zero by the edges of the simulation region to avoid edge effects such as reflection from the absorbing boundary .
however , computer memory limits the mesh density which can be achieved in these large simulation regions .
we used linear interpolation to reconstruct the results on a finer grid to improve the detail visible in the figures .
as before , we checked the validity of our simulations by removing all structures from the simulation region and checking that only a gaussian intensity profile remained in such cases .
in our analytical derivation of the trapping potential we will use the axes and variables defined in fig . [ fig : setup](b ) . in particular , we assume a grating structure with period @xmath3 , where @xmath18 is the bragg grating period , @xmath4 is an integer chosen so that the trapping grating is a first order grating .
the grating structure has a depth @xmath5 and the grating slats have thickness @xmath6 , where @xmath7 is the duty cycle .
the grating is assumed to have refractive index @xmath118 at the trapping light wavelength , and for simplicity the grating structure is assumed to be fabricated on top of a silica substrate which has much larger dimensions than any of the other scales in the problem .
( in typical experiments the substrate has been of size 15 mm x 5 mm x 2 mm @xcite ) .
the axes are defined as shown in fig .
[ fig : setup](b ) and the unit vectors @xmath30 , @xmath31 , and @xmath32 are aligned with the @xmath33 , @xmath8 , and @xmath34 axes respectively . a plane wave is assumed to be incident on the structure originating from the @xmath0 direction at an angle @xmath9 as shown in fig .
1(b ) of our manuscript .
the grating diffracts the incident wave into @xmath24th and @xmath119st orders with angles @xmath25 and @xmath120 respectively . in
what follows , standard harmonic time dependence @xmath121 is assumed for all fields and is not explicitly shown .
we first consider the electric field at the output of the grating _ in the absence of the nanofiber_. this problem was solved by knop in @xcite but due to a typographical error , the result given was ambiguous . for clarity
, we therefore reproduce the results of @xcite below giving the correct fromula which we have checked by rederiving the results . for an incident field @xmath122\mathbf{e}_y\ ] ] the field outside the grating is given by @xmath123\mathbf{e}_y,\ ] ] where @xmath124 is the @xmath125axis wavenumber and @xmath126 is the @xmath108axis wavenumber for diffraction order @xmath46 .
the coefficients @xmath45 give the ( complex ) amplitude of each diffracted component .
although they do not contribute to the trapping potential , we also define the @xmath108axis wavenumbers of the plane wave components reflected at the grating interface which are given by @xmath127 also , note that the phase term @xmath128 , which does not appear in @xcite , is necessary to retard the phase of the solution since the grating output is offset from the origin along the @xmath33-axis by @xmath2 ( the output of the grating is at @xmath29 in @xcite ) .
the coefficients @xmath45 of the different plane wave components are given by @xmath129 where @xmath130 is a matrix with components @xmath131 where @xmath132 is a matrix of eigenvectors and @xmath133 are the eigenvalues of the eigenvalue equation @xmath134 where @xmath135 , and @xmath136 is the @xmath46th fourier coefficient of the refractive index profile of the grating @xcite .
the vector @xmath137 is made up of zeros except for a single element at @xmath138 which has the value @xmath139 it is well known that the numerical inversion of @xmath130 required to calculate @xmath140 is unstable when significant numbers of evanescent diffraction orders are included in the calculation @xcite . in this paper , we will only consider first order gratings , i.e. all but the @xmath119th and @xmath24th diffraction orders are evanescent . as noted in @xcite ,
a simple way to avoid the problems due to inversion of near - singular matrices is simply to truncate the eigenvalue problem
. therefore we approximate @xmath140 as a length three vector and matrices @xmath130 , @xmath132 and @xmath141 are taken to have size @xmath142 .
this means that we neglect all evanescent orders of the grating which is a good approximation when the distance from the grating is more than the wavelength of the trapping light .
this approximation is justified by the good agreement it produces when compared to the results of fdtd numerical simulations .
we now review the analytical form of the field created when a plane wave is incident on a dielectric cylinder at oblique incidence with angle @xmath143 .
the treatment follows exactly that of ref .
@xcite and is included here only for convenience , not as a new result of our study .
the following results are valid for the case where the propagation vector @xmath144 of the incident wave lies in the @xmath92 plane and the polarization is parallel to the @xmath8 axis .
we first define cylindrical coordinates for our system .
the azimuthal coordinate @xmath145 is given by @xmath146 , and the radial coordinate is given by @xmath147 .
for the cylindrical unit vectors , we define @xmath148 , the radial unit vector , and @xmath149 , the azimuthal unit vector .
to proceed , we introduce the vector cylindrical harmonics which are given by the following formulae : @xmath150,\ ] ] @xmath151,\ ] ] where @xmath152 , @xmath153 and @xmath154 is the @xmath49th order hankel function .
note that although the arguments to the cylindrical harmonics are given as cartesian coordinates , calculations take place in polar coordinates .
the field scattered by the cylinder can now be written as @xmath155,\ ] ] where the coefficients @xmath50 and @xmath51 are defined as follows .
@xmath156 @xmath157 the @xmath49 dependent coefficients have the following defintions : @xmath158,\ ] ] @xmath159 @xmath160 @xmath161,\ ] ] @xmath162,\ ] ] where @xmath163 , and @xmath164
. finally , the total field is found by adding the incident field @xmath165 to the scattered field , i.e. @xmath166
we can now write down the formula for the electric field outside the device as used in our manuscript . we will approximate this field by the output field of the grating as scattered by the nanofiber .
noting that @xmath50 and @xmath51 depend on the incident angle @xmath143 , we write @xmath167 and @xmath168 .
then , by applying eq . [ eq : etot ] along with eq .
[ eq : escat ] for each diffraction order of the grating , we can write the field outside the device in the region @xmath42 for the case of single illumination as @xmath169.\end{aligned}\ ] ] 99 h.j .
kimble , the quantum internet , " nature * 453 * , 1023 - 1030 ( 2008 ) .
vahala , optical microcavities , " nature * 424 * , 839 - 346 ( 2003 ) .
t. aoki , b. dayan , e. wilcut , w. p. bowen , a. s. parkins , t. j. kippenberg , k. j. vahala , and h. j. kimble , observation of strong coupling between one atom and a monolithic microresonator , " nature 442 , 671 - 674 ( 2006 ) . f. le kien , v. i. balykin , and k. hakuta , atom trap and waveguide using a two - color evanescent light field around a subwavelength - diameter optical fiber , " phys .
a * 70 * , 063403 ( 2004 ) .
e. vetsch , d. reitz , g. sagu , r. schmidt , s. t. dawkins , and a. rauschenbeutel , optical interface created by laser - cooled atoms trapped in the evanescent field surrounding an optical nanofiber , " phys .
lett . * 104 * , 203603 ( 2010 ) .
a. goban , k. s. choi , d. j. alton , d. ding , c. lacrote , m. pototschnig , t. thiele , n. p. stern , and h. j. kimble , demonstration of a state - insensitive , compensated nanofiber trap , " phys .
lett . * 109 * , 033603 ( 2012 ) .
m. daly , v.g .
truong , c.f .
phelan , k. deasy and s. nic chormaic , nanostructured optical nanofibres for atom trapping , " new j. phys .
* 16 * , 053052 ( 2014 ) .
j. d. thompson , t. g. tiecke , n. p.
de leon , j. feist , a. v. akimov , m. gullans , a. s. zibrov , v. vuletic , and m. d. lukin , coupling a single trapped atom to a nanoscale optical cavity , " science * 340 * 1202 - 1205 ( 2013 ) .
l , hung , s.m .
meenehan , d.e .
chang , o. painter , and h.j .
kimble , trapped atoms in one - dimensional photonic crystals , " new j. phys * 15 * , 083026 ( 2013 ) .
s. kato , , and t. aoki , strong coupling between a trapped single atom and an all - fiber cavity , " phys .
lett . * 115 * , 093603 ( 2015 ) .
k. j. arnold , m. p. baden , and m. d. barrett , self - organization threshold scaling for thermal atoms coupled to a cavity , " phys .
* 109 * , 153002 ( 2012 ) .
j. lee , g. vrijsen , i. teper , o. hosten , and m.a .
kasevich , many - atom cavity qed system with homogeneous atom cavity coupling , " opt . lett . *
39 * , 4005 - 4008 ( 2014 ) .
o. hosten , n.j .
engelsen , r. krishnakumar , and mark a. kasevich ,
measurement noise 100 times lower than the quantum - projection limit using entangled atoms , " nature * 529 * , 505 - 508 ( 2015 ) .
f. le kien , p. schneewies , and a. rauschenbeutel , dynamical polarizability of atoms in arbitrary light fields : general theory and application to cesium , " eur .
j. d * 67 * , 92 ( 2013 )
. f. le kien , s. dutta gupta , k. p. nayak , and k. hakuta , nanofiber - mediated radiative transfer between two distant atoms , " phys . rev .
a * 72 * , 063815 ( 2005 ) .
m. sadgrove , r. yalla , k. p. nayak , and k. hakuta , photonic crystal nanofiber using an external grating , " opt . lett . * 38 * , 2542 - 2545 ( 2013 ) .
r. yalla , m. sadgrove , k. p. nayak , and k. hakuta ,
cavity quantum electrodynamics on a nanofiber using a composite photonic crystal cavity , " phys .
lett . * 113 * , 143601 ( 2014 ) .
j. keloth , m. sadgrove , r. yalla , and k. hakuta , diameter measurement of optical nanofibers using a composite photonic crystal cavity , " opt .
, * 40 * , 4122 - 4125 ( 2015 ) . c. lacroute , k.s .
choi , a. goban , d.j .
alton , d. ding , n.p .
stern , and h.j .
kimble , a state - insensitive , compensated nanofiber trap , " new j. phys . * 14 * , 023056 ( 2012 ) .
k. knop , j. opt .
* 68 * , 1206 ( 1978 ) . c. f. bohren and d. r. huffman , absorption and scattering of light by small particles " , wiley - vch ( 2004 ) .
m. g. moharam , eric b. grann , and drew a. pommet , j. opt .
* 12 * , 1068 ( 1995 ) . | we propose a method of trapping atoms in arrays near to the surface of a composite nanophotonic device with optimal coupling to a single cavity mode .
the device , comprised of a nanofiber mounted on a grating , allows the formation of periodic optical trapping potentials near to the nanofiber surface along with a high cooperativity nanofiber cavity .
we model the device analytically and find good agreement with numerical simulations .
we numerically demonstrate that for an experimentally realistic device , an array of traps can be formed whose centers coincide with the antinodes of a single cavity mode , guaranteeing optimal coupling to the cavity .
additionally , we simulate a trap suitable for a single atom within 100 nm of the fiber surface , potentially allowing larger coupling to the nanofiber than found using typical guided mode trapping techniques . |
the interaction of molecular species with surfaces is of critical importance in astrophysical environments .
much interstellar chemistry occurs on or in the icy layers which cover dust grains in molecular clouds , the birthplace of stars and planets @xcite .
these dust grains are believed to be composed of silicates and carbonaceous material @xcite .
the silicate is at least @xmath1 95 % amorphous , as determined from the breadth of the 9.7 @xmath2 m band @xcite , and although its composition has not been determined precisely , is well approximated by amorphous olivine ( mgfesio@xmath3 ) @xcite .
the silicates are believed to be fluffy in nature , with a large surface area upon which reactions can occur @xcite . at the low temperatures in molecular clouds ( t @xmath4 20 k ) atoms and molecules freeze out onto these silicate surfaces , forming the icy mantles .
amorphous solid water ( asw ) is by far the largest component of this ice , with abundances of @xmath1 1 @xmath5 10@xmath6 with respect to the total h column density @xcite , equivalent to coverages of up to 100 monolayers ( ml ) .
given that the extinction threshold for h@xmath0o mantles is a@xmath7 @xmath1 3.3 mag @xcite , it is reasonable to assume that between the dense pre - stellar cores ( where dust grains are completely coated by this icy mantle ) and the cloud edges ( where competition between ice formation and photodesorption of h@xmath0o yields a population of bare silicate grains @xcite ) , there must be a region where icy surfaces and bare silicates co - exist .
as interstellar regions evolve , the icy grains are further processed , either by gentle heating or cyclic desorption - deposition events @xcite , producing crystalline h@xmath0o .
this has been detected in various stellar and pre - planetary environments , including , for example , m giant stars @xcite , quaoar in the kuiper belt @xcite , trans neptune objects @xcite , comets @xcite , and in the outer disks around t tauri stars @xcite . as gas - grain modelling of interstellar environments becomes more sophisticated @xcite ,
key questions impeding the full implementation of surface chemistry include : what effect does the underlying grain surface have on the desorption characteristics of key molecules , and to what extent is this desorption affected as we move from the multilayer to sub - monolayer coverage regime ?
a third issue is how to realise the transition between the multilayer and monolayer regimes in a gas - grain model without overloading the model with many more layers of complexity .
co , o@xmath0 and co@xmath0 are all interstellar molecules that potentially could populate bare interstellar grains at sub - monolayer coverages . on ice , only one previous study focussed on sub - monolayer coverages of co , highlighting the spectroscopic rather than desorption characteristics of the porous asw : co system @xcite .
to date , no desorption studies of co , o@xmath0 or co@xmath0 have been made on a silicate surface , at either multilayer or sub - monolayer coverages .
co is the second most abundant interstellar molecule @xcite , and is known to form in the gas phase , then freeze - out on to h@xmath0o covered grains to form overlayers of pure co ice , the layer thickness critically depending on gas density rather than grain temperature @xcite .
consequently , extensive temperature programmed desorption ( tpd ) studies have been made of multilayer co coverages on various surfaces , including au @xcite , h@xmath0o @xcite , a meteorite sample @xcite and highly oriented pyrolytic graphite ( hopg ) @xcite .
it is widely accepted that both co@xmath0 and h@xmath0o form on dust grains in molecular clouds , and recent studies suggest that at least some of the co@xmath0 and h@xmath0o is formed concurrently @xcite .
since co is the key precursor to co@xmath0 formation , such mechanisms would require co freeze - out onto bare grain surfaces long before co ice , or even large quantities of h@xmath0o ice , are detected .
although tenuous , spectroscopic observational and experimental evidence exists for such a freeze - out process @xcite , but an investigation of the desorption behaviour of co at sub - monolayer coverages on ice and silicate surfaces is vital if we want to be certain that it can reside at an interstellar grain surface long enough to form more complex species . similarly , o@xmath0 is a potential precursor in h@xmath0o formation @xcite , and is also a species likely to form in the gas phase then freeze - out onto grain surfaces , rather than forming on the grain itself .
however , as a homonuclear diatomic o@xmath0 is infrared inactive , so its detection in interstellar ice , though occasionally claimed via a forbidden transition @xcite remains elusive .
it is not clear whether this is related to the weak transition probability , lack of a significant o@xmath0 population in the ice , or that the o@xmath0 has rapidly reacted ( upon adsorption ) to exclusively form h@xmath0o .
nevertheless , the multilayer desorption behaviour of o@xmath0 has been studied previously alongside co and n@xmath0 @xcite , as well as on au , porous asw @xcite and tio@xmath0 @xcite .
these studies show that the multilayer desorption characteristics of o@xmath0 are very similar to co , so it is interesting from both a chemical and astrophysical viewpoint to also investigate the desorption behaviour of o@xmath0 at sub - monolayer coverages on ice and silicate surfaces .
the desorption characteristics of co@xmath0 have not been extensively studied on any surface , despite it being one of the most abundant solid phase molecular species in the interstellar medium .
recently the desorption characteristics of multilayer co@xmath0 were reported from porous asw , and have been previously studied on hopg @xcite , porous asw and au @xcite .
however , given that the observational @xcite and experimental @xcite evidence shows that a fraction of co@xmath0 ice must form concurrently with the water ice layer , some co@xmath0 molecules must populate both the bare silicate grains and the ice layers at sub - monolayer coverages . here
we present an experimental study of the desorption of co , o@xmath0 and co@xmath0 from three different surfaces ; non - porous asw , crystalline ice , and amorphous olivine - type silicate . as these are , for the first time , all undertaken in the same experimental set - up
formolism @xcite we are able to investigate both the individual effect of each surface on the desorption characteristics , as well as determining whether the molecular composition or morphology of the surface is most relevant in determining the desorption behaviour of molecules . for each of the nine combinations of co , o@xmath0 and co@xmath0 on each surface ,
our study has encompassed both the multilayer and sub - monolayer regimes .
these data are modelled to determine a simple analytical expression which accurately calculates both the sub - monolayer and the multilayer desorption energies . by changing the model to incorporate interstellar , rather than experimental , heating rates ,
we are able to address the key questions above , namely : what effect does the underlying surface have on the desorption characteristics of the molecules adsorbed there , and how is this desorption modified in the sub - monolayer coverage regime ?
the experiments were conducted using the formolism set - up ( formation of molecules in the interstellar medium ) , described in detail elsewhere @xcite . briefly , the set - up consists of an ultra - high vacuum chamber ( base pressure @xmath1 10@xmath8 mbar ) , containing a silicate - coated copper sample surface , operating at temperatures between 18 and 400 k. the system is equipped with a quadrupole mass spectrometer ( qms ) , which is used for the tpd experiments . a sample of either o@xmath0 , co or co@xmath0 was deposited onto the surface at 18 k via the triply differentially pumped beam line ; a linear temperature ramp was then applied to the surface , and the qms used to measure the desorption of each species into the gas phase , as a function of temperature .
.description of experimental exposures . [ cols= " < ,
< , < " , ] @xmath9 full wetting behaviour is seen for molecules which fill all the lowest energy sites on the surface before forming a multilayer . intermediate
behaviour is when the molecules start to fill the surface , but form islands before it is full .
non - wetting behaviour is when islands are almost immediately formed by molecules on the surface .
( dot - dashed ) and co@xmath0 ( dashed ) from h@xmath0o@xmath10 ( black ) , h@xmath0o@xmath11 ( dark grey ) and sio@xmath12 ( light grey ) at a heating rate of 1 kcentury@xmath13 . for comparison ,
the black dotted line shows the modelled desorption characteristics of co on h@xmath0o@xmath10 if only multilayer desorption is included in the model.,scaledwidth=50.0% ] under astrophysical conditions , the heating rate is likely to be on the order of 1 kcentury@xmath13 ( hot core heating rate @xcite ) , many orders of magnitude slower than the laboratory value of 10 kmin@xmath13 employed here . by simulating the desorption profiles of the molecule - surface combinations here ,
it is possible to investigate whether surface type and coverage is likely to affect desorption characteristics on astrophysically relevant timescales . figure [ fgr : astro ] presents the results of such a simulation for all molecule - surface combinations , assuming a 2 ml coverage of each species on each grain ( i.e. one multilayer and one monolayer ) .
this seems reasonable as none of these species ( except possibly co ) form large pure ice overlayers in the ism . in figure
[ fgr : astro ] , the dot - dashed line shows the modelled desorption characteristics of 2 ml of co from h@xmath0o@xmath10 if only multilayer desorption is considered .
it is evident that without including the sub - monolayer component of the model , the desorption time is vastly underestimated ( by over a few @xmath5 10@xmath14 years or 10 15 k ) , particularly for co and co@xmath0 .
this is a comparable error to that introduced when gas - grain models treat multilayer desorption as first order , as suggested previously @xcite , and illustrates that sub - monolayer desorption characteristics are fundamental to describing the overall desorption profile of species on astrophysical timescales .
figure [ fgr : astro ] shows that for o@xmath0 there is the least difference between the time ( or temperature ) at which desorption would be completed in the multilayer only model versus the combined multilayer - monolayer model .
however , monolayers of both co and co@xmath0 are able to reside directly at the surface for thousands of years , or at equivalent temperatures 10 15 k higher , than those predicted by multilayer desorption alone , particularly on the amorphous silicate surface . in quiescent regions of the interstellar medium or at the edges of molecular clouds , where grains are warmer , this would suggest that small molecules , like co and co@xmath0 , could easily be present on the grain surfaces , potentially undergoing chemistry to generate larger chemical species , long before detectable ice layers really form .
such processes have been postulated previously @xcite , and it remains an interesting follow - on from this work to see which chemical species will form in astrochemical models when these sub - monolayer desorption data are considered .
what is also clear from figure [ fgr : astro ] is that , while the molecules desorb at different times ( and temperatures ) , there is a clear pattern to the order of desorption from the various substrates .
the onset of desorption from the amorphous surfaces , h@xmath0o@xmath10 and sio@xmath12 , occurs almost concurrently for all species , while desorption from h@xmath0o@xmath11 occurs later , at a higher temperature .
this delay in desorption from crystalline surfaces indicates that in regions where crystalline water is predicted to dominate , for example on processed ices inside disks around ysos @xcite , or even cometary surfaces @xcite , volatile species will not start desorbing from grain surfaces for a few thousand years longer ( or over temperature ranges of a few kelvin higher ) than anticipated from zeroth order desorption kinetics models .
such differences are subtle , but when one considers that these molecules will be highly mobile at such temperatures , the potential for complex chemistry to occur on the grain surfaces is increased .
furthermore , it is exactly these inner disk regions , where crystalline water ice is predicted to be found , from which hot core and hot corino gases are populated by desorbing molecules(e.g . * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
similarly , the simulation results in figure [ fgr : astro ] show the residence time of molecules on a h@xmath0o@xmath11 surface is significantly higher than on the amorphous surfaces . starting from a 2 ml coverage ,
the time taken for 50 % of the o@xmath0 on sio@xmath12 or h@xmath0o@xmath10 to desorb is 8000 years , while on h@xmath0o@xmath11 this value is 8700 years , an increase of 8.75 % .
similarly , the difference between desorption of 50 % of the molecules from h@xmath0o@xmath10 and from h@xmath0o@xmath11 is 8.82 % for co and 6.97 % for co@xmath0 .
a key aim of this work was to address what effect the underlying dust grain surface could have on the desorption characteristics of molecules adsorbed there , and how this desorption is modified in the sub - monolayer coverage regime compared to the multilayer regime .
this work used temperature programmed desorption to measure the desorption characteristics of o@xmath0 , co and co@xmath0 from h@xmath0o@xmath10 , h@xmath0o@xmath11 and sio@xmath12 over sub - monolayer to multilayer coverages .
the experimental data were modelled using the polanyi - wigner equation , combining different approaches to reproduce the sub - monolayer and multilayer tpd traces , in particular by varying e@xmath15 as a function of coverage in the sub - monolayer regime .
the desorption can be categorised as full wetting , intermediate or non - wetting behaviour , and the switching point is well described by n@xmath16 , an empirical measure of the coverage at which pure monolayer desorption stops . on both a laboratory and an astrophysically relevant timescale , the desorption characteristics of molecules from the amorphous substrates h@xmath0o@xmath10 and sio@xmath12
were found to be very similar , while on the crystalline surface h@xmath0o@xmath11 , molecules desorbed at higher temperatures and on a longer timescale .
it would be relatively trivial to implement the data presented here into astrochemical models ; it would require only a subroutine to describe the sub - monolayer , coverage dependent , e@xmath15 , which can be calculated using the parameters presented in this paper , and the coverage n@xmath16 , at which the true monolayer desorption ends and beyond which a fixed e@xmath15 , and zeroth order desorption can be implemented .
previously , @xcite concluded that surface type is relatively unimportant , and rather it is the heating rate and grain size that dominate desorption kinetics ; we do not disagree that both these factors are important , but by not considering the sub - monolayer case , this previous work overlooks certain intricacies .
firstly , the results presented here suggest that the surface type is the dominant factor controlling desorption under sub - monolayer coverage conditions .
this definition of surface type must include both the surface material and the degree of crystallinity , and not simply the material alone .
from the data presented here , it is clear that the desorption characteristics of molecular species from h@xmath0o@xmath10 and sio@xmath12 , two amorphous surfaces , are much more similar than those from h@xmath0o@xmath10 and h@xmath0o@xmath11 , which are composed of the same underlying material .
secondly , while the size of the grain is critical to desorption characteristics , it is both the surface area and the surface - adsorbate interaction which must be considered . without including information on whether the surface - adsorbate interaction is wetting or non - wetting , it is not possible to know the coverage at which behaviour switches from sub - monolayer ( to intermediate ) to the multilayer regime , i.e. the value of n@xmath16 . as it is notoriously difficult to calculate physisorption surface - adsorbate energies theoretically , this is necessarily an empirically determined value , highlighting the importance of experiments such as those presented here .
j.a.n . gratefully acknowledges the financial support of the leverhulme trust , the scottish international education trust , the university of strathclyde and the scottish universities physics alliance , without whom this work would not have been carried out .
the research leading to these results has received funding from the european community s seventh framework programme fp7/2007 - 2013 under grant agreement no .
we acknowledge the support of the national pcmi programme founded by the cnrs , the conseil regional dile de france through sesame programmes ( contract i-07 - 597r ) , the conseil gnral du val doise and the agence nationale de recherche ( contract anr 07-blan-0129 ) . the authors would like to thank h. mokrane for her assistance in the laboratory , as well as louis dhendecourt and zahia djouadi for preparing the silicate surface used in these experiments .
we are also immensely grateful to the anonymous referee , whose insightful comments have greatly improved this manuscript
. 99 accolla , m. , et al . 2011 , physical chemistry chemical physics ( incorporating faraday transactions ) , 13 , 8037 acharyya , k. , fuchs , g. w. , fraser , h. j. , van dishoeck , e. f. , & linnartz , h. 2007 , , 466 , 1005 acharyya , k. , hassel , g. e. , & herbst , e. 2011 , , 732 , 73 amiaud , l. , fillion , j. h. , baouche , s. , dulieu , f. , momeni , a. , & lemaire , j. l. 2006 , , 124 , 094702 andersson , p. u. , nagard , m. b. , witt , g. , & pettersson , j. b. c. 2004 , j. phys . chem . a , 108 , 4627 bisschop , s. e. , fraser , h. j. , berg , k. i. , van dishoeck , e. f. , & schlemmer , s. 2006 , , 449 , 1297 bojan , m. j. , & steele , w. a. 1987 , langmuir , 3 , 1123 brown , w. a. , & bolina , a. s. 2007 , , 374 , 1006 burke , d. j. , & brown , w. a. 2010 , , 12 , 5947 cazaux , s. , tielens , a. g. g. m. , ceccarelli , c. , castets , a. , wakelam , v. , caux , e. , parise , b. , & teyssier , d. 2003 , , 593 , l51 collings , m. p. , dever , j. w. , fraser , h. j. , mccoustra , m. r. s. , & williams , d. a. 2003a , , 583 , 1058 collings , m. p. , dever , j. w. , fraser , h. j. , & mccoustra , m. r. s. 2003b , , 285 , 633 collings , m. p. , anderson , m. a. , chen , r. , dever , j. w. , viti , s. , williams , d. a. , & mccoustra , m. r. s. 2004 , , 354 , 1133 collings , m. p. , dever , j. w. , & mccoustra , m. r. s. 2005 , chem .
, 415 , 40 cuppen , h. m. , & garrod , r. t. 2011 , , 529 , a151 dohnalek , z. , ciolli , r. l. , kimmel , g. a. , stevenson , k. p. , smith , r. s. , & kay , b. d. 1999 , j. chem .
phys . , 110 , 5489 dohnalek , z. , kimmel , g. a. , joyce , s. a. , ayotte , p. , smith , r. s. , & kay , b. d. 2001 , j. phys . chem .
b , 105 , 3747 dohnalek , z. , kim , j. , bondarchuk , o. , white , j. m. , & kay , b. d. 2006 , j. phys .
b , 110 , 6229 draine , b. t. 2003 , , 41 , 241 elsila , j. , allamandola , l. j. , & sandford , s. a. 1997 , , 479 , 818 ertl , g. , neumann , m. , & streit , k. m. 1977 , surf .
, 64 , 393 favre , c. , despois , d. , brouillet , n. , baudry , a. , combes , f. , gulin , m. , wootten , a. , & wlodarczak , g. 2011 , , 532 , a32 fayolle , e. c. , berg , k. i. , cuppen , h. m. , visser , r. , & linnartz , h. 2011 , , 529 , a74 fraser , h. j. , collings , m. p. , mccoustra , m. r. s. , & williams , d. a. 2001 , , 327 , 1165 fraser , h. j. , mccoustra , m. r. s. , & williams , d. a. 2002 , astronomy and geophysics , 43 , 020000 fraser , h. j. , collings , m. p. , dever , j. w. , & mccoustra , m. r. s. 2004 , , 353 , 59 fraser , h. j. , bisschop , s. e. , pontoppidan , k. m. , tielens , a. g. g. m. , & van dishoeck , e. f. 2005 , , 356 , 1283 frenklach , m. , huang , d. , thomas , r. e. , rudder , r. a. , & markunas , r. j. 1993 , appl .
lett . , 63 , 3090 fuchs , g. w. , et al .
2006 , faraday discussions , 133 , 331 glvez , o. , ortega , i. k. , mat , b. , moreno , m. a. , martn - llorente , b.,herrero , v. j. , escribano , r. , & gutirrez , p. j. 2007 , , 472 , 691 goumans , t. p. m. , uppal , m. a. , & brown , w. a. 2008 , , 384 , 1158 green , s. d. , bolina , a. s. , chen , r. , collings , m. p. , brown , w. a. , & mccoustra , m. r. s. 2009 , , 398 , 357 ioppolo , s. , cuppen , h. m.,romanzin , c. , van dishoeck , e. f. , & linnartz , h. 2008 , , 686 , 1474 ioppolo , s. , van boheemen , y. , cuppen , h. m. , van dishoeck , e. f. , & linnartz , h. 2011 , , 413 , 2281 jewitt , d. c. , & luu , j. 2004 , , 432 , 731 lakhlifi , a. , & killingbeck , j. p. 2010
, 604 , 38 lattelais , m. , et al .
2011 , , 532 , a12 lemaire , j. l. , vidali , g. , baouche , s. , chehrouri , m. , chaabouni , h. , & mokrane , h. 2010 , , 725 , l156 li , a. , & draine , b. t. 2002 , , 564 , 803 lisse , c. m. , et al .
2006 , science , 313 , 635 matar , e. , bergeron , h. , dulieu , f. , chaabouni , h. , accolla , m. , & lemaire , j. l. 2010 , , 133 , 104507 mathis , j. s. 1998 , , 497 , 824 mautner , m. n. , abdelsayed , v. , el - shall , m. s. , thrower , j. d. , green , s. d. , collings , m. p. , & mccoustra , m. r. s. 2006 , faraday discussions , 133 , 103 merlin , f. , guilbert , a. , dumas , c. , barucci , m. a. , de bergh , c. , & vernazza , p. 2007
, , 466 , 1185 noble , j. a. , dulieu , f. , congiu , e. , & fraser , h. j. 2011 , , 735 , 121 oba , y. , miyauchi , n. , hidaka , h. , chigai , t. , watanabe , n. , & kouchi , a. 2009 , , 701 , 464 berg , k. i. , van broekhuizen , f. , fraser , h. j. , bisschop , s. e. , van dishoeck , e. f. , & schlemmer , s. 2005 , , 621 , l33 omont , a. , forveille , t. , moseley , s. h. , glaccum , w. j. , harvey , p. m. , likkel , l. , loewenstein , r. f. , & lisse , c. m. 1990 , , 355 , l27 pontoppidan , k. m. , et al .
2003 , , 408 , 98 pontoppidan , k. m. 2006 , , 453 , l47 roush , t. l. 2001 , , 106 , 33315 raval , r. , haq , s. , harrison , m. a. , blyholder , g. , & king , d. a. 1990 , chem .
lett . , 167 , 391 sandford , s. a. , & allamandola , l. j. 1990 , , 355 , 357 schegerer , a. a. , & wolf , s. 2010 , , 517 , a87 semenov , d. , et al .
2010 , , 522 , a42 smith , r. g. , sellgren , k. , & brooke , t. y. 1993 , , 263 , 749 sofia , u. j. , & meyer , d. m. 2001 , , 554 , l221 stirniman , m. j. , huang , c. , smith , r. s. , joyce , s. a. , & kay , b. d. 1996 , , 105 , 1295 terlain , a. , & larher , y. 1983 , surf .
, 125 , 304 tielens , a. g. g. m. 2005 , the physics and chemistry of the interstellar medium , cambridge university press ulbricht , h. , zacharia , r. , cindir , n. , & hertel , t. 2006 , carbon , 44 , 2931 van der tak , f. f. s. , van dishoeck , e. f. , & caselli , p. 2000 , , 361 , 327 visser , r. , van dishoeck , e. f. , doty , s. d. , & dullemond , c. p. 2009
, , 495 , 881 viti , s. , & williams , d. a. 1999 , , 305 , 755 wakelam , v. , et al .
2010 , , 156 , 13 wang , s .- x .
, yang , y. , sun , b. , li , r .- w . , liu , s .- j .
, & zhang , p. 2009
, , 80 , 115434 whittet , d. c. b. , bode , m. f. , longmore , a. j. , adamson , a. j. , mcfadzean , a. d. , aitken , d. k. , & roche , p. f. 1988 , , 233 , 321 williams , d. , & herbst , e. 2002 , surface science , 500 , 823 | the desorption characteristics of molecules on interstellar dust grains are important for modelling the behaviour of molecules in icy mantles and , critically , in describing the solid - gas interface . in this study ,
a series of laboratory experiments exploring the desorption of three small molecules from three astrophysically relevant surfaces are presented .
the desorption of co , o@xmath0 and co@xmath0 at both sub - monolayer and multilayer coverages was investigated from non - porous water , crystalline water and silicate surfaces .
experimental data was modelled using the polanyi - wigner equation to produce a mathematical description of the desorption of each molecular species from each type of surface , uniquely describing both the monolayer and multilayer desorption in a single combined model .
the implications of desorption behaviour over astrophysically relevant timescales are discussed .
[ firstpage ] astrochemistry ism : molecules methods : laboratory . |
the extragalactic background light ( ebl ) is an inherently average or `` global '' property of the present - day universe .
it is a direction - averaged , time - integrated record of the emission and absorption of photons since recombination ( @xmath0 ) . as such , the ebl is closely related to the global history of star formation and , potentially , that of other processes , such as the accretion of matter onto black holes .
the global history of star formation in turn is closely related to the global histories of gas consumption and metal production in galaxies .
the purpose of this article is to explore some of these relations . in this approach ,
one largely ignores the individuality of galaxies and their internal complexity .
the primary focus , instead , is on the average properties of galaxies .
the global approach may be contrasted with others in which the individuality and diversity of galaxies are regarded as paramount . in the latter ,
one attempts to predict , or at least to describe , the full variety of evolutionary paths followed by galaxies of different types , masses , and other properties .
this is necessary to interpret luminosity functions and number counts .
the ebl and other global quantities are then obtained by integrating over the these functions .
this approach is more challenging and perhaps more fundamental than the global approach in that it ties the interpretation of the ebl to a fairly complete understanding of the formation and evolution of galaxies .
the global approach forfeits some of this insight but is much simpler . that is both its main limitation and its main advantage .
one of the grand themes to emerge in the last few years is the idea that we may be able to determine the global histories of star formation , gas consumption , and metal production in galaxies from high redshifts to the present .
this idea is illustrated in figure 1 , where we sketch the evolution of the contents of a large , comoving box .
we can conveniently quantify the masses of the different constituents of the box by the corresponding mean comoving densities normalized to the present critical density .
we are especially interested in the comoving densities of stars , gas ( both inside and outside galaxies , i.e. , ism and igm ) , metals , dust , and black holes : @xmath1 , @xmath2 , @xmath3 , @xmath4 , @xmath5 , and @xmath6 , respectively .
we are also interested in the cosmic emissivity @xmath7 , the power radiated per unit comoving volume per unit rest - frame frequency @xmath8 , and the ebl intensity @xmath9 , the power received per unit solid angle of sky per unit area of detector per unit observed frequency @xmath8 .
after recombination , our comoving box is filled with neutral , metal - free gas with nearly uniform density .
perturbations in this intergalactic medium ( igm ) eventually condense , probably by gravitational clumping and inflow , into protogalaxies .
stars then form in the resulting interstellar medium ( ism ) .
they produce metals and may drive outflows of gas from galaxies . in this way , both the ism and igm may be enriched with metals .
some of the metals remain in the gas phase ; others condense into solid dust grains .
black holes form in the nuclei of some , perhaps even all , galaxies and , when fueled , can power active galactic nuclei ( agn ) .
some of the radiation emitted by stars and agn propagates freely , while the rest is absorbed and then emitted at longer wavelengths by dust .
thus , the radiation we detect from galaxies tells us primarily about their star , agn , and dust contents .
the spectra of high - redshift quasars contain signatures of the absorption and scattering of radiation by the intervening ism and igm ( absorption lines , reddening , etc ) .
such observations tell us primarily about the composition and comoving densities of the ism and igm .
exactly how all this happens is not yet known , of course .
it should be clear from figure 1 and the commentary above , however , that the constituents of our comoving box , including the radiation that propagates through it , are very much interrelated .
in fact , the corresponding comoving densities must obey a series of coupled conservation - type equations .
clearly , @xmath1 , @xmath2 , and @xmath3 must add up to @xmath10 , a constant .
similarly , @xmath11 , @xmath12 , and @xmath13 , the comoving densities of metals in stars , the ism , and the igm , must add up to @xmath4 .
moreover , @xmath4 remains a fixed fraction of @xmath1 on the assumption that the global yield is constant and that delayed recycling is negligible ( a good approximation in the present context ) .
the cosmic emissivity @xmath7 depends on the star formation rate @xmath14 , black hole fueling rate @xmath15 , and the amount of reprocessing by dust and hence @xmath5 .
the ebl intensity is related to the cosmic emissivity by a radiative transfer equation ; for wavelengths at which the igm is transparent , this gives @xmath16 where @xmath17 is the present age of the universe .
it is likely that agn make relatively small contributions to @xmath7 and @xmath9 and can be neglected in a first approximation .
to obtain a set of equations for the evolution of the average stellar and interstellar contents of galaxies , we begin with the familiar equations of galactic chemical evolution . in this approach
, the igm is regarded as a reservoir with which a galaxy can exchange material by inflow or outflow . following standard practice
, we denote the mass of stars and ism ( gas and dust ) in a galaxy by @xmath18 and @xmath19 ( @xmath20 ) , the interstellar metallicity by @xmath21 ( @xmath22 ) , the inflow or outflow rate by @xmath23 , the metallicity of inflowing or outflowing material by @xmath24 , and the nucleosynthetic yield of heavy elements , averaged over the stellar initial mass function , by @xmath25 . in the approximation of instantaneous recycling and @xmath26 ,
these quantities are related by @xmath27 @xmath28 equation ( 3 ) gives the total rate of change of the mass of metals in the ism ; on the right - hand side , the first term is the gain by stellar production and ejection ( proportional to the rate of star formation since most of the metals are synthesized in short - lived stars ) , the second term is the loss by removal of the ism ( including its metals ) by forming stars , while the third term is the gain or loss by exchange with the igm ( see tinsley 1980 or pagel 1997 for complete explanations and derivations ) .
if we now sum equations ( 2 ) and ( 3 ) over all galaxies in a large comoving volume and then divide by that volume and the present critical density , we obtain what are sometimes called the equations of cosmic chemical evolution : @xmath29 @xmath30 here , the effective yield @xmath31 depends on the nucleosynthetic yield @xmath25 and the spread in the interstellar metallicities of galaxies weighted by their star formation rates . in the hypothetical case that all galaxies have the same interstellar metallicity , @xmath31 is the same as @xmath25 . in general ,
the two yields will differ , but this is not a serious problem because one can adjust @xmath31 so that the mean metallicity in the models agrees , at some particular time , such as the present , with the observed value .
this is the same approach taken in nearly all models of galactic chemical evolution . indeed ,
if one were modeling the evolution of a chemically inhomogeneous galaxy , the same issues would arise in the derivation and solutions of equations ( 2 ) and ( 3 ) ; the parameter @xmath25 would then have to be regarded as an effective yield , not equal to the nucleosynthetic yield .
the average interstellar properties of galaxies can be determined from the statistics of quasar absorption - line systems as follows .
let @xmath32 be the column density distribution of particles of any type @xmath33 that absorb or scatter light .
these might , for example , be hydrogen atoms ( @xmath34 hi ) , metal ions ( @xmath35 ) , or dust grains ( @xmath36 ) . by definition
, @xmath37 is the mean number of absorption - line systems with column densities of @xmath33 between @xmath38 and @xmath39 and redshifts between @xmath40 and @xmath41 along the lines of sight to randomly selected background quasars .
these lines of sight are very narrow ( the projected size of the continuum emitting regions of quasars , less than a light year across ) and pierce the absorption - line systems at random angles and impact parameters .
one can show that the mean comoving density of @xmath33 is given by @xmath42 where @xmath43 is the mass of a single particle ( atom , ion , or grain ) .
equation ( 6 ) plays a central role in this subject .
it enables us to estimate the mean comoving densities of many quantities of interest without knowing anything about the structure of the absorption - line systems .
in particular , we do not need to know their sizes or shapes , whether they are smooth or clumpy , and so forth .
a corollary of equation ( 6 ) is that the global interstellar metallicity , @xmath44 , is given simply by an average over the metallicities of individual absorption - line systems weighted by their total column densities .
the absorption - line systems of most interest in the present context are the damped ly@xmath45 ( dla ) systems .
it is widely believed that they constitute the ism of galaxies and protogalaxies and hence are the principal sites of star formation in the universe ( wolfe et al .
there are excellent reasons to adopt this as a working hypothesis .
first , the dla systems have , by definition , @xmath46 @xmath47 , and this , at least at low redshifts , is near or slightly below the threshold for star formation ( kennicutt 1989 ) .
second , the dla systems contain at least 80% of the hi in the universe and appear to be mostly neutral .
the other absorption - line systems , those with @xmath48
@xmath47 , probably contain more gas in total than the dla systems , but this must be diffuse and mostly ionized . in the following , we regard non - dla systems as belonging to the igm , even though some of them might actually be associated with the outer , tenuous parts of galaxies .
this distinction between the mostly - neutral ism , where stars form , and the mostly - ionized igm , where they do not is certainly valid at the present epoch ( zwaan et al .
thus , the dla systems are often referred to as dla galaxies .
the precise nature of these objects whether they are large or small , disk or spheroid remains to be determined , probably by direct imaging ( le brun et al .
. however , as we have already emphasized , the global properties derived from equation ( 6 ) are not affected by these ambiguities .
the sample of known dla galaxies now includes about 100 objects .
most of these have redshifts in the range @xmath49 ; only a few are known at @xmath50 ( which require observations from space ) and at @xmath51 ( which require bright quasars at higher redshifts ) .
we now have measurements of the column densities of neutral atomic hydrogen ( hi ) in all these systems and of the metals and dust in many of them ( for hi , see lanzetta , wolfe , & turnshek 1995 ; wolfe et al .
1995 ; storrie - lombardi , mcmahon , & irwin 1996 ; rao & turnshek 2000 ; storrie - lombardi & wolfe 2000 ; for metals , see pettini et al .
1994 , 1997a , 1999 ; lu et al .
1996 ; boiss et al . 1998 ; prochaska & wolfe 2000 ; for dust , see pei , fall , & bechtold 1991 ; kulkarni , fall , & truran 1997 ; pettini et al . 1997b ; welty et al . 1997 ; vladilo 1998 ) .
it is believed that the dla galaxies have low fractions of ionized hydrogen ( hii ) because the hi layers are opaque to ionizing radiation . in the few dla galaxies that have been searched
, the abundance of molecular hydrogen ( h@xmath52 ) lies below the value in the milky way ( levshakov et al . 1992 ; ge & bechtold 1997 ) .
one factor that potentially biases all these observations is the presence of dust in dla galaxies .
quasars behind dusty dla galaxies will be obscured , thus reducing the chances that they will be included in optically selected samples ( ostriker & heisler 1984 ; fall & pei 1993 ; boiss et al .
gravitational lensing has the opposite effect , but this appears to be negligible in the existing samples ( le brun et al . 1997 ; perna , loeb , & bartelmann 1997 ; smette , claeskens , & surdej 1997 ) .
finally , we emphasize that estimates of the comoving densities of hi , metals , and dust are dominated by relatively few systems those with the highest column densities . as a result
, they are less certain than is often recognized .
global evolution models of the type described above have been constructed by lanzetta et al .
( 1995 ) , pei & fall ( 1995 , hereafter pf ) , malaney & chaboyer ( 1996 ) , and pei , fall , & hauser ( 1999 , hereafter pfh ) . some related issues are discussed by songaila , cowie , & lilly ( 1990 ) , timmes , lauroesch , & truran ( 1995 ) , fall , charlot , & pei ( 1996 ) , calzetti & heckman ( 1999 ) , and harwit ( 1999 ) . in the following
, we present highlights from the pf and pfh models . at the time the pf models were published ,
most of the successful surveys for emission from galaxies were confined to @xmath53 , and nothing was known about the global history of star formation , @xmath54 . indeed , one of the goals of the pf models was to predict @xmath54 from absorption - line observations of dla galaxies , particularly @xmath55 and @xmath56 . in 199495 , the comoving density of hi was found to be a smoothly increasing function of redshift from @xmath57 to @xmath58 , while the mean metallicity was known at only two redshifts : @xmath59 at @xmath57 and @xmath60 at @xmath61 . these observations
the first pertaining to gas consumption , the second to metal production strongly suggested that most stars in the universe formed after @xmath61 .
the motivation behind the global evolution models was to provide a framework to interpret such observations more quantitatively .
the pf models are based on the following simplifying assumptions : 1 .
hi is a tracer of the total interstellar content of galaxies , both atomic and molecular ( @xmath62 ) .
the dust content of galaxies is proportional to the metal content , both locally and globally ( @xmath63 and @xmath64 ) .
the interaction between galaxies and the igm takes one of three idealized forms : closed box ( @xmath65 ) , metal - free inflow ( @xmath66 , @xmath67 ) , or metal - enriched outflow ( @xmath68 , @xmath69 ) .
assumptions 1 and 2 are consistent with , and indeed motivated by , the observations summarized in the previous section .
assumption 2 couples the internal absorption and the obscuration of background quasars to the chemical evolution of galaxies ; this enables self - consistent corrections for biases in the the census of dla galaxies and hence in the estimates of @xmath55 .
assumption 3 is the cosmological analog of the standard treatment of inflow and outflow in models of galactic chemical evolution , beginning with the work of larson ( 1972 ) and hartwick ( 1976 ) .
while one might question the precise validity of these assumptions , they should be realistic enough to reveal a first glimpse of the global evolution of galaxies .
the pf models were designed to reproduce ( as input ) the observed evolution in @xmath55 . as a bonus , they also matched ( as output ) the available estimates of @xmath56 without any fine tuning of parameters .
figure 2 shows the predicted evolution of the comoving rate of metal production @xmath70 in the standard pf models .
( the results displayed in this and subsequent figures are for @xmath71 kms@xmath72mpc@xmath72 , @xmath73 , and @xmath74 . )
this is just another way of expressing the star formation rate : @xmath75 .
figure 2 also shows subsequent emission - based estimates of @xmath70 from several surveys , including the canada - france redshift survey and the hubble deep field ( gallego et al .
1995 ; lilly et al .
1996 ; madau et al . 1996 , 1998 ; connolly et al .
these were derived from rest - frame h@xmath45 and uv emissivities and the approximate proportionality between uv emission and metal production in stellar populations . evidently , the predicted and observed rates are in broad agreement ( within factors of about two ) .
this is remarkable because the pf models were constructed only with absorption - line observations in mind , before the emission - based estimates of @xmath76 became available .
the pf models , when combined with stellar population synthesis models , also predicted a far - ir / sub - mm background ( fall et al . 1996 ) that was consistent with early empirical limits and estimates from the dirbe and firas experiments on _ cobe _ ( hauser 1996 ; puget et al .
the pfh models were designed to improve upon the pf models by addressing several subsequent developments in this rapidly advancing field .
first , it was realized that estimates of the comoving star formation and metal production rates from rest - frame uv emissivities were potentially biased by absorption by dust .
the proposed corrections to the estimates of @xmath76 in figure 2 ranged from nearly nothing to more than an order of magnitude .
the large uncertainty in these corrections stemmed from the difficulty of determining the dust content and wavelength dependence of the effective absorption in high - redshift galaxies .
second , it was found from a large - area , ground - based survey that even the uncorrected uv emissivity at @xmath77 was probably about twice as high as had initially been estimated from the small - area hubble deep field ( steidel et al .
third , the limits on , and measurements of , the ebl intensity have improved , especially in the far - ir / sub - mm spectral region ( fixsen et al .
1998 ; hauser et al . 1998 ; schelgel , finkbeiner , & davis 1998 ) .
these are shown in figure 3 .
it now appears that there is at least as much energy in the long - wavelength hump of the ebl spectrum ( @xmath78 @xmath79 m ) as in the short - wavelength hump ( @xmath80 @xmath79 m ) .
this , of course , provides a valuable constraint on the amount of uv / optical emission that has been absorbed and reradiated by dust . in the pfh models ,
the observed emissivity at rest - frame uv wavelengths @xmath81 is treated as an input function , along with @xmath55 .
the emissivity at other wavelengths is computed from stellar population synthesis models and a template spectrum of dust emission derived from the local emissivity at several ir wavelengths .
the absorption per mass of dust , assumed proportional to the mass of interstellar metals , is then adjusted to satisfy the empirical constraints on the ebl intensity .
this procedure closes the equations of cosmic chemical evolution without invoking a specific relation between the star formation rate and the inflow or outflow rate , as in the pf models .
indeed , in the pfh models , both @xmath54 and @xmath82 are independent output solutions of the equations .
the net effect of this procedure is to determine the correction for absorption by dust between the observed and true uv emissivities , i.e. , between the observed and true star formation rates , as a function of redshift , in a way that is consistent with the production of metals and dust and the consumption or replenishment of gas by a combination of star formation and inflow or outflow .
the results are also guaranteed , by construction , to be consistent with a wide variety of observations , including the uv emissivity and comoving density of hi as functions of redshift and the relative amounts of energy in the long- and short - wavelength humps of the ebl spectrum . as figure 3 shows , the ebl intensity in the models
does indeed match the observations .
( in figures 35 , the short - dash , solid , and long - dash lines indicate the range of solutions permitted by uncertainties in @xmath9 at sub - mm wavelengths . )
figure 4 shows the evolution of the comoving rate of star formation @xmath54 before and after the corrections for absorption by dust . evidently , these corrections are factors of 24 at most redshifts .
the star formation rates in the pfh models have the same qualitative behavior as those in the pf models , including the rapid rise from @xmath57 to @xmath83 12 , but the decline from @xmath83 12 to @xmath84 is now shallower or possibly non - existent .
figure 5 shows the predicted evolution of the mean interstellar metallicity and the comoving density of interstellar metals , @xmath56 and @xmath85 .
the former increases monotonically with decreasing redshift , while the latter first increases and then decreases as the interstellar medium is consumed by star formation .
the effective yield in the models has been adjusted to give @xmath86 at @xmath87 , the approximate mean interstellar metallicity in present - day galaxies ( averaged over internal gradients and the luminosity function of galaxies ) . at higher redshifts ,
the mean metallicities in the models are marginally consistent with those in the dla galaxies , although the observed enrichment appears to be slower than predicted .
this may in part be a consequence of the bias against selecting dusty and hence metal - rich dla galaxies , an effect that tends to make the observed @xmath56 relation artificially flat .
the main conclusion to be drawn from this article is that we now have the apparatus needed to interpret simultaneously a wide variety of observations pertaining to the average properties of galaxies , including the comoving densities of stars and interstellar gas , metals , and dust .
an attraction of the global evolution models is that they relate observations that would otherwise appear disparate , namely , those of the light galaxies emit and those of the light they absorb ( from background quasars ) .
moreover , this approach is perhaps the simplest and most natural one in which to interpret the ebl .
simplicity is achieved by focusing on the average properties of galaxies and , for the most part , ignoring their individuality and internal complexity .
the global evolution models have now been developed to the point that the accuracy of the results is limited more by observational than by theoretical uncertainties .
the pfh models incorporated the best data available in 1998 .
since then , there have been debates about the evolution of the comoving density of hi and the mean metallicity in dla galaxies .
the evolution of the uv emissivity , and hence the comoving rate of star formation , is probably equally uncertain , even before the corrections for absorption by dust .
the problem is that all these observational relations are based on relatively small samples and are affected by a variety of selection biases ( e.g. , dust for absorption , surface brightness for emission ) .
thus , it should be a high priority in the next few years to improve the observations , by increasing the sizes of the samples ( e.g. , by factors of 10 or more ) and by reducing the selection biases ( e.g. , by using radio rather than optically selected quasars ) .
it should then be possible to repeat much of the analysis described here with greater confidence in the accuracy of the results .
boiss , p. , le brun , v. , bergeron , j. , & deharveng , j .- m . 1998 , a&a , 333 , 841 calzetti , d. , & heckman , t. m. 1999 , apj , 519 , 27 connolly , a. j. , szalay , a. s. , dickinson , m. , subbarao , m. u. , & brunner , r. j. 1997 , apj , 486 , l11 fall , s. m. , charlot , s. , & pei , y. c. 1996 , apj , 464 , l43 fall , s. m. , & pei , y. c. 1993 , apj , 402 , 479 fixsen , d. j. , dwek , e. , mather , j. c. , bennett , c. l. , & shafer , r. 1998 , apj , 508 , 123 gallego , j. , zamorano , j. , aragn - salamanca , a. , & rego , m. 1995 , apj , 455 , l1 ge , j. , & bechtold , j. 1997 , apj , 477 , l73 hartwick , f. d. a. 1976 , apj , 209 , 418 harwit , m. 1999 , apj , 510 , l83 hauser , m. g. 1996 , in unveiling the cosmic infrared background , ed .
e. dwek ( woodbury : am .
inst . phys . ) , 11 hauser , m. g. , arendt , r. g. , kelsall , t. , dwek , e. , odegard , n. , weiland , j. l. , freudenreich , h. t. , reach , w. t. , silverberg , r. f. , moseley , s. h. , pei , y. c. , lubin , p. , mather , j. c. , shafer , r. a. , smoot , g. f. , weiss , r. , wilkinson , d. t. , & wright , e. l. 1998 , apj , 508 , 25 kennicutt , r. c. 1989 , apj , 344 , 685 kulkarni , v. p. , fall , s. m. , & truran , j. w. 1997 , apj , 484 , l7 lanzetta , k. m. , wolfe , a. m. , & turnshek , d. a. 1995 , apj , 440 , 435 larson , r. b. 1972 , nature phys .
, 236 , 7 le brun , v. , bergeron , j. , boiss , p. , & deharveng , j. m. 1997 , a&a , 321 , 733 levshakov , s. a. , chaffee , f. h. , foltz , c. b. , & black , j. h. 1992 , a&a , 262 , 385 lilly , s. j. , le fvre , o. , hammer , f. , & crampton , d. 1996 , apj , 460 , l1 lu , l. , sargent , w. l. w. , barlow , t. a. , churchill , c. w. , & vogt , s. s. 1996 , apjs , 107 , 475 madau , p. , ferguson , h. c. , dickinson , m. e. , giavalisco , m. , steidel , c. c. , & fruchter , a. 1996 , mnras , 283 , 1388 madau , p. , pozzetti , l. , & dickinson , m. 1998 , apj , 498 , 106 malaney , r. a. , & chaboyer , b. 1996 , apj , 462 , 57 ostriker , j. p. , & heisler , j. 1984 , apj , 278 , 1 pagel , b. e. j. 1997 , nucleosynthesis and chemical evolution of galaxies ( cambridge : cambridge university press ) pei , y. c. , & fall , s. m. 1995 , apj , 454 , 69 ( pf ) pei , y. c. , fall , s. m. , & bechtold , j. 1991 , apj , 378 , 6 pei , y. c. , fall , s. m. , & hauser , m. g. 1999 , apj , 522 , 604 ( pfh ) perna , r. , loeb , a. , & bartelmann , m. 1997 , apj , 488 , 550 pettini , m. , ellison , s. l. , steidel , c. c. , & bowen , d. v. 1999 , apj , 510 , 576 pettini , m. , king , d. l. , smith , l. j. , & hunstead , r. w. 1997b , apj , 478 , 536 pettini , m. , smith , l. j. , hunstead , r. w. , & king , d. l. 1994 , apj , 426 , 79 pettini , m. , smith , l. j. , king , d. l. , & hunstead , r. w. 1997a , apj , 486 , 665 prochaska , j. x. , & wolfe , a. m. 2000 , apj , 533 , l5 puget , j .- l .
, abergel , a. , bernard , j .-
p . , boulanger , f. , burton , w. b. , dsert , f .- x .
, & hartmann , d. 1996 , a&a , 308 , l5 rao , s. m. , & turnshek , d. a. 2000 , apjs , 130 , 1 schelgel , d. j. , finkbeiner , d. p. , & davis , m. 1998 , apj , 500 , 525 smette , a. , claeskens , j .- f .
, & surdej , j. 1997 , new astron . , 2 , 53 songaila , a. , cowie , l. l. , lilly , s. j. 1990 , apj , 348 , 371 steidel , c. c. , adelberger , k. l. , giavalisco , m. , dickinson , m. , & pettini , m. 1999 , apj , 519 , 1 storrie - lombardi , l. j. , mcmahon , r. g. , & irwin , m. j. 1996 , mnras , 283 , l79 storrie - lombardi , l. j. , & wolfe , a. m. 2000 , apj , in press ( astro - ph/0006044 ) timmes , f. x. , lauroesch , j. t. , & truran , j. w. 1995 , apj , 451 , 468 tinsley , b. m. 1980 , fund .
cosmic phys . , 5 , 287 vladilo , g. 1998 , apj , 493 , 583 welty , d. e. , lauroesch , j. t. , blades , j. c. , hobbs , l. m. , & york , d. g. 1997 , apj , 489 , 672 wolfe , a. m. , lanzetta , k. m. , foltz , c. b. , & chaffee , f. h. 1995 , apj , 454 , 698 wolfe , a. m. , turnshek , d. a. , smith , h. e. , & cohen , r. d. 1986 , apjs , 61 , 249 zwaan , m. a. , briggs , f. h. , sprayberry , d. , & sorar , e. 1997 , apj , 490 , 173 | this article outlines a method to interpret the extragalactic background light in terms of the large - scale , average properties of galaxies , including the comoving densities of stars and interstellar gas , metals , and dust .
these quantities are related by a series of coupled conservation - type equations analogous to the equations of galactic chemical evolution .
this approach enables us to combine observations of the emission and absorption in galaxies and thus to relate their average stellar and interstellar contents .
applications of the method include predictions of the global history of star formation from absorption - line observations and corrections to the cosmic uv emissivity for absorption by dust . |
we consider relative kauffman bracket skein module was defined in @xcite and it was noted there that , @xmath5 for @xmath6 with @xmath7 points on the boundary is a free module with the basis consisting of catalan connections . for an extensive discussions of theory of kauffman bracket skein modules
see @xcite . ]
( @xmath5 ) of @xmath8 where @xmath3 is @xmath2 parallelogram with @xmath1 points on the boundary arranged as it is shown in @xmath9 and @xmath10-tangle @xmath0 shown in @xmath11 that we will refer to as @xmath2__-lattice crossing__. if @xmath12 denotes the set of all catalan states for @xmath3 ( crossingless connections between boundary points ) then @xmath0 in @xmath5 of @xmath4 can be uniquely written in the form @xmath13 where @xmath14 .$ ] let @xmath15 be the set of all kauffman states ( choices of positive or negative markers for all crossings as shown in @xmath16.@xmath17 ) , after applying skein relations we have in @xmath5 of @xmath18@xmath19where , following standard conventions , for each kauffman state @xmath20 @xmath21 ( resp .
@xmath22 ) denotes the number of crossings with positive ( resp .
negative ) marker , @xmath23 the number of trivial components of the diagram obtained from @xmath0 by smoothing crossings according to markers defined by @xmath20 and @xmath24 denotes the diagram corresponding to state @xmath25 after removing trivial components . in @xmath5 of @xmath26
we have@xmath27so , for each catalan state @xmath28 define@xmath29we call @xmath30 a _ realizable catalan state _
if @xmath31 and denote by @xmath32 the set of all realizable catalan states .
analogously , we call @xmath30 a _ forbidden _ _ catalan state _ if @xmath33 and denote by @xmath34 the set of all forbidden catalan states@xmath35 in this paper we will consider the following problems : 1 ) : : which catalan states @xmath30 are realizable as kauffman states of @xmath36 2 ) : : what is an explicit formula for @xmath37 3 ) : : what is the number of elements of @xmath38 in the second section we show which catalan states @xmath30 can be realized by kauffman states of @xmath0 . in the third section , for a catalan state @xmath30 with no arc that starts and ends on the same _ side _ of @xmath0 , we give an explicit formula for the corresponding coefficient @xmath39 . finally , in the forth section , we obtain an explicit formula for the number of realizable catalan states .
enumerating horizontal and vertical lines as in @xmath40 let @xmath41 @xmath42 @xmath43 @xmath44 be _ horizontal lines _ and @xmath45 @xmath46 @xmath43 @xmath47 be _ vertical lines_. for a catalan state @xmath48 we let@xmath49where @xmath50 is the minimal number of intersections of @xmath51 and @xmath30 for @xmath52 or @xmath53 [ lemma 2.1]let @xmath28 and assume that @xmath54 or @xmath55 then @xmath56 observe that for any kauffman state @xmath25 of @xmath57 corresponding diagram @xmath58 obtained by smoothing crossings according to choice of markers defined by @xmath25 cuts horizontal and vertical separating lines precisely @xmath59 times and @xmath60 times . hence for the catalan state @xmath61 obtained from @xmath58 we must have @xmath62 and @xmath63 .
we say that a catalan state @xmath30 satisfies _ horizontal - line condition , _
@xmath64 if @xmath62 and , correspondingly , @xmath65 satisfies _ vertical - line condition , _
@xmath66 if @xmath67 we show that if @xmath28 satisfies both conditions @xmath68 and @xmath66 then @xmath30 is a realizable catalan state .
the following lemma gives sufficient conditions for a catalan state @xmath28 to meet condition @xmath66 or / and @xmath68 .
[ lemma 2.2]for catalan state @xmath69 @xmath70 : : if @xmath30 has no returns on the top horizontal boundary then @xmath30 satisfies condition @xmath71 @xmath72 : : if @xmath69 in addition , has no returns on the right side , then @xmath30 satisfies both @xmath68 and @xmath66 . for @xmath73
we argue by case distinction .
let @xmath74 be boundary points on top of @xmath30 , and @xmath75 be a separating line between @xmath76 and @xmath77 ( see @xmath78 ) .
if @xmath79th connection cuts @xmath75 for some @xmath80 then all @xmath81 connections do not cut @xmath75 .
since there are @xmath82 points on the right of @xmath75 and among them at least @xmath7 points yield connections in @xmath30 without cutting @xmath75 , it follows that there are at most @xmath60 points on the right of @xmath75 which can form connections with points on the left of @xmath75 . using similar reasoning , if @xmath83 then at most @xmath60 points on the left of @xmath75 can form connections with points on the right of @xmath75 . therefore , statement @xmath70 follows from basic counting .
if each @xmath84 connection does not cut @xmath75 for all @xmath85 @xmath86 @xmath43 @xmath87 we observe that , in particular , @xmath88 connections do not cut @xmath89 thus , as in the previous case , we argue that at most @xmath60 points on the left of @xmath75 can be connected with points on the right of @xmath75 .
statement @xmath90 follows directly from @xmath91 since @xmath92 can be applied to any side of the rectangle . for two catalan states
@xmath93 and @xmath94 with no returns on the bottom of the first and the top for the second one defines an operation @xmath95 of vertical composition of catalan states ( see @xmath96 ) . the operation
@xmath97 allows us to define catalan state @xmath98 for which the number of intersections with its vertical lines can easily be counted provided that the number of intersections with vertical lines of the factor states is known .
analogously , one defines horizontal composition @xmath99 of catalan states .
[ lemma 2.3]let @xmath93 and @xmath94 and assume that bottom of @xmath100 and top of @xmath101 are no return sides respectively @xmath102see @xmath103 .
then the vertical composition @xmath104 of states @xmath100 and @xmath101 is a catalan state with the following properties@xmath105 @xmath70 : : every vertical line @xmath75 intersects @xmath106 at most @xmath107 times and every horizontal line @xmath108 either in @xmath100 or @xmath109 cuts @xmath30 the same number of times as in @xmath100 or @xmath109 respectively@xmath35 @xmath72 : : catalan states @xmath100 and @xmath109 satisfy conditions @xmath110 and @xmath111 , for @xmath112 respectively if and only if @xmath30 satisfies conditions @xmath113 and @xmath66 . since @xmath100 and @xmath109
have no return sides on the bottom and the top respectively , vertical composition @xmath104 of states @xmath100 and @xmath109 ( as it is shown in @xmath96 ) is a catalan state .
furthermore , if @xmath75 is a vertical line in @xmath69 then from lemma lemma 2.2@xmath92 it follows that the number of times that @xmath75 cuts @xmath30 is bounded above by the number @xmath114 .
finally , if @xmath108 is a horizontal line either in @xmath100 or @xmath109 then @xmath108 is also a separating line in the composed state @xmath30 and , since @xmath100 and @xmath115 are composed along sides with no returns , the number of times @xmath116 cuts @xmath30 is identical as the number of times @xmath108 cuts either @xmath100 or @xmath109 .
this finishes our proof of @xmath117 statement @xmath72 follows from @xmath92 . for catalan states
@xmath30 satisfying conditions @xmath68 and @xmath118 we apply induction on @xmath119 to prove that such states could be obtained as kauffman states of @xmath120 an important step in our inductive proof is an observation that catalan states with @xmath121 ( or @xmath122 ) could be obtained as a vertical ( horizontal ) composition of catalan states .
[ lemma 2.4]let @xmath28 and assume that @xmath30 satisfies conditions @xmath68 and @xmath123 and @xmath121 .
then @xmath70 : : @xmath98 where @xmath124 , @xmath125 , @xmath126 and @xmath72 : : catalan states @xmath100 and @xmath109 satisfies conditions @xmath110 and @xmath111 , for @xmath112 respectively . since @xmath127
, there is a horizontal separating line @xmath128 that cuts @xmath30 precisely @xmath59 times . splitting @xmath30 along @xmath75 yields a pair of catalan states @xmath93 and @xmath94 , where @xmath129 since @xmath59 is the number of intersections of @xmath75 with catalan state @xmath69 it follows that @xmath100 has the bottom side with no returns and @xmath109 has the top side with no returns .
therefore , state @xmath30 can be expressed as @xmath104 . since @xmath30 satisfies conditions @xmath68 and @xmath130 it follows from lemma [ lemma 2.3]@xmath72 that each @xmath131 satisfies conditions @xmath110 and @xmath111 , for @xmath132 respectively . obviously , the statement of _ split lemma _ holds when @xmath133 . in such a case ,
@xmath134 where @xmath135 , @xmath136 , @xmath137 and each @xmath131 satisfies conditions @xmath138 and @xmath139 , for @xmath132 respectively .
now , we state and prove the main result for this section . [ theorem 2.5]assume that @xmath30 satisfies conditions @xmath68 and @xmath66 , then @xmath30 can be obtained as a kauffman state of @xmath140 we proceed by induction on @xmath119 .
clearly , statement is vacuously true for @xmath141 , and it is obvious when @xmath142 .
we assume that statement is true for all pairs @xmath143 such that @xmath144 @xmath145 and @xmath146 where @xmath147 we show that statement is true for @xmath144 @xmath145 and @xmath148 and wlog we can assume @xmath149 .
let @xmath28 and @xmath30 satisfies conditions @xmath64 and @xmath66 .
we have the following cases 1 ) : : @xmath127 or @xmath122 if @xmath150 then applying _ split lemma , _ we have @xmath151 , where @xmath93 , @xmath94 @xmath152 and each @xmath131 satisfies conditions @xmath110 and @xmath111 , for @xmath85 @xmath153 . since @xmath154 , applying induction hypothesis , we obtain corresponding kauffman states @xmath155 and @xmath156 for @xmath157 and @xmath158 respectively such that @xmath159 and @xmath160 .
let @xmath25 be kauffman state obtained by putting markers according to @xmath155 in the first @xmath161 rows of @xmath0 and according to @xmath156 in the remaining @xmath162 rows of @xmath0 .
clearly , for kauffman state @xmath20 we have @xmath163 .
proof in the case when @xmath122 is analogous .
2 ) : : @xmath164 and @xmath165 that is , each horizontal line cuts @xmath30 less than @xmath59 times and each vertical line cuts @xmath30 less than @xmath60 times . in this case we will construct a kauffman state @xmath25 such that @xmath166 using an algorithm given in the following lemma .
[ lemma 2.6]assume that @xmath164 and @xmath167 , then @xmath30 can be obtained as a kauffman state of @xmath168 by assumption , each horizontal line cuts @xmath30 less than @xmath59 times and each vertical line cuts @xmath30 less than @xmath60 times , so the top boundary side of @xmath30 must have at least one return ( see @xmath169 ( left ) ) , or at least one corner connection ( see @xmath169 ( right ) ) .
if @xmath30 has no corner connection , we start construction by defining kauffman state @xmath170 ( first row of desired kauffman state @xmath25 ) for the returning connection of @xmath65 between pair of points @xmath84 and @xmath171 on the top of @xmath30 with minimal index @xmath172 , @xmath173 as shown in @xmath174 ( left ) .
if there is a corner connection , we take as kauffman state @xmath155 ( first row of kauffman state @xmath25 ) the state shown in @xmath174 ( right ) . using isotopy
, we deform state @xmath30 into an equivalent state @xmath175 with top as shown in @xmath174 and , for which all remaining connections form @xmath176-catalan state@xmath35 since every vertical line cuts @xmath30 less than @xmath60 times , we observe that it must cut each vertical line no more then @xmath177 times .
it follows that @xmath178-catalan state @xmath179 obtained from @xmath175 by removing its top ( see @xmath174 ) cuts all vertical separating lines no more than @xmath180 times . in particular ,
catalan state @xmath181 satisfies conditions @xmath182 and @xmath183 and one may apply induction hypothesis for the state @xmath181 .
therefore , there is a kauffman state @xmath184 of @xmath185 such that @xmath186 . in order to complete our construction ,
we take for @xmath187 the state obtained by putting markers according to state @xmath155 in the first row of @xmath0 and putting markers according to @xmath188 in its remaining @xmath189 rows . for the choice of markers determined by @xmath25
, we clearly have @xmath190 proof of lemma [ lemma 2.6 ] also completes our proof of theorem theorem 2.5 .
in this section we prove result concerning coefficients of realizable catalan states for @xmath0 for the case of catalan states with no returns . this result was proved in @xmath191 by j. hoste and j. h. przytycki while they were working on the kauffman bracket skein module of the whitehead type manifolds .
[ theorem 3.1]in @xmath5 of @xmath192 with @xmath1 points we have@xmath193where @xmath194 @xmath195 is the catalan state with @xmath7 positive diagonal arcs , @xmath196 negative diagonal arcs and @xmath197 vertical arcs @xmath102see @xmath198 and the _ remainder _ is a polynomial @xmath102over @xmath199 $ ] @xmath200 in variables consisting of realizable catalan states that have at least one return . by jones - wenzl idempotents @xmath201 and @xmath202 ( respectively ) then reminder is @xmath203 since @xmath204 for any @xmath172 and @xmath205 . ]
before we give a proof of theorem [ theorem 3.1 ] , we need to recall the notation @xmath206-analogue of an integer @xmath60 and @xmath206-analogue of binomial coefficient@xmath35 a @xmath206-analogue of an integer @xmath60 is defined as follows @xmath207_{q}=1+q+q^{2}+ ... +q^{m-1},$ ] and if @xmath207_{q}!=\left [ 1\right ] _ { q}\cdot \left [ 2\right ] _ { q}\cdot ... \cdot \left [ m\right ] _ { q}$ ]
then @xmath206-analogue of binomial coefficient is defined by@xmath208_{q}!}{[k]_{q}!\cdot \lbrack m - k]_{q}}\]]one shows using the following identity for @xmath206-analogue of binomial coefficient@xmath209and induction that @xmath210 $ ] .
applying kauffman bracket skein relation ( see @xmath211 ) for the crossing in top left corner of @xmath0 ( see @xmath212 ) we obtain the following recursive formula @xmath213 now , we let @xmath214 and @xmath215 one verifies that @xmath216 satisfies the above recurrence[multiblock footnote omitted ] .
namely , we have by the property of @xmath206-analogue of binomial coefficient with @xmath217@xmath218as needed
. one of the goals for future research is to find a `` _ _ relatively good _ _ '' formula for coefficients of all realizable catalan states of @xmath219 in @xmath5 of @xmath192 with @xmath1 points .
in particular , theorem [ theorem 3.1 ] gives an explicit formula for all coefficients of catalan states with no returns .
a natural generalization of this family catalan states would be families of catalan states with three , two , or just one side that has no return . as we proved in lemma [ lemma 2.2 ]
even if only one side @xmath102left or right@xmath200 of the catalan state has no return then any horizontal separating line can cut such a catalan state in at most @xmath59 points .
in this section , we find a closed formula for @xmath220 the number of catalan states that could be obtained as kauffman states of @xmath57 or alternatively , we find number of elements of the set @xmath221 of all _ forbidden catalan states . _ as we showed in theorem [ theorem 2.5 ] , a catalan state @xmath222 is realizable if and only if @xmath30 satisfies conditions @xmath64 and @xmath223 therefore , every forbidden state @xmath30 must intersects at least one of the horizontal or vertical lines more than @xmath59 or @xmath60 times respectively .
let @xmath224 and @xmath225 denote sets of catalan states that are excluded by horizontal and vertical condition respectively .
more precisely , we have for the lines in @xmath226 @xmath227we call them respectively _ horizontally excluded _ and _ vertically excluded catalan states_. clearly , we see that@xmath228however , more is true , i.e. @xmath229 , which is a consequence of the following proposition .
[ proposition 4.1]let @xmath230 be a circle with an even number of points @xmath231 and consider a pair of lines intersecting at one point and splitting @xmath230 into sectors @xmath232 with @xmath233 points each sectors , @xmath85 @xmath86 @xmath234 @xmath235 @xmath102see @xmath236 .
let @xmath30 be a catalan state and @xmath237 be the number of intersections of @xmath30 with the pair of lines .
( i ) : : if points from sectors @xmath238 and @xmath239 are not connected in @xmath240 then @xmath241 ( ii ) : : if points from sectors @xmath242 and @xmath243 are not connected in @xmath240 then @xmath244 ( iii ) : : for any @xmath69 @xmath245 .
we observe that in a catalan state @xmath30 we can not have points from @xmath242 connected with points from @xmath246 and points from @xmath238 connected with points from @xmath239 at the same time .
for @xmath247 we notice that since points from sectors @xmath238 and @xmath239 are not connected in @xmath69 if @xmath248 is the number of connections in @xmath30 between points from sectors @xmath242 and @xmath243 then these connections intersect pair of lines at @xmath249 points and involve @xmath249 points from sectors @xmath242 and @xmath243 in total .
consequently , there is a total of @xmath250 points left in sectors @xmath242 and @xmath243 that may form connections intersecting the pair of lines at most once each .
it follows that total number of intersections of catalan state @xmath30 with the pair of lines is bounded above by @xmath251 .
proof of @xmath252 follows by changing the roles of @xmath253 @xmath243 and @xmath254 @xmath255 and statement @xmath256 is a direct consequence of @xmath257 and @xmath258 .
proposition [ proposition 4.1 ] has the following important consequence .
[ corollary 4.2]horizontally and vertically excluded catalan states are disjoint . for a pair of lines consisting of a vertical and a horizontal line as in @xmath259
we see that @xmath260 applying result of proposition [ proposition 4.1 ] for such a pair of lines , for any catalan state @xmath69 we have @xmath261 therefore , if the horizontal line intersects @xmath30 more then @xmath59 times then the vertical line must intersect @xmath30 less than @xmath60 times .
analogously , if the vertical line intersects @xmath30 more then @xmath60 times then the horizontal line must intersect @xmath30 less than @xmath59 times . as a consequence of corollary [ corollary 4.2 ] and theorem [ theorem 2.5 ] , we obtain the following result .
[ corollary 4.3]let @xmath262 and @xmath263 be the number of horizontally and vertically excluded catalan states .
the number of forbidden catalan states equals @xmath264 it is worth to note that following relation between numbers @xmath265 and @xmath266 holds@xmath267consequently , from now on , we will consider only @xmath262 . in order to find @xmath268 we start by constructing bijection between the set @xmath269 and the set of dyck paths @xmath270 of semilength @xmath119 from @xmath271 to @xmath272 that never pass below the line @xmath273 and consists only the vertical steps @xmath274 ( _ ascents _ ) and horizontal steps @xmath275 ( _ descents _ ) as it is shown in @xmath276 .
consider @xmath277 lines @xmath278 @xmath279 @xmath43 @xmath280 as in @xmath281 .
any catalan state @xmath282 uniquely determines the minimal number of intersections @xmath283 of @xmath30 with each line @xmath284 @xmath285 and let @xmath286 [ lemma 4.4]for all @xmath287 we have @xmath288 we observe that for any three successive lines @xmath289 @xmath290 and @xmath291 where @xmath292 @xmath86 @xmath43 @xmath293 , we have @xmath235 possibilities ( see @xmath294 ) for the intersections of @xmath30 with lines @xmath289 @xmath290 and @xmath295 @xmath105 for instance , in the second case shown in @xmath296 , we have@xmath297analogously , for the remaining cases shown in @xmath298 @xmath299 and @xmath300 therefore , for all @xmath301 @xmath302 @xmath43 @xmath303 we have @xmath304 .
for the remaining lines @xmath278
@xmath279 @xmath305 and @xmath306 @xmath307 @xmath308 we see that@xmath309therefore , we have @xmath310 @xmath311 or @xmath312 and analogously @xmath313 or @xmath314 @xmath315 .
hence , we have @xmath316 for all @xmath317 as we claimed .
now , let @xmath28 and let@xmath318we define a lattice path @xmath319 by putting@xmath320clearly , for any @xmath48 lattice path @xmath321 of semilength @xmath119 from @xmath271 to @xmath322 never pass below the line @xmath323 thus @xmath321 is a dyck path .
furthermore , we have the following result .
[ proposition 4.5]the map @xmath324 given by@xmath325is a bijection@xmath35 it suffices to show that @xmath326 is injective .
let @xmath327 @xmath328 and suppose that@xmath329therefore , for every @xmath292 @xmath86 @xmath43 @xmath330 @xmath331 that is@xmath332since @xmath333 and @xmath334 we have @xmath335 by induction , we have@xmath336it follows that @xmath100 and @xmath109 have identical minimal number of intersections with all lines @xmath278 @xmath279 @xmath43 @xmath337 hence @xmath338 .
following the notations used in @xcite ( see pp .
@xmath339 and lemma @xmath340 pp .
@xmath341 ) , let @xmath342 denote the set of paths from @xmath343 to @xmath344 not touching the line @xmath345 where @xmath346 and let @xmath347 denote the set of paths from @xmath348 to @xmath143 not touching the lines@xmath349where @xmath350 @xmath351 as it was shown in @xcite , we have@xmath352 , \end{aligned}\]]where@xmath353 [ theorem 4.6]the number @xmath262 of horizontally excluded catalan states @xmath102see also @xmath354 is given by@xmath355\ ] ] before we prove theorem [ theorem 4.6 ] , we note that if @xmath356 lines @xmath357 @xmath358 and @xmath359 may intersect a catalan state @xmath30 satisfying horizontal line condition @xmath68 in the pattern : @xmath360 @xmath361 and @xmath362 ( see @xmath363)@xmath35 recall , for a lattice path @xmath364 that starts from @xmath271 , one defines@xmath365where @xmath366 and @xmath367obviously , path @xmath368 for any @xmath292 @xmath86 @xmath43 @xmath369 we have@xmath105@xmath370 using bijection @xmath371 from proposition [ proposition 4.5 ] , we have:@xmath372since @xmath373 and @xmath374 after standard algebraic simplifications , we have @xmath375% .\end{aligned}\ ] ] this finishes our proof .
[ corollary 4.7]the number of all realizable catalan states by @xmath376 lattice of crossings @xmath377 is given by@xmath378 in particular , for @xmath379 @xmath380
j. h. przytycki was partially supported by the nsa - ams 091111 grant , and by the gwu ref grant .
authors would like to thank ivan dynnikov and krzysztof putyra for many useful computations / discussions . | for a lattice crossing @xmath0 we show which catalan connection between @xmath1 points on boundary of @xmath2 rectangle @xmath3 can be realized as a kauffman state and we give an explicit formula for the number of such catalan connections . for the case of a catalan connection with no arc starting and ending on the same side of the tangle ,
we find a closed formula for its coefficient in the relative kauffman bracket skein module of @xmath4 .
_ keywords : _ _ _ _ _ _ _ |
the precision of the experimental data on electroweak interactions and qcd is now very high and it is expected to become significantly higher within the next few years .
this has triggered a substantial refinement in the corresponding theoretical calculations . yet ,
already now for certain experimental quantities the theoretical uncertainty is one of the major open questions in the interpretation of the data and in the search for signals of physics beyond the standard model .
a striking example is the need for a precise determination of the gauge couplings at the weak scale , which is the prerequisite for investigation of possible unification of couplings at some gut scale .
one of the reasons for this current state of affairs in the relation between the theory and experiment is that computation of high orders in perturbation theory for quantum field theories , and especially non - abelian gauge theories in @xmath0 dimensions is extremely hard .
state - of - the - art calculations available today for this kind of theories have reached , after a very large effort , the 3-rd and the 4-th order in @xmath1 , for observables and for the @xmath2-function , respectively @xcite . without a major breakthrough in the relevant techniques it is unlikely that exact results for the next order will become available in the foreseeable future . moreover , even if explicit expressions for very high order terms do become available , we still have to deal with the fact that the perturbative series of interest are asymptotic , with zero radius of convergence and usually are not even borel summable . in this talk
i will review an approach which has been recently suggested to deal with some of these problems .
as mentioned in the introduction , the perturbation series in qcd is expected to be asymptotic with rapidly growing coefficients : s(x ) = ^_n=0 c_n x^n , x , c_n n!k^n n^[genericseries ] for some coefficients @xmath3 @xcite .
anyone who wants to make use of qcd perturbation theory to carry out precision analysis of observables has to face several practical problems : * only few first orders in ( [ genericseries ] ) are known for any observable ( @xmath4 ) * the series has zero radius of convergence * the series is usually not _
borel summable_. borel summation is a trick that sometimes works for summing series with factorial divergence .
consider the series for @xmath5 in eq .
( [ genericseries ] ) .
we can define a new function , @xmath6 , whose series is obtained from ( [ genericseries ] ) by dividing the @xmath7-th term by @xmath8 , ( y ) = ^_n=0 ( c_nn ! ) y^n [ borelseries ] if the new series is convergent , the original function @xmath5 can be obtained by the so - called inverse borel transform , s(x ) = 1x_0^e^-y / x s(y ) dy [ invboreltransform ] provided @xmath6 has no singularities along the integration path . 0.5truecm
the presence of singularities along the integration path makes the integral ( [ invboreltransform ] ) ill - defined .
one can try to define it by going around the poles , but this introduces an ambiguity proportional to the pole residue , since different deformations of the integration path will give different results .
* renormalization scale dependence : finite - order perturbative predictions depend on the arbitrary renormalization scale @xmath9 through the coupling , @xmath10 .
this renormalization scale is most pronounced at leading order in perturbation theory and decreases with the inclusion of higher order terms .
* renormalization scheme dependence : in principle , the theory can be renormalized in any valid renormalization scheme , yielding the same predictions for any physical observable . in practice , when we work with a finite number of perturbative terms , the results depend on the renormalization scheme .
there is no miracle cure " which would solve these problems completely .
however , we should and can minimize their effect .
thus the practical issue is how to get the best possible precision , given a fixed number of terms in the perturbative expansion . in the following ,
i will discuss one method which has already shown considerable progress towards this goal .
the method is based on the so - called pad approximants ( pa - s ) @xcite-@xcite .
pad approximants @xcite are rational functions chosen to have a taylor expansion equal the perturbative series to the order calculated .
given a series s(x ) = c_0 + c_1 x + c_2 x^2 + c_3 x^3 + + c_n x^n + o(x^n+1 ) [ ston ] one can always find a rational function : l+m = n [ padedef ] such that @xmath11 $ ] has the taylor series = c_0 + c_1 x + c_2 x^2 + c_3 x^3 + + c_n x^n + c_n+1 x^n+1 + [ lmseries ] the rational function in ( [ padedef ] ) is called the @xmath11 $ ] pad approximant .
it is important to keep in mind that at a given finite order in @xmath12 the @xmath11 $ ] pad in ( [ padedef ] ) is formally as valid representation of @xmath5 as the original perturbation expansion .
moreover , in practice the pa - s turn out to posses many important and useful properties which are absent in the straightforward perturbation theory .
thus , even though pa is constructed to reproduce the series ( [ ston ] ) only up to order @xmath13 , it turns out that under rather mild conditions the next term in the taylor expansion of the pa in eq .
( [ lmseries ] ) , @xmath14 , provides a good estimate , @xmath15 .
we call it the _ pade approximant prediction _ ( pap ) , of the next coefficient @xmath16 in the series ( [ ston ] ) : -1 1 [ esti ] and for sufficiently large @xmath13 the relative error decays exponentially fast , -1 ~ e^-n,= [ estii ] let us consider some simple examples , starting with the trivial case of a single - pole geometric series = _ n=0^c_n x^n [ singlepole ] it is easy to convince oneself that in this case the @xmath11 $ ] pad is _ exact _ for @xmath17 , @xmath18 .
for example , if we attempt to construct a [ 10/10 ] pad of ( [ singlepole ] ) , we will find that the _ a priori _ 10-th degree polynomials in numerator and denominator reduce to a degenerate case of a constant and 1-st degree polynomial , respectively , = [ tenten ] once this is clear , the extension to a sum of finite number of poles in obvious , _ i=1^k = _
n=0^c_n x^n l k-1 , m k [ npoles ] one can also show that for an infinite number of isolated poles , i.e. when @xmath19 is a meromorphic function , the sequence of @xmath20 $ ] for @xmath21 fixed converges to @xmath22 as @xmath23 , _
l f(x ) ; k = 0 , 1 , 2 , a somewhat less intuitive , but very important result is that in certain cases the pad sequence @xmath20 $ ] converges exponentially fast in @xmath24 to the correct function even for a factorially divergent asymptotic series with zero radius of convergence .
a classical example @xcite is the function g(x ) = _ 0^ d t = _ 0^(-x)^nn ! [ factorialseries ] here again it turns out that @xmath20 \ , \longrightarrow \ , g(x)$ ] as @xmath25 .
the crucial property of the series in ( [ factorialseries ] ) which makes this possible is that it has alternating signs .
it is easy to show that this implies that all the poles of the borel transform of ( [ factorialseries ] ) are on the negative real axis , and hence that the series is borel summable . more generally ,
when the series is borel summable , pad will converge to the correct result .
it is interesting to note that the exponentially fast convergence of pa - s is not limited to meromorphic functions . as a simple example , consider the hypergeometric function @xmath26 @xcite , which has a cut for @xmath27 .
[ hyperfig ] shows that despite the cut , the diagonal pa - s evaluated at @xmath28 converge exponentially fast .
while such mathematical examples are instructive , in order to gain confidence in the method , we need to see how it fares on high - order series taken from quantum field theory . as the first test case
, we consider the scalar field theory with gaussian propagators .
high - order perturbation expansions of green s functions in this theory have been computed in ref .
[ phi4fig ] demonstrates the convergence of pap for the relevant coefficients for the 4-point green s function in @xmath29 @xcite .
the relative error is @xmath30 at 5-th order . for comparison
also shown are relative errors of estimates based on asymptotic behavior of large orders in perturbation theory , as given in @xcite . clearly , at 5-th order pad does better by about 4 orders of magnitude .
if information is available about the asymptotic behavior of @xmath31 , it is possible to obtain an explicit expression for the the error formula on the r.h.s . of ( [ estii ] ) .
for example , we have demonstrated that if _
n - 1 1n , [ epsdef ] as is the case for any series dominated by a finite number of renormalon singularities , then @xmath32}$ ] defined by _
[ l / m ] [ efour ] has the following asymptotic behaviour _ [ l / m ] - , where k = l+m+am [ efive ] and where @xmath33 is a number of order 1 that depends on the series under consideration . for large @xmath34 eq .
( [ efive ] ) yields an exponential decrease of the the error , as in eq .
( [ estii ] ) .
this prediction agrees very well with the known errors in the pap s @xcite for the qcd vacuum polarization d function calculated in the large @xmath35 approximation @xcite , as seen in fig .
[ figiii]a .
one can repeat this exercise also for the borel transform of the d function series .
as mentioned earlier , a generic borel transform is characterized by the presence of poles ( more generally , branch points ) on the real axis . in view of this
, we expect an even faster convergence in this case , since the pad , being a rational function , is particularly well - suited to reproduce this analytic structure .
indeed , it turns out that in this case @xmath36 and _ [ m / m ] - [ eeight ] which agrees very well with the corresponding pap results shown in fig .
[ figiii]b @xcite .
the high degree of agreement between the analytical error estimates in eqs .
( [ efive ] ) and ( [ eeight ] ) and the actual errors in pap suggest that one can substantially improve the pap method by systematically including the error estimates @xmath32}$ ] _ as a correction _ , yielding the asymptotic pad approximant predictions ( apaps ) : @xmath37 where @xmath38 is the original pap prediction without the additional correction , as in eq .
( [ efour ] ) , and @xmath39 is obtained by fitting ( [ efive ] ) to the known lower orders @xcite .
the apap method results not only in a substantial improvement of the pap estimates , but also significantly reduces the difference between the predictions based on different @xmath11 $ ] values at a given order , .
the method has been applied to the bjorken sum rule for the difference of first moments of proton and neutron structure functions @xmath40 in polarized deep inelastic scattering @xcite .
for @xmath41 the sum rule reads ^1_0 [ g_1^p(x , q^2 ) - g_1^n(x , q^2 ) ] = 1 6 |g_a|-0.2em [ bjf ] where @xmath42 and the exact expression for @xmath43 is still unknown .
the pap and the corresponding apap estimates of @xmath43 are lccc [ 2/1 ] : & c_4^pap 114 & & c_4^apap 131 + + [ 1/2 ] : & c_4^pap 111 & & c_4^apap 130 [ c4bjsr ] clearly , apap estimates show significantly less spread than the corresponding pap estimates .
remarkably , the apap estimates of @xmath44 show an almost perfect agreement with an independent estimate , based on a completely different method @xcite : @xmath45 ! as already mentioned , a typical finite order perturbative series such as ( [ bjf ] ) exhibits a spurious renormalization scale and scheme dependence .
schematically we have , _ =
_ + _ [ observable ] replacing a finite - order perturbative series by a pad is equivalent to adding an infinite series of estimated terms generated by the rational approximant .
if such an estimate is accurate , we expect to see a reduction in the renormalization scheme and scale dependence . as shown in fig .
[ parcol ] , this expectation is fully realized when pad is applied @xcite to the bjorken sum rule series in eq .
( [ bjf ] ) .
it turns out that this dramatic reduction in the scale and scheme dependence can also be understood on a deeper level . in ref .
@xcite it was shown that in the large-@xmath46 limit , i.e. when the @xmath2 function is dominated by the one loop contribution , the scale dependence is removed completely .
this is because in this limit the renormalization scale transformation of @xmath1 reduces to a homographic transformation of the pad argument .
diagonal pa s are invariant under such transformations @xcite . non - diagonal pa s are not totally invariant , but they reduce the rs dependence significantly @xcite . in the real world
the usual @xmath2 function includes higher - order terms beyond @xmath46 .
still , in qcd with @xmath47 , the 1-loop running of the coupling is dominant and therefore pa s are still almost invariant under change of renormalization scale .
a further related interesting development is the observation @xcite that the pade approximant approach for resummation of perturbative series in qcd provides a systematic method for approximating the flow of momentum in feynman diagrams . in the large-@xmath46 limit , diagonal pa
s generalize the brodsky - lepage - mackenzie ( blm ) scale - setting method @xcite to higher orders in a renormalization scale- and scheme - invariant manner . -1truecm
although no qcd observables have been calculated exactly beyond @xmath48 , in fall of 1996 we had learned that a calculation of the 4-loop contribution to the qcd @xmath2 function was under way , and likely to be published soon .
the prediction of the unknown 4-loop coefficient @xcite was therefore an important challenge and excellent testing ground for the new apap method .
as a warm - up exercise one can test the apap method on the 4-loop @xmath2 function of the @xmath49 theory with @xmath50 global vector symmetry @xcite , the latter being analogous to the @xmath51 global symmetry of qcd .
the results @xcite for @xmath52 in @xmath50 @xmath49 theory are shown in figure [ betaphi4 ] .
clearly , the variant of apap method denoted as @xmath53 ( see @xcite for details ) is markedly superior to the naive pap .
the 5-loop @xmath2 function in this theory is also known @xcite and the corresponding apap estimates also turn out to be very precise @xcite .
consequently this was the method of choice for the qcd 4-loop @xmath2 function .
the strategy for computing @xmath52 , the 4-loop @xmath2 function coefficient , is as follows .
we recall that @xmath52 is a cubic polynomial in the number of flavors @xmath35 : _
3 = a_3 + b_3 n_f + c_3 n_f^2 + d_3 n_f^3 , [ beta3form ] where @xmath54 ( for @xmath55 ) is known from large-@xmath35 calculations @xcite .
the known exact expressions for the 1- , 2- and 3-loop @xmath2 function are used as input to apap , to predict the value of @xmath52 for a range of @xmath35 values .
the predictions for @xmath56 are then obtained from fitting the apap results for @xmath57 to a polynomial of the form ( [ beta3form ] ) . shortly after the apap prediction of @xmath52 @xcite
the exact result was published in ref .
one important lesson from the exact results is that they contain qualitatively new color factors , corresponding to quartic casimirs , analogous to light - by - light scattering diagrams in qed .
such terms are not present at 2- and 3-loop level , and therefore can not be estimated using the pad method .
numerically the correction due to these new color factors is not very large , but in principle the pap estimates should be compared with the rest of the exact expression , as is done in the first three columns of table i. # 1 the apap estimates for @xmath58 , @xmath60 and @xmath61 seem quite satisfactory , until one realizes that the @xmath58 and @xmath60 terms in ( [ beta3form ] ) have opposite signs and their magnitude is such that they almost cancel each other at @xmath62 .
this means that for numerical prediction of @xmath52 as function of @xmath35 in the physically interesting range @xmath63 a better precision is required .
fortunately , it is possible to obtain such precision .
this is accomplished by formally using negative values of @xmath35 in the fitting procedure , so that no cancellation occurs , and making a careful choice of the range of negative values of @xmath35 used for the fit .
once the values of @xmath58 , @xmath60 and @xmath61 are obtained this way , one can use them to compute the physical predictions at positive @xmath35 @xcite .
this procedure has been referred to as wapap , for weighted apap " .
the corresponding results are shown and compared with exact results in the last two columns of table i. we see a dramatic improvement in the precision .
figure [ beta3_error ] displays graphically predictions for @xmath52 , as a function of @xmath35 for the most interesting case @xmath64 .
we plot the percentage relative errors obtained using various apap - based estimation schemes @xcite : naive apap s fitted with positive @xmath65 ( diamonds ) , naive apap s fitted with negative @xmath66 , wapap s compared to the exact value of @xmath52 including quartic casimir terms , and wapap s compared to @xmath52 without quartic casimir terms ( crosses ) .
we see that the latter are the most accurate for @xmath52 in qcd . in figure [ beta3_vs_nf_vs_nc ]
we show the error in the wapap prediction for @xmath52 as a function of @xmath35 , for @xmath673 , 4 , 5 , 6 , 7 and 10 , once again omitting quartic casimir terms from the exact result .
the accuracy of these predictions is our best evidence for believing in the utility of the wapap method .
the wapap method does very well on the four - loop qcd @xmath2 function , but the details of the method were fine - tuned after the exact results became available . in ref .
@xcite predictions were also given for yet unknown 5-loop @xmath2 function in qcd and 4- and 5-loop @xmath2 function in @xmath68 supersymmetric qcd .
it is extremely interesting to see how well our predictions will do for these quantities .
this paper is devoted to the memory of mark samuel , a wonderful collaborator and friend , who passed away suddenly last november .
the results described here were obtained in a close collaboration with him , together with stan brodsky , john ellis , einan gardi , ian jack and tim jones .
this research was supported in part by the israel science foundation administered by the israel academy of sciences and humanities , and by a grant from the g.i.f . , the german - israeli foundation for scientific research and development . for a review ,
see for example : t. van ritbergen , j.a.m .
vermaseren , s.a .
larin and p. nogueira , ( 1995)513;a.l .
kataev , talk at _ second workshop on continuous advances in qcd _ , minneapolis , 1996 , hep - ph/9607426
. t. van ritbergen , j.a.m .
vermaseren and s.a .
larin , hep - ph/9701390 , ( 1997)379 . j.a.m .
vermaseren , s.a .
larin and t. van ritbergen , hep - ph/9703284 , ( 1997)327;k.g .
chetyrkin , hep - ph/9703278 , ( 1997)161 .
a. mueller ( 1985)327 ; a.i .
vainshtein and v.i .
zakharov , ( 1994)1207 . | i review the recent work on obtaining precise estimates of higher - order corrections in qcd and field theory . |
one of the main challenges faced today by particle physics is to elucidate the nature of electroweak symmetry breaking . this is the only aspect of the standard model that still has not been directly tested by experiment .
moreover , there is a general agreement in the sense that , due to the hierarchy and triviality problems , the standard higgs sector is not satisfactory and it really points to the existence of new physics at the tev scale .
one possibility for such new physics is the existence of a new strong interaction that dynamically breaks the electroweak symmetry @xcite .
the best known realization of this elegant idea , technicolor , has been challenged by lep precision measurements .
in fact , the original models , which are scaled up versions of qcd , are ruled out .
a great progress toward the construction of potentially more successful models was the introduction of walking technicolor which basically means that the interaction is quasi - conformal over a large range of energy .
nevertheless , in general these models predict that the composite scalar ( the higgs ) in the spectrum must be heavy , while the current experimental data seem to point to the existence of a light higgs .
very recently a new kind of technicolor models has been proposed @xcite whose main characteristic is that technifermions are not in the fundamental representation of the technicolor group . in these models
the walking behavior of the coupling constant appears naturally and they are not in conflict with the current limits on the oblique parameters . but
the most remarkable feature of these models is that they predict the existence of light composite higgs with a mass around 150 gev . inspired by such models , we write down an effective lagrangian which describes the standard model with a light higgs and vector resonances which are a general prediction of dynamical symmetry breaking models @xcite .
the model is minimal in the sense that we assume that any other composite state would be heavier than the vector resonances , and so they are not taken into account , and there are no physical technipions in the spectrum .
in what follows we will study some aspects of such a model .
section [ sec : lag ] is devoted to the construction of the lagrangian . in section [ sec
: res ] we compute the cross section for the associate production of a higgs and a gauge boson and we compare the results with the standard model . finally in section
[ sec : con ] we present our conclusions .
we start by noticing that , in general , dynamical electroweak symmetry breaking models predict the existence of composite vector particles ( the so called technirho and techniomega ) that mix with the gauge bosons of the standard model . in order to describe this mixing
, we use a generalization of vector meson dominance @xcite introduced in @xcite and developed in @xcite . in this approach
we choose a representation where all vector fields transform as gauge fields and they mix through a mass matrix . on the other hand , gauge invariance imposes that the mass matrix has a null determinant . in our case
, the lagrangian for the gauge sector can be written as : @xmath0 where @xmath1 and @xmath2 ( @xmath3 ) is the hard mass of the proto - technirho ( proto - techniomega ) .
notice that our lagrangian is written in term of non - physical fields .
the physical ones will be obtained by diagonalizing the mass matrix . by construction ,
lagrangian ( [ eq : vectlag ] ) is invariant under @xmath4 .
the symmetry breaking to @xmath5 will be described by mean of the vacuum expectation value of a scalar field , as in the standard model . in other words
, we will use an effective gauged linear sigma model as a phenomenological description of the electroweak symmetry breaking . as usual , fermions are minimally coupled to gauge bosons through a covariant derivative . because in our scheme all the vector bosons transform as gauge fields , it is possible to include the proto - technirho and the proto - techniomega in an `` extended '' covariant derivative @xcite , resulting in the following lagrangian for the fermion sector : @xmath6 with @xmath7 and @xmath8 the parameters @xmath9 ( @xmath10 ) play a role similar to fermion delocalization in deconstruction models @xcite . a direct coupling between fermions and the vector resonances can appear naturally in technicolor due to extended technicolor interactions .
nevertheless , they must be proportional to the fermion mass and then , they are not important except for the top quark . in what follows , for simplicity , we take @xmath11 . in our effective model
the higgs sector is assumed to be the same as in the standard model except by the possibility of including a direct coupling between the higgs doublet and the vector resonances through an `` extended '' covariant derivative as was shown for the fermion sector .
that is , the lagrangian for the higgs sector is : @xmath12 where , as usual @xmath13 and @xmath14 this direct coupling may seem natural in a context where the higgs and the vector resonances are both composite states of the same underlying strong sector . nevertheless ,
in this work we will avoid it ( we choose @xmath15 ) because , in principle , it can introduce dangerous tree level corrections to the @xmath16 parameter .
once the electroweak symmetry has been broken , the mass matrix of the vector bosons takes contributions from ( [ eq : vectlag ] ) and from the higgs mechanism . for the neutral vector bosons ,
the resulting mass matrix is : @xmath17{c c c c } ( 1+\alpha)g^2 & -\alpha g g_2 & -g g ' & 0 \\ -\alpha g g_2 & \alpha g_2 ^ 2 & 0 & 0 \\ -g g ' & 0 & ( 1+\alpha ' ) g'^2 & - \alpha ' g ' g_2'\\ 0 & 0 & - \alpha ' g ' g_2 ' & \alpha ' g'^2 \end{array } \right ] \ ] ] where @xmath18 and @xmath19 in what follows , in order to simplify our analysis , we will assume that @xmath20 and @xmath21 .
notice that this assumption implies that the technirho and the techniomega will have the same mass .
we will also assume that @xmath2 is proportional to @xmath22 which is quit natural when we realize that @xmath2 is a dynamical mass produced by the underlying strong interaction .
on the other hand , the mass matrix for the charged vector bosons can be written as : @xmath23{c c } ( 1+\alpha)g^2 & -\alpha g g_2 \\ -\alpha g g_2 & \alpha g_2 ^ 2 \end{array } \right ] \ ] ] we diagonalize the mass matrices in the limit @xmath24 and we obtain the following expressions for the physical fields ( we have dropped the lorentz indexes ) .
we write the transformation to physical fields in the limit @xmath25 and we keep terms up to order @xmath26 . @xmath27 where @xmath28 and @xmath29 the relevant feynman rules , in terms of the physical fields , for the associate production of a higgs and a gauge boson can be found in table [ tab : feynmanrules ] while figure [ fig : diagrams ] shows the feynman diagrams for the processes studied in this work .
.feynman rules for the relevant couplings of the vector resonances for the associated production of a higgs and a gauge bosons .
the couplings of the @xmath30 and @xmath31 to the quarks are identical , in our limit , to the sm [ cols="<,<",options="header " , ] we have build our lagrangian based on an extension of vector meson dominance . another possibility for studying models with vector resonances is to consider the so called hidden local symmetry@xcite . in this case
, the model would correspond to a nonlinear sigma model based on the coset space @xmath32 .
the archetype of this kind of model applied to a strong electroweak model is the so called bess model @xcite ( which is based on @xmath33 and hence does not have a techniomega ) .
of course such a model does not include a higgs , nevertheless a `` linear '' version of the bess model exists @xcite which include scalars in the spectrum .
we use lanhep @xcite and comphep @xcite in order to compute the cross section of the associate production of a higgs and a gauge boson at the lhc and the tevatron ( run ii ) .
we choose to work with @xmath34 because for values of @xmath35 of this order , the vector resonances can be light ( i.e. @xmath36 gev ) while @xmath37 is still much bigger than one . as @xmath35 approaches to one ,
the vector resonances became too heavy and their observation is increasingly difficult . on the other hand , if @xmath35 is too small the coupling of the vector resonances to the sm fields are suppressed .
the decay width of the vector resonances were also computed with comphep . in the range of masses considered here ,
the techniomega and the charged technirho decay mainly into fermions while the neutral technirho decays mainly into a pair of @xmath38 s .
notice that @xmath39 and @xmath40 vertices do not exist in this model .
indeed , gauge invariance avoid the generation of such couplings from yang - mills kinetic terms .
for example , a @xmath40 vertex coming from yang - mills term would violate gauge symmetry in the process @xmath41 .
a discussion about this coupling in a similar model can be found in @xcite .
our aim in this work is to study the impact of the presence of vector resonances on the higgs production in association with a gauge boson . in particular , we want to compare the predictions of this scenario with the standard model .
for this reason we define the following quantity as a measure of the enhancement of the signal over the standard model : @xmath42 where @xmath43 is the cross section predicted by our effective lagrangian and @xmath44 is the cross section predicted by the standard model . in figure
[ fig : lhc_wh ] we show the value of @xmath45 as a function of the mass of the technirho ( @xmath46 ) for three values of the higgs mass ( @xmath47 gev ( solid line),@xmath48 gev ( dashed line ) and @xmath49 gev ( dotted line ) ) for the process @xmath50 at the lhc .
observe that in this case , the cross section is significatively enhanced with respect to the standard model when the technirho has a mass between 200 gev and 350 gev .
the variation of this enhancement as function of @xmath35 is shown in figure [ fig : alfa ] for @xmath51 gev and @xmath52 on the other hand , when a higgs and a @xmath31 are produced ( figure [ fig : lhc_zh ] ) , the cross section is less enhanced and we expect that this channel will not be sensible to the presence of the vector resonances
. we also compute the cross section of the process @xmath53 at the tevatron ( run ii ) .
the result is shown in figure [ fig : tev ] for @xmath51 gev .
notice that the enhancement in this case is comparable to the prediction for the lhc .
the point in the parameter space we use for studying our model was chosen in order to maximize the deviation from the standard model for the selected channel .
this procedure allows us to evaluate the possibility of testing the model .
let s now consider the restrictions imposed by data on the electroweak parameters @xmath54 and @xmath55 .
this is not the place for performing a detailed calculation of these parameters in our model , nevertheless we expect that the final result must be similar to the one obtained in the minimal bess model .
@xcite . written in the notation of the original authors ,
this result is : @xmath56 where @xmath57 are proportional to @xmath58 and @xmath59 respectively , @xmath60 is our coupling constant @xmath61 and @xmath62 corresponds to our parameter @xmath63 which represents a direct coupling of the proto - technirho to fermions . for the set of parameters used above we obtain @xmath64 which is disfavored by precision measurements .
nevertheless , we can be consistent with the constrains imposed by precision data by choosing @xmath65 . in this case ,
our results on @xmath66 are modified by a factor @xmath67 and an important enhancement remains in the channel @xmath50 for @xmath46 around @xmath68 gev . at this point
a word must be said about the direct search of our vector resonances and the mass limits imposed to them by current data .
in general the results of direct search , performed at the tevatron and lep , of the technirho and techniomega predicted in usual walking technicolor models applies to our case , except by the fact that we have suposed there are no physical technipions in the spectrum . in this case , considering the technirho decay to charged leptons and a pair of @xmath38 s , the technirho ( as well as the techniomega ) is excluded for @xmath69 gev@xcite .
we have constructed an effective lagrangian which represents the standard model with a light higgs bosons and vector resonances that mix with the gauge bosons .
we fixed the parameter of the model that connects the mass of the new vector bosons with their coupling constant , in such a way that the model were compatible with light resonances .
we want to emphasize that , although this effective lagrangian is inspired by a new technicolor scenario recently proposed in @xcite , we do not claim that the values chosen for our study represent the low energy limit of the specific models constructed there .
the most obvious process for searching differences between our model and the predictions of the standard model is the associate production of a higgs and a gauge boson .
we found that the most sensitive channel is the production of the higgs and @xmath38 . for a range of resonance s mass between @xmath49 gev and @xmath70
gev the enhancement of the cross section is significant at both , the lhc and the tevatron .
the author is grateful to rogrio rosenfeld for valuable comments for a recent review see c. t. hill and e. h. simmons , phys .
rept . * 381 * , 235 ( 2003 ) [ erratum - ibid . * 390 * , 553 ( 2004 ) ] [ arxiv : hep - ph/0203079 ] .
d. d. dietrich , f. sannino and k. tuominen , phys .
d * 72 * , 055001 ( 2005 ) [ arxiv : hep - ph/0505059 ] .
+ for a review of effective models of a strong electroweak symmetry breaking sector with scalar and vector resonances , see d. dominici , riv .
nuovo cim . *
20 * , 1 ( 1997 ) [ arxiv : hep - ph/9711385 ] . for a modern review on vector meson dominance see ,
h. b. oconnell , b. c. pearce , a. w. thomas and a. g. williams , prog .
phys . *
39 * , 201 ( 1997 ) [ arxiv : hep - ph/9501251 ] .
r. casalbuoni , s. de curtis , d. dominici and r. gatto , ; ; + r. casalbuoni , p. chiappetta , a. deandrea , d. dominici and r. gatto , ; + r. casalbuoni , p. chiappetta , s. de curtis , f. feruglio , r. gatto , b. mele and j. terron , ; + r. casalbuoni , p. chiappetta , m.c .
cousinou , s. de curtis , f. feruglio , r. gatto , ; + l. antichini , r .
casalbuoni and s. de curtis , .
a. v. semenov , arxiv : hep - ph/0208011 .
+ a. semenov .
lanhep - a package for automatic generation of feynman rules .
user s manual .
inp msu preprint 96 - 24/431 , moscow , 1996 ; hep - ph/9608488 + a. semenov . nucl.inst . & meth .
a393 ( 1997 ) p. 293 .
+ a. semenov .
lanhep - a package for automatic generation of feynman rules from the lagrangian .
updated version 1.3 .
inp msu preprint 98 - 2/503 . + home page : http://theory.sinp.msu.ru/ semenov / lanhep.html e.boos et al .
[ comphep collaboration ] , comphep 4.4 : automatic computations from lagrangians to events , nucl .
instrum . meth .
a534(2004 ) , p250 [ hep - ph/0403113 ] . +
comphep - a package for evaluation of feynman diagrams and integration over multi - particle phase space .
user s manual for version 3.3 , hep - ph/9908288 + home page : http://theory.sinp.msu.ru/comphep | motivated by new models of dynamical electroweak symmetry breaking that predict a light composite higgs boson , we build an effective lagrangian which describes the standard model ( with a light higgs ) and vector resonances .
we compute the cross section for the associate production of a higgs and a gauge boson .
for some values of model parameters we find that the cross section is significantly enhanced with respect to the standard model .
this enhancement is similar at the lhc and the tevatron for the same range of resonance mass . |
classically , wave motion in elastic media is a phenomenon associated with either / both the transverse or longitudinal vibrations ; if there is a wave , then something material should be waving .
this notion led 19th century scientists to introduce the concept of the luminiferous medium ( field , aether , etc . ) . the first attempt to explain the propagation of light as a field phenomenon was by cauchy circa 1827 ( see the account in @xcite ) , who postulated the existence of an elastic continuum through which light propagates as a shear wave . subsequently came the contributions of faraday and ampere , which eventually led to the formulation of the electromagnetic model as known today
the crucial advance in understanding the phenomena of electromagnetism phenomena was achieved , however , when @xcite added the ` displacement current ' , @xmath0 , to ampere s law .
this term was similar to the time - derivative of the stress in his constitutive relation for elastic gases @xcite .
since the electric field vector is a clear analogue of the stress vector in continuum mechanics , one can say that maxwell postulated an elastic constitutive relation by adding the displacement current to ampere s law . indeed , the new term transformed the system of equations ( already established at that time in electrostatics ) into a hyperbolic system with a characteristic speed of shear wave propagation .
maxwell and hertz identified the characteristic speed with the speed of light .
however , soon after maxwell formulated his equations , it was discovered that his model was not invariant with respect to translational motion of the coordinate frame .
@xcite realized that the cause of non - invariance was the use of partial time derivatives .
he proposed to use the convective time derivative _ in lieu _ of the former .
the maxwell
hertz equations ( called also ` progressive wave equations ' ) are clearly the correct model for electromagnetic phenomena in moving bodies .
but the primordial question is whether the progressive - wave equations can be construed to hold also _ in vacuo_. the answer is obviously in the affirmative if one accepts that what is currently called ` physical vacuum ' must be regarded as a material continuum .
voigt ( see @xcite ) , @xcite , and independently lorentz , spotted the fact that the wave equation can be made invariant if , in the moving frame , the time variable is changed together with the spatial variables .
nowadays , this is known as the lorentz transformation .
the success of the latter stems from the fact that it tacitly restores some parts of the convective derivative , i.e. it emulates the material invariance for non - deformable frames in rectilinear motion @xcite .
although researchers usually speak about lorentz covariance as a general covariance , it has to be pointed out that the lorentz transformation has no meaning for accelerating frames ( nor for generally deforming frames ) .
hence the search for the truly covariant formulation should continue .
the present paper summarizes the efforts from the last decade and a half towards identifying the mechanical construct behind the phenomenon of what is termed ` electromagnetic field ' .
we show here that the maxwell equations and the laws of electromagnetism ( biot savart , oersted ampere , and lorentz force laws ) , all can be derived from the governing equations of maxwell s elastic fluid , that includes relaxation of the stress . in doing so
, we also present a concise frame indifferent formulation of the maxwell elastic fluid model .
the linearized governing equations of elastic medium are valid only for infinitesimal deformations , when the referential description @xmath1 and spatial configuration @xmath2 coincide .
the linear constitutive relationship for an elastic body relates the stress tensor @xmath3 ( we reserve the notation @xmath4 for the deviatoric stress tensor ) to the deformation tensor @xmath5 via the generalized hooke s law @xmath6,\ ] ] where @xmath7 is the trace of the deformation tensor , and the superscript @xmath8 denotes the transpose .
the above constitutive law yields the so - called navier equations ( see , e.g. , ( * ? ? ?
117 ) for the displacement vector @xmath9 : @xmath10 where @xmath11 and @xmath12 are the lam coefficients . here
one can use the ` nabla ' operator @xmath13 because the equations are written in the current configuration .
the speeds of propagation of shear and compressional disturbances , are given respectively by @xmath14 where the ratio @xmath15 is introduced for convenience . in a compressible elastic medium
, both the shear and the dilational / compressional waves should be observable . since the groundbreaking works of young and fresnel
, it is well established that electromagnetic waves ( e.g. , light ) are a purely transverse ( shear ) phenomenon .
this observation requires us to reduce the complexity of the model and to find a way to eliminate the dilational modulus @xmath11 .
cauchy assumed that @xmath16 and ended up with the theory of so - called ` volatile aether ' ( see @xcite ) . upon a closer examination
, we found that such an approach can not explain maxwell s equations .
let us now assume that the dilational waves are not observable because the other extreme situation is at hand : @xmath17 which is equivalent to @xmath18 or @xmath19 .
it is convenient to rewrite eq . in terms of the speed of light , @xmath20 , and parameter @xmath15 , namely @xmath21 and to expand the speed of light @xmath20 , displacement @xmath22 , and velocity @xmath23 into asymptotic power series with respect to @xmath15 , namely @xmath24 introducing into and combining the terms with like powers we obtain for the first two terms @xmath25 where @xmath26 is introduced for convenience and is the coefficient of the spherical part of the internal stresses .
the variable @xmath26 has dimension of @xmath27 and plays the same role as the pressure in an incompressible medium : @xmath26 is an implicit function in eq . that provides the necessary degree of freedom to enforce the satisfaction of the ` incompressibility ' condition , eq .. the latter can also be rewritten as @xmath28 which requires that the velocity field be solenoidal within the zeroth - order of approximation of the small parameter @xmath15 . from now on ,
the subscript ` @xmath29 ' will be omitted form the variables without fear of confusion .
now , eq . can be rewritten as @xmath30 where @xmath31 is the tangential part of the stress vector in the elastic continuum .
the normal part of the stress vector is given by the pressure gradient @xmath32 .
let us now introduce the following notations [ eq : elmag_definitions ] @xmath33 and call them the ` electric field ' and ` magnetic induction ' vectors .
thus the electric field is the negative tangential stress vector , while magnetic field is the ` dynamic ' vorticity .
naturally , @xmath34 is the ` kinematic ' vorticity . in the virtue of the above definition
, one has the following equation for the magnetic field @xmath35 now taking the @xmath36 of eq . and acknowledging the definitions given in eqs . , we arrive at faraday s law @xmath37 on the other hand , taking the time derivative of eq .
, we obtain the second of the dynamic maxwell s equations @xmath38 the fact that the governing equations of any elastic medium in the linear limit admit a maxwell form also can be considered as an indication that the electromagnetic field is in itself an elastic medium . in
what follows , we shall call the mechanical object equivalent mathematically to the electromagnetic field the _ metacontinuum _ , to distinguish it from continuous media in technical applications , such as fluids and elastic solids . note that the inverse of the elastic shear coefficient plays the role of the electric permittivity _ in vacuo _ , while the density of the metacontinuum acts as the ` magnetic permeability . '
the results of this section unequivocally show that the ` field ' described by maxwell s equations is equivalent to an elastic material .
to best of author s knowledge , the connection of maxwell s equation to the equations governing elastic media was first established in @xcite .
the common theme of these earlier papers is that the metacontinuum is considered as an elastic _ solid_. in such a model , no _ stationary _ magnetic fields can exist , since no steady velocities are possible for a solid continuum without discontinuities .
in this section we outline the next decisive step in developing the model : we consider an elastic liquid _ in lieu _ of an elastic solid .
this means that for shear deformations , the metacontinuum must be an elastic _ fluid _ for which the time derivative of the deviatoric stress tensor is related to the deviatoric rate of deformation tensor @xmath39 via the relation @xmath40 this constituitve relationship ( rheology ) can be rewritten for the negative stress vector @xmath41 and deformation vector , @xmath42 ) , since both of these vectors are the divergences of the respective tensors involved in the elastic rheology , eq . .
then @xmath43 where @xmath44 is called ` elastic viscosity ' by @xcite , @xmath45 is the relaxation time of the stresses , and the apparent elastic shear modulus is given by @xmath46 .
note that the above elastic - liquid rheology concerns just the shear deformations . for compressional / dilational motions
, the metacontinuum can still behave as a virtually incompressible solid .
a more general formulation of the shear part of the constitutive relation would be as in the viscoelastic liquid @xmath47 where @xmath48 can be called the ` conductivity ' of the viscoelastic liquid .
note that in @xcite , the conductivity is set to unity , which precludes treating purely elastic ( non - viscous ) liquids . here
, we prefer to keep the flexibility offered by the presence of the coefficient @xmath48 .
setting the appropriate terminology is an uneasy task because , when @xmath49 , then @xmath44 does indeed have a meaning of a viscosity coefficient , while for @xmath50 , it loses its independent meaning and enters the picture through the coefficient of apparent shear elasticity @xmath12 . the case @xmath51 gives a viscoelastic rheology , but it does _ not _ lead to a model governed by the navier - stokes equations with additional elasticity , because no retardation term ( time derivative of the deformation tensor / vector ) is present . in this sense , adding the conductivity does not introduce dispersive dissipation , but rather a linear attenuation parameterized by the conductivity coefficient . for the effects connected with the attenuation /
conductivity we refer the reader to @xcite and @xcite . the constitutive relationship given in eq .
can be interpreted as ohm s law for _
vacuo_. although , this stipulation is made in mainstream texts ( e.g. , @xcite ) , the cause of ohm s law in matter is not necessarily the intrinsic resistance of the metacontinuum : it is the result of the thermal fluctuations of the atoms that obstruct the free passage of charges through a conductor .
clearly , a more in - depth argument is needed to justify having ohm s law in _
vacuo _ , which goes beyond the scope of the present paper .
it should be pointed out here that eq .
is concerned merely with the rheology for the shear motions . at this point
it is not of importance if for dilational / compressional motions the metacontinuum is solid or liquid , provided that the dilational coefficient is much larger than the shear coefficient .
then , @xmath52 the intuitive argument is that if the metacontinuum is a liquid with respect to the dilational motions , it may lose its integrity during the motion . since there is no information on the electromagnetic field being ` ruptured , ' we are guided by the above argument and assume that the solid rheology applies to the dilational / compressional motions .
note that in the above notations , @xmath53 is the dilational viscosity coefficient , and @xmath44 is the shear viscosity coefficient defined in eq .
@xmath54 .
the system , @xmath55 specify the combined constitutive relation for the metacontinuum ( electromagnetic field ) .
the closest analogy to an ubiquitous continuous medium is that of jelly or pine pitch .
if compression / dilation waves are sent through the metacontinuum , it behaves as a elastic solid with very large dilational modulus , while if a shear deformation is applied , it flows as a incompressible liquid .
in terms of the velocity vector , the cauchy balance can be rewritten as follows : @xmath56 where the left - hand side is the material ( convective ) derivative of the velocity vector @xmath57 in the current configuration ( called ` convective ' or ` total ' derivative ) .
remember that in the referential description , it is just the partial time derivative . note also that for an incompressible metacontinuum , the density is the same constant in both the referential and spatial descriptions and we denote it by @xmath58 .
the concept of frame indifference ( general covariance of the system ) requires that the partial derivative of the stress variable ( in our case the stress vector @xmath59 ) in eq .
is replaced by the appropriate invariant rate . since it is a vector density ,
see the argument by @xcite , the rate has to leave the integral of the stress vector invariant .
it is argued in @xcite that the pertinent invariant rate is the so - called oldroyd upper - convected derivative , namely @xmath60 here we come to one of the most crucial assumptions of the present work , namely , the way the constitutive relation has to be written when a relaxation of the stress is present .
it is usually assumed that the invariant rate to be added to the constitutive law of viscous liquids should be of the stress tensor , i.e. the upper - convected oldroyd derivative ( @xcite ) : @xmath61 the problem with this constitutive conjecture is that the tensor @xmath4 does not play an independent role in the cauchy balance equation .
rather , the deviatoric stress vector @xmath62 appears there .
then , does eq . ensure that the time rate of @xmath63 is invariant ? to find out , we take the operation div of eq . , arriving at @xmath64 which differs from the invariant time rate of @xmath63 by the term @xmath65 , the latter being the contraction of the third rank tensor of the repeated gradient ( the hessian ) of @xmath57 and the second rank stress tensor @xmath4 . in order to establish which constitutive relation
is correct ( the one involving the stress tensor or stress vector ) , one has to have data for flows in which the hessian of the velocity vector field is measured .
while for the case of elastic liquids it is still possible to devise such an experiment , electromagnetism clearly indicates that the constitutive relation at play is the one involving the stress vector .
here we propose an alternative formulation for models involving stress relaxation by replacing the partial time derivative of the stress vector with the invariant rate @xmath66 which was proposed in @xcite , and then successfully applied to the generalization of the maxwell cattaneo model of second sound in @xcite .
apart from providing insight into the possible constitution of the electromagnetic field , the above model has practical significance for the theory of maxwell elastic fluids .
note that we add the condition of incompressibility ( recall eq .
with the subscript 0 " omitted ) : @xmath67 collectively , we can term the system eqs . , and as the ` frame indifferent incompressible maxwell fluid model ' .
the alternative formulation of the constitutive relation based on the stress vector concept proposed here has a very important consequence : it allows the stress vector @xmath59 to be eliminated between eqs . and ( when @xmath68 ) , to obtain a system that does not contain the stress variable : @xmath69 where the following notations are used @xmath70 note that @xmath71 is the upper convected derivative of a vector defined in eq . .
. contains the implicit function @xmath26 that has to ensure the satisfaction of the incompressibility condition eq . .
the advantage here is that , the only unknown functions are @xmath57 and @xmath26 , and the hyperbolicity of the model is now easily seen .
according to the principal of frame indifference @xcite , the laws of physics ( including the laws of continuum physics ) must have the same form in any reference frame ( coordinate system ) .
unlike what is called ` lorentz covariance ' , the laws in the referential description are _ frame indifferent _ , i.e. they are truly covariant . however , experimental measurements are _ always _ connected with a current frame in the geometric space .
this means that an observational frame can not detect the material variables ( the referential description ) , but rather can merely measure their counterparts in the current ( geometric ) frame .
this is a typical situation in mechanics of continuous media where the reference configuration is often not related to any measurable frame . for this reason
we need to reformulate the model from section [ sec : maxwell ] in the current description making use of euler variables .
this is the objective of the present section . the cauchy balance , eq .
, can be rewritten in the so - called ` gromeka lamb form ' @xcite : @xmath72 now , taking the _ curl _ of eq . , and using our definitions eq . , we get : @xmath73 = - { \frac{\partial \bm{b}}{\partial t } } , \label{eq : faraday_lorentz}\ ] ] which is faraday s law with an additional term representing the force exerted by a moving magnetic field on each point of the medium
. it is induced by the convective part of the acceleration at that point .
the reaction to this force is the force acting on a moving point ( charge ) , @xmath74 , known as the ` lorentz force ' . in other words ,
the material invariant version of faraday s law presented here automatically accounts for the physical mechanism that causes the lorentz force .
the latter is nothing more than the inertial force given by the convective part of the total derivative .
this is a very important result because it tells us that the lorentz force is not an additional , empirically observed force that has to be grafted onto maxwell s equations , but rather , it is connected to the material time derivative , specifical , to its convective part . under the incompressibility condition ,
eq . can be recast as @xmath75 which we can call the ` hertz form ' of the faraday lorentz law . note the presence of the third term on the left - hand - side .
it is not in hertz s formulation , nor does it appear in @xcite .
evidently , eq . does not give any special advantage over eq .
, but the form of eq . shows that one can not just add the convective part of the derivative to faraday s law to make it frame indifferent .
this follows from the fact that the magnetic field is not a primary variable ( primary variables in fluids are the velocity components and the pressure ) .
rather , it is proportional to the @xmath36 of the velocity vector , @xmath57 . by using the vector identity ( see , e.g. , ( * ? ? ?
180 ) @xmath76 eq .
yields the following generalization of the second of the dynamical maxwell s equation : @xmath77 where the definition given in eq .
is already acknowledged .
now , eqs .
, , , form a system which can be termed the _ frame indifferent electromagnetodynamics _ ( fiem ) .
this system generalizes hertz s program from 1890 and _ rigorously _ fulfills the requirements for ` general covariance , ' because fiem is frame - indifferent ; it is invariant when changing to another coordinate frame that can accelerate and can even deform .
a very simple , limiting case of frame indifference is that of galilean invariance . the vector of absolute velocity @xmath57 is the primary variable , but the absolute velocity in the referential description
can not be measured ( only relative velocities can be observed ) . in principle
it can be restored from the magnetic field eq .
, provided a boundary condition is known .
clearly , in the limit of small velocities the convective and convected terms can be neglected and in the limit one obtains maxwell s system .
a remarkable feature of eq .
is that it incorporates terms ( with under - braces ) that , collectively , form the progenitor of the biot savart law . indeed , setting aside the possible singularities connected to point charges , then we can neglect the term with @xmath78 .
let us also consider the case of stationary electric field , @xmath79 , and no resistance @xmath80 . then , considering a surface @xmath81 inside a closed contour @xmath82 , we can integrate eq . over @xmath82 and use the stokes theorem to obtain @xmath83\operatorname{d\!}s
= 0,\ ] ] where @xmath84 is an areal element on the surface @xmath81 .
since the surface @xmath81 is arbitrary , then the only possibility is that the integrand must be equal to zero .
thus one arrives at the form of the biot
savart law as it is stipulated in relativistic electrodynamics @xcite , @xmath85 where @xmath57 is the velocity of the point of the field at which @xmath86 is measured .
the underlined terms in eq . give that in _ vacuo _ @xmath87 which is discussed in ( * ? ? ?
* ch.15 ) in dimensional form .
this can be interpreted as a combination of ohm s and ampere s laws of electromagnetism in _ vacuo_. the terms with hats over them in eq .
give as a corollary the ampere law if the following definition of a charge in _ vacuo _ is introduced : @xmath88 in particular , the ` chargedness ' , @xmath89 , of the displacement / velocity field is defined as the divergence of the electric field at the point . in order not to confuse this property of the field _ in vacuo _ with the localized pattern , called electron or a proton , we call the above defined function @xmath89 the _ metacharge_. for the latter , a continuity equation is readily derived upon applying the operation @xmath90 to eq . , namely @xmath91
which , for the case @xmath50 , is the standard continuity equation for charge . here , is to be mentioned that eq
was derived by @xcite directly from the oldroyd form , eq . , but the derivation here is much more straightforward because of the application of the identity eq . .
we shall refer to eq . as the continuity equation for the _
metacharge_. in terms of the above introduced _
_ metacharge _ _ , we obtain @xmath92 which can be called ampere s law in _ vacuo_. the important conclusion form
the frame - indifferent formulation of the displacement current is that similarly to the lorentz force law , the convective / convected terms are related to phenomena that are embodied in ampere s and biot
savart s laws , thus unifying them with maxwell s electrodynamics .
all three electromagnetic - force laws ( called alternatively the ` laws of motional electromagnetism ' ) are manifestations of the inertial forces in the metacontinuum .
@xcite put forward the idea that both of these laws may actually follow from a single law , similar to what is presented in eq . .
a debate is still ongoing in the literature about wether these two laws are identical ( see , e.g. , @xcite ) or independent ( see , e.g. , @xcite ) .
our results seem to favor whittaker s original idea that both laws have to be interwowen in the correct formulation . in our work
they are merely the corollaries form the inertial terms embodied in the convected derivative .
the frame indifferent model of the electromagnetic field ( called here the metacontinuum ) , succeeds in unifying , in a single nexus , all known phenomena of electromagnetism : faraday s law , displacement - current law , lorentz - force law , oersted
ampere s law , and biot savart law .
it is a significant step forward from maxwell s model , in which only the first two were explained by the equations themselves , and the latter three appeared as additional empirically observed relations between the main characteristics of the field . since in our model these new terms are valid _ in vacuo _ , they are clearly the progenitors of the related phenomena in moving media .
as shown above , the model that yields maxwell s equations is one of an incompressible elastic fluid . the obvious way to extend the validity of the model is to assume that the fluid is compressible .
this means that it has to be a medium with vanishing compressibility , which happens when the dilational elasticity or viscosity coefficient is extremely large in comparison with the shear viscosity coefficient .
as already mentioned , at this stage it is not clear whether the medium behaves as a liquid or solid when the compressional / dilational motions are considered .
this question can not be answered without staging an experiment in which the fabric of the metacontinuum could eventually be ` ripped ' in order to settle this question . actually ,
if the compressional / dilational motion is oscillatory , then it does not really make much difference if the metacontinuum is a solid or a fluid . as shown in the precedence , it has to be a fluid under shearing , but this does not impose any restriction on its behavior in the oscillatory compression / dilation motions .
now , combining the two parts of the constitutive relation as given by eqs .
the momentum equation can be written as @xmath93 which is to replace eq . .
. remains unchanged , while the incompressibility condition , eq .
, has to be replaced by the following : @xmath94 alternatively , one can use the continuity equation in the from @xmath95 ( see @xcite ) , where @xmath96 is the determinant of the gradient of deformation tensor and @xmath97 is the _ constant _ density in the referential description . note that , the form chosen above , is consistent with the eulerian description . note also that if the solid rheology is assumed in eq . , then one has to add the defining equation for the velocity components : @xmath98 , eqs . , , and form a coupled system for the compressible metacontinuum , which is highly nonlinear ( even for the linear rheology ) because of the dependence of the density @xmath58 on the motion .
there are no conceptual difficulties to limit the model to the case of slight compressibility of the metacontinuum , and look for the propagation of harmonic compression waves therein .
this raises the question about speed of the compressional waves ( ` sound ' ) of the metacontinuum . in order to avoid ambiguous terminology we will not use the term _ sound _ when referring to these waves .
rather we will borrow a coinage from the ancient school of stoa ( see @xcite ) and will call the compressional / dilational motions the _ pneuma_. one obvious implication of the existence of waves of a different kind than the electromagnetic ( shear ) waves is that , in fact , there is more energy in the physical vacuum than that detected from electromagnetic interactions . mechanically speaking ,
the _ pneuma _ waves are ` orthogonal ' to the shear ( em ) waves and they may not be detectable by devices based on electromagnetic interactions . indeed , they perfectly fit the bill of what is currently called ` dark energy ' ( see , e.g. , @xcite ) .
one can begin thinking of how to detect pneuma waves only after some solutions for coupled compressional and shear waves of the system eqs .
, , and become available .
let us mention in the conclusion of this section that the above derived system is suitable for any ( visco)elastic medium , such as the ager - gelatin base phantom ( see @xcite ) .
the approach proposed here presents a self - consistent point of view based on the continuum mechanics of the electromagnetic field whose shear deformations are perceived as the phenomena of electromagnetism .
it is shown that the linearized governing equations of any incompressible elastic medium admit a ` maxwell form ' .
conversely , the maxwell equations of electrodynamics are a _ strict _ mathematical corollary of the linearized governing equations of the incompressible maxwell ( visco)elastic fluid .
this idea is further elaborated upon by deriving the frame indifferent formulation of the model of elastic fluids .
it is argued that in some cases ( as in the formulation of electromagnetism considered here and many other technologically significant applications ) the constitutive relation can be written for the stress vector rather than for the stress tensor , because what actually enters the momentum equations is the point - wise stress vector , which is the divergence of the stress tensor . from the frame indifferent governing system of elastic fluids , a ` maxwell form '
is derived that includes the terms of the original maxwell equations along with terms stemming from the convective and convected invariant time rates .
these inertial terms are the progenitors of the so - called laws of ` motional electrodynamics ' : biot savart , oersted ampere , and lorentz force laws .
the latter are usually assumed as additional empirical hypotheses to maxwell s equations .
this makes for a unified model of electromagnetism , which is truly covariant , by virtue of the fact that they are frame indifferent . in other words ,
the new model is invariant to changes to frames that accelerate and deform .
the continuum - mechanics formulation of the electromagnetic field proposed here opens a new avenue of research connected with the possible compressibility of elastic fluids .
consequently two kinds of linear propagating waves can co - exist : shear waves ( light , when in the visible spectrum ) and compressional waves ( called _ pneuma _ in this paper ) .
it must be stressed that the speed of compressional waves is necessarily much larger than the speed of light . as a consequence , the metacontinuum appears virtually incompressible to observes using tools designed to detect its shear motions ( i.e , electromagnetic phenomena ) .
32 natexlab#1#1url # 1`#1`urlprefix[2][]#2 [ 2]#2 , , , . .
, , , , . . ,
. . , . , . , in : , . pp . .
, . , in : ( ed . ) , , ,
, . , a. . , . , b. . , . , .
. , . . , . ,
. , . . , . ,
. . , . , ,
. . , . , , ,
. , . . , . , | we show that the linearized equations of the incompressible elastic medium admit a ` maxwell form ' in which the shear component of the stress vector plays the role of the electric field , and the vorticity plays the role of the magnetic field .
conversely , the set of dynamic maxwell equations are strict mathematical corollaries from the governing equations of the incompressible elastic medium .
this suggests that the nature of ` electromagnetic field ' may actually be related to an elastic continuous medium .
the analogy is complete if the medium is assumed to behave as fluid in shear motions , while it may still behave as elastic solid under compressional motions .
then the governing equations of the elastic fluid are re - derived in the eulerian frame by replacing the partial time derivatives by the properly invariant ( frame indifferent ) time rates .
the ` maxwell from ' of the frame indifferent formulation gives the frame indifferent system that is to replace the maxwell system .
this new system includes terms already present in the classical maxwell equations , alongside terms that are the progenitors of the biot
savart , oersted
ampere s , and lorentz force laws .
thus , a frame indifferent ( truly covariant ) formulation of electromagnetism is achieved from a single postulate : the electromagnetic field is a kind of elastic ( partly liquid , partly solid ) continuum .
frame indifference , maxwell s elastic fluid , maxwell s equations of electrodynamics , lorentz force , biot savart law |
the x - ray transient xte j0421 + 560 was discovered on mar .
31 , 1998 with all sky monitor aboard rxte . according to asm data
the flux from the source in 1 - 12 kev energy range was rising quickly from 40 mcrab mar .
31.36 to 1880 mcrab apr .
1.04(smith&remillard 1998 ) .
pca and hexte observations started apr .
1.08 , 1998 and confirmed the asm results ( marshall&strohmayer , 1998 ) .
the new x - ray source was localized by pca with coordinates : ra=@xmath0 , dec=@xmath1 ( marshall&strohmayer,1998b ) .
the lightcurve of the source was characterized by extremely fast rise up to the maximum with subsequent , also unusually fast decay of the flux .
the observations by asca and bepposax satellites showed the presence of the strong emission line around 6.7 kev .
the radio observations of apr . 1
, 1998 revealed the bright radio source at the location r.a.=
@xmath2 + /-
@xmath3 , dec .
= @xmath4 + /-
@xmath5 , 2000 equinox ( hjellming&mioduszewski , 1998a ) .
the radio source , which position is coincident with that of symbiotic star ci cam ( wagner&starrfield , 1998 ) was proposed to be the radio counterpart of x - ray transient xte j0421 + 560 .
the radio images obtained on apr 5 and 6 , 1998 revealed that the ci cam radio source had become resolved .
the vla images showed the extended radio emission that had the appearance of a roughly symmetrical s - shaped twin - jet , strikingly similar to the radio jets of ss 433 ( hjellming&mioduszewski , 1998c ) .
the outermost pair of emission components was separated by @xmath6 .
assuming the expansion started on mar .
31.6 , 1998 the proper motion of the components is 26 mas / day .
assuming a distance of 1 kpc ( chhikvadze 1970 ) , this corresponds to apparent tangential velocity @xmath7 .
optical spectroscopy of ci cam showed the strong emission lines of h , he i , fe ii .
however , none of the lines showed double peak profile , which is typical for the accretion disk emission in x - ray binaries and cataclysmic variables .
these results are consistent with ci cam / xte j0421 + 560 , but suggest that it might be unusual x - ray symbiotic star(garsia et al . , 1998 ) .
in the present work we used the data obtained by rossi x - ray timing explorer ( bradt , swank&rothschild , 1993 ) .
the data was retrieved from the xte gof at gsfc .
the asm light curves were used as provided by the xte gof .
the rxte has a payload of two coaligned spectrometers pca and hexte that provides the broad band spectral coverage for energy range from 3 to @xmath8 kev and all sky monitor , that give a possibility to follow the long term behavior of the source with nearly complete time coverage within 1 - 12 kev energy band .
the pca and hexte data was treated with standard ftools v.4.1.1 package tasks . to estimate the pca background we run the latest version of task _ pcabackest _
v.2.0 ( stark 1998 ) .
we applied new @xmath9 model to the observations , when the source was extremely weak ( apr .
89 , 1998 ) , and vle based background model for others . in our spectral analysis we used pca rmf matrix v.3.3 ( jahoda 1998 , see also the @xmath10 jahoda 1998a ) .
the analysis of the crab nebula spectra had confirmed , that the systematic uncertainties of the matrix are less than 1% .
we would like to note , that pca effective area is substantially decreased after @xmath1134 kev to the lower energies , so the estimation of @xmath12 value with pca should be treated with care . to account
roughly for the uncertainty in the knowledge of the spectral response , 1% of systematic error had been added to the statistical errors for each pca channel .
all spectra had been corrected for deadtime according to zhang&jahoda , 1996 . for the spectral analysis of the hexte data the version 2.6 ( released mar .
20 , 1997 ) of the response matrix had been taken .
the background for each cluster of hexte detectors was estimated using the off - source observations .
only the data above 15 kev were taken into account because of the uncertainties of the response matrix below this energy . at the high energy end the spectrum was cut at @xmath13 kev depending on the brightness of the source in order to avoid significant influence from the background subtraction uncertainty .
the deadtime correction had been performed using hxtdead ftools task for all observations .
the brief information about the rxte observations of the source are presented in table [ obslog ] .
the lightcurve of the source in various energy bands is shown in fig .
[ lcurve ] .
the peculiar point in this figure is that the asm data in the lowest energy band 1.3 - 3.0 kev show the feature that looks like secondary peak around the 5th day after the beginning of the outburst ( fig.[lcurve]a ) .
similar `` kick '' feature is frequent for the lightcurves of the x - ray novae in standard x - ray band , but in our case the feature is seen only in the soft x - ray band ( @xmath14 3 kev ) while the flux at higher energies is decreasing monotonically .
note , that flaring in softer energy band was observed also by asca observatory around the same date ( ueda et al .
, 1998b ) in the lower panels the hardness ratios of the source are presented .
the flux decay during the period apr.1 - 3 in the total asm energy band can be approximated with the exponential law with e - fold time @xmath110.6 day , later on the e - fold time increases to @xmath111.1 day .
the example of the source flux decay during the first session of observation by the pca and hexte is shown in the fig.[pca_hexte ] . during this observation
e - fold times were @xmath110.57 day for the pca energy band and @xmath110.28 day for the hexte energy band .
the source spectrum in the 3100 kev energy band can by approximated by a power law with the exponential cutoff at higher energies : @xmath15 .
extremely strong emission line is present in the spectrum at energies around 6.46.7 kev .
similar spectral shape was found by asca and bepposax observatories for their observations ( ueda et al . , 1998 , orlandini et al . , 1998 ) .
however , for the pca spectra this model leaves the significant residual at the energies around 8 kev with the amplitude ( @xmath1146% , see fig.[8kev ] ) , which is sufficiently higher than the response matrix uncertainty . because of the presence of very bright emission line at the energies 6.46.7 kev and low energy resolution of pca in this energy band ( @xmath111 kev ) , the quality of the data approximation at the energies @xmath118 kev will strongly depend on the accuracy of used photon redistribution matrix ( definite part of the response matrix ) .
the analysis , carried out with collaboration of keith jahoda ( rxte / pca team ) have showed , that the unaccuracy of the matrix is not worse than @xmath111% and the observed excess around 8 kev should be considered as real intrinsic spectral feature of the source .
an addition of the emission line at energy @xmath118.1 kev significantly improves the quality of the data approximation in terms of @xmath16 statistics
@xmath16 value reduces by @xmath1140 with the systematic uncertainties included and by @xmath119000 with statistical errors only . as a result , the best fit analytical model consists of a power law with high energy exponential cutoff , low energy absorption , and two emission lines .
the parameters of this model for different sessions are given in tables [ cutoff ] and [ lines ] .
we would like to note , that the spectral results of pca with response matrix 3.3 have relative uncertainties in energy around 0.02 kev at the energies 68 kev , but the uncertainties in absolute values is slightly higher , it can be estimated to be @xmath17 kev ( jahoda 1998 , private communication ) .
one should bear this in mind , while tring to localize the lines in the spectrum .
combined spectra of pca and hexte experiments for three different observations are presented in fig.4 ( in the fig.4 @xmath18 model was used to approximate the line emission , see below ) .
the spectra can be approximated satisfactorily by the aforementioned model .
note , hovewer , that because of calibration uncertainties the photon indexes of the crab spectra of pca and hexte have slightly different values ( on @xmath11 0.050.1 ) , therefore some deviations in spectral shapes between two experiments can be expected .
due to the decrease of the flux and the softening of the spectrum from xte j0421 + 560 the source had not been detectable by hexte after apr.4 , 1998 .
the shape of the spectra is close to the spectrum of the emission of optically thin plasma ( @xmath19 ) , but can not be satisfactorily fitted by this model with single temperature .
possible interpretations of this fact could be fast change of the temperature affecting the intergral spectrum or non - equilibrium state of the emitting plasma .
it is worth to note several features of the obtained spectra : 1 ) the absorbtion value @xmath12 , that characterizes the low energy cutoff in the spectra is changing from @xmath20 on apr.1.04 , @xmath21 apr.1.35 , to @xmath22 the day after and then became insignificant ; 2 ) the slope of the power law fit increases and the cutoff energy decreases with time , which shows the cooling of the emission medium ; 3 ) the center of the broad emission line 6.56.7 kev fluctuates to the higher energies .
( see fig .
[ 6kev ] and [ lines ] ) .
below we discuss the possible interpretation of these and other observational results .
the x - ray transient source xte j0421 + 560/ci cam has several featues that make it different from the other known x - ray transients . the most striking difference , which has been repeatedly mentioned in the related publications , is an extremely fast evolution .
mean value of e - folding decay time for know x - ray transient is @xmath11 30 days ( see wan chen et al . 1997 ) , whereas for xte j0421 + 560 it is @xmath11 0.51 day .
possibly , the main difference is the nature of the compact object in these systems .
ci cam is known to be symbiotic binary , i.e. the binary which composed by red giant and an accreting white dwarf .
the most probable cause of the flares in these systems is considered to be the thermonuclear outburst on the surface of the white dwarf .
unstationary accretion in the system is proposed as an alternative mehanism for the flares generation ( see e.g , review of the models by mikolajewska&kenyon 1992 ) .
it should be noted that the x - ray observations of xte j0421 + 560 do not demonstrate the features , which is usually attributed to the presence of the accretion disk in the system , qpo , or characteristic very low frequency noise in the power spectrum , or typical soft component in the energy spectrum .
moreover , the analysis of the light curve of the source on the various time scales ( from hundreds of microseconds up to hundreds and thousands of seconds ) do not reveal the significant variability of the source flux different from the main trend of the flux decay .
this result is consistent with observations of asca and sax ( ueda et al .
, 1998b , frontera et al.,1998 ) .
hovewer , according to sax and asca data x - ray flux in the 0.5 - 1 kev energy band do has variability on the scales of hundreds of seconds and hours .
the character of the source flux variability could be an evidence that the soft and hard x - ray components are forming in the geometrically different parts of the system .
the dimension of the region , where the hard component is formed , should be larger than @xmath23 cm , here @xmath24 the sound velocity in the medium with @xmath25 kev , while the region , where the soft component is formed , must be less than @xmath26 cm in size . as it has been mentioned above , the best fit model for continuum spectrum is the power law with high energy cutoff at the energies @xmath11513 kev .
the spectrum of this type can be formed in the cloud of optically thin hot plasma , that has non - uniform temperature distribution . it should be noted that this form of the spectra is completely different from that of the many other x - ray novae - black hole candidates , where x - ray emission was detected up to hundreds kev ( see e.g. , sunyaev et al . , 1994 ; tanaka&shibazaki , 1997 ) .
the spectra , obtained during the first two days after the beginning of the outburst demonstrate the presense of significant low energy absorption .
the @xmath12 value was increased from @xmath27 on apr.1.04 , 1998 ( model dependent value ) to @xmath28 on apr . 1.35 , 1998 and then decreased to @xmath29 one day later and
had become insignificant during later observations .
it is quite natural to interpret that at the beginning of the outburst the x - ray source had been embedded within dense and cold cloud , which later on became less dense and , consequently , more transparent as a result of mass loss and/or temperature heating .
lightcirves of the object , obtained in x - ray , optical and radio bands may also be considered as an evidence of the cloud density decrease with time(frontera et al.,1998 ) .
the radio observations of the relativistic jet structures ( hjellming&mioduszewski , 1998c ) supports an idea of significant outflow of the matter from the central object of the system xte j0421 + 560 during the outburst .
it is interesting to estimate the dimensions and the density of the cloud assuming its cylindrical shape .
let us assume , that the cloud forms a cylinder with radius @xmath30 , length @xmath31 expanding along the cylinder axis , perpendicular to the line of sight ( see below the discussion of orientation of the jets ) .
we accept for simplicity that the cloud has the uniform distribution of the density .
the low energy absorption observed can be caused either by bremsstrahlung self absorption or by photoabsorption in the cool outer region of the cloud . with soft x - ray instrument in hand
it would be possible to distinguish between these two possibilities because of its different influence on the source spectrum .
unfortunately , pca do not allows to do it .
if the low energy cutoff in the spetra of xte j0421 + 560 is due to photoabsorption , then we can make some simple estimations on the cloud dimensions .
the value of photoabsorption can be approximated as @xmath32 , where @xmath33 part of the cloud which contribute to the @xmath12 value , @xmath34 density of the cloud .
emission measure in the cloud @xmath35 is given by : @xmath36 then we obtain : @xmath37 taking into account that this value is just order - of - magnitude evaluation of cloud length we can say , that our estimation is consistent with the observed expansion rate of the cloud in this direction @xmath38 cm / hour if @xmath39 . it should be noted , that if the radiation goes through the cold surrounding medium , the weak fluorescent line of fe at the energy @xmath40 kev should be formed ( see the discussion of this item below and also in ueda et al . , 1998b ) .
we emphasize that the discussed model should not be considered as ultimate , but as one of the possible ones . under assumptions
that there is no energy supply to the cloud after the maximum of the lightcurve is reached , and that to the moment of sufficient flux decrease ( after @xmath1134 days after the peak ) the cloud have lost a significant part of its energy , we can estimate the density of the cloud .
let us take the characteristic time of the flare to be @xmath4112 days ( @xmath42 s ) . from the energy balance of the cloud
( see also ueda et al .
, 1998b ) : @xmath43 and @xmath44 accepting this value of the density , we can evaluate the transverse size of the cloud @xmath45 cm and total mass of the erupted matter @xmath46g@xmath47 .
the obtained estimated value of @xmath30 is consistent with the free expansion of the cloud in this direction @xmath48 ( @xmath24 the sound velocity of the gas with the temperature @xmath49 kev ) .
the most likely , that the observed lines in the spectrum of xte j0421 + 560 has been generated in the optically thin emission of the hot plasma cloud .
alternative mechanisms of line formation the propagation though the medium or the reflection is less probable , because of huge equivalent width of the observed line . besides
, if the propagation or the reflection would be responsible for the line formation then the strong absorption edge ( ew@xmath50 ev ) would be formed at energies above the line .
the lack of such a feature moves us to believe that both the continuum spectrum , that extends up to the 100 kev and the observed emission features are formed in the hot optically thin plasma cloud .
however , the pca data show that the model of emission of optically thin plasma with lines with @xmath511015 kev does nt give a good approximation of the spectra ( @xmath52 for 41 d.o.f for the first observation ) .
high @xmath16 value is mainly driven by the fact that the model fails to predict the right ratio between the fluxes detected from different lines . if we take two - component model that consists of the emission of optically thin plasma cloud without emission lines ( @xmath19 ) with @xmath5114 kev and the emission of the plasma cloud with emission lines ( @xmath18 ) with @xmath515 kev , then we will get much better fit in terms of @xmath16 ( @xmath53 for 38 d.o.f . ) .
it should be noted that in order to get the good approximation one should take into account the doppler shift @xmath54 ( see .
table [ bremss_meka ] and the discussion below ) .
so , we believe that the spectrum of xte j0421 + 560 consists of ( at least ) two components , one of which describes the continuum spectrum extended up to @xmath11100 kev and the second one represents the region , where emission lines are formed . the difference between the characteristic temperatures of these two components during whole outburst may hint on the fact that they are emitted by the geometrically different regions of the system ( see also analysis of the flux variability above ) .
the best fit parameters for the discussed model are presented in table [ bremss_meka ] .
the combined pca and hexte spectra shown in fig.4 .
the pca data demonstrate that the spectrum of the source has strong emission line features at the energies around 6.56.7 kev and around 8 kev .
the existence of the strong line at the energy @xmath116.7 kev has been reported by all instruments which observed xte j0421 + 560 and had needed capabilities in this energy range .
new and interesting results of the present work is the shift of the central energy for fe - line complex and the detection of additional significant line - shaped emission around 8 kev .
model ] in the light of recent radio observations of relativistic jets it looks attractive to interpret the observed drift of fe - line in terms of the change of doppler effect value for relativistically moving plasma .
this interpretation is in agreement with results obtained by asca .
estimated velocity of the expansion is equal to @xmath55 for the observation on april 1 , with following decrease of this value leading to observed shift of line centrum to higher energies . if an initial shift is really caused by doppler effect , then an orientation of the cloud must be equal to @xmath56 relative to the line of sight .
thus the expected shape of the jets should be close to symmetrical as it was really observed in radio band .
our expansion velocity estimation is in principal agreement with asca results ( ueda et al .
, 1998b ) .
one serious argument against this interpretation is the lack of blue - shifted line component in the spectra .
emission line @xmath11 8 kev is hardly can be interpreted as such a component , because , even taking into account an uncertainty in its position , its intensity is still much weaker than for 6.5 - 6.7 kev line .
hovewer , one should take into account that pca energy resolution ( @xmath57 ) does not allow to distinguish between multiple lines which might present in this energy band .
it is remarkable , that similar shift in emission line energy up to @xmath58 kev had been observed for ss433 source during its observation by exosat and had been interpreted in terms of doppler effect ( watson et al . , 1986 ) .
but in that case only blue - shifted component had been detected , and the absense of red - shifted line was explained for the screening of the emitting region by the accretion disk .
if the change in the energy observed for 6.7 kev emission line is due to the relativistic motion of an emitting cloud of optically thin plasma , then the decrease in @xmath59 with time can be attributed to the decrease of the velosity of the cloud or to the precession of the jet axis , in the analogy with ss433 system .
hovewer , the time dependence of the emission measure ( see below ) does not show any significant decrease in the expansion velosity , which is an argument in favor of the precession . in this case
a period of the precession is @xmath60 days ( in comparison with @xmath61 days for ss433 ) .
s - shaped structure observed in radio band is another argument , which suggest that the change in observed expansion velosity is due to the change of the projection of the expansion vector to the line of sight .
as an alternative explanation of the observed shift of the line to lower energies one can consider a presence in the spectrum of fluorescent fe - line at 6.4 kev , which is generated during the passage of x - rays through cold dense cloud surrounding the x - ray source .
it is possible that an equivalent width of this line has decreased significantly during later observations , which resulted to the shift of the centrum of broad emission feature detected by pca .
the trend of low - frequency absorption discussed above is in favor of this interpretation .
hovewer , it looks like higher optical depth of cold cloud is needed to form 6.4 kev line with sufficient width .
this model is also discussed by ueda et al.(1998 ) as a possible interpretation of asca results .
the spectral feature observed around 8 kev could be attributed to the emission line of highly ionized ni or to fe @xmath62 line complex .
while an emission in fe line with energies @xmath63 kev has been widely observed from many x - ray sources , the detection of ni - line is quite unusual result .
we believe , that this detection became possible in this case only because of extreme brightness of the source and unusually high significance of line emission component in its spectrum combined with high sensitivity of pca experiment . to trace the changes of physical parameters for an emitting system we have applied two - component model compoused of the component of optically thin plasma emission without lines and another component of optically thin plasma with line emission ( as has been discussed above ) .
we have used pca data only , because its sensitivity is high enough to trace spectral parameters change for small exposure times .
we find best fit parameters for spectra integrated over 400-sec time intervals for the first three observational sessions , when the brightness of the source have been at maximum , and one integrated spectrum for the rest of the sessions .
the dependencies obtained for emission measure and temperature parameters are shown in figures 7 and 8 .
it is evident from fig.7 that after @xmath64 hours after the beginning of the outburst emission measure for both components is decreasing as @xmath65 , while overall behaviour of this parameter can be described as @xmath66 , where @xmath67 2.17 for the first component ( bremsstrahlung without lines ) and @xmath67 2.25 for the second component with lines ( meka ) .
there is no evidence in fig.7 that expansion velosity is decreasing with time .
if we assume that cloud mass @xmath68 is constant , then its internal thermal energy will depend of the temperature t only , and emission measure - of density n only . in this case
the volume of the cloud is proportional to the emission measure : @xmath69 . from the observations the time dependence of the emission measure parameter
is given by @xmath70 .
then the cloud volume will increase with time as @xmath71 .
we note , that index value @xmath72 is an indication of non - isotrop or collimated expansion of the cloud .
the temperature vs time dependence is quite complicated ( fig.8 ) , but one can mention two main features to describe it . during the first 20 - 30 hours after march 31.6
the temperature of the components remain relatively constant in comparison with significant change of the emission measure .
after @xmath64 hours , the temperature is decreasing as @xmath73 . while xte j0421 + 560 have a set of prominent differencies from other x - ray transients , radio observations
have shown an evident similarity between this source and ss433 , well - known galactic object emitting relativistic plasma jets .
an analogy between two sources is revealed itself in x - ray band also , because both of them have similar continuum spectra and strong emission lines aroun 6.7 kev .
furthermore , the shift of line centra with time is detected for both sources which might be explained in terms of doppler shift in line energy .
the important difference between two sources is a fact that xte j0421 + 560 , contrary to ss433 , is a transient object , which had been active in x - ray and radio bands for few days only .
we would like to note , that rapid decrease of x - ray flux with simultaneous softening of the spectrum , as well as shift of the maximum of energy emission with time for different bands of electromagnetic spectrum is typical also for gamma - burst afterglows .
the observations of x - ray transient xte j0421 + 560 , which is an x - ray counterpart of symbiotic binary ci cam , by rxte experiments give a set of important and unexpected results .
first of all , the source attracted the attention because of unusually fast rise of its x - ray flux , which was followed by also too fast decline after the maximum passage . during 10 days of observations by pca and hexte experiments
the flux from the source had been decreased by more than two orders of magnitude .
one interesting result is the detection of soft x - ray flare , which was not correlated with the flux change in harder energy range .
the source spectrum in 3 - 150 kev band can be approximated by power - law with an exponential cut - off at higher energies .
the spectrum of this shape can be generated in the cloud of non - thermalized optically thin plasma .
the softening of the spectrum ( or decrease of the effective temperature ) with the decrease of flux has been mentioned .
the changes in low - energy absorption in the spectrum is probably due to the decrease in density of the cold cloud , which surrounds x - ray source .
initial mass of the cloud can be estimated as @xmath74 .
the emission lines have been detected at energies 6.5 - 6.7 kev and around 8 kev .
discovered shift of the 6.7-kev line to lower energies can be explained either as doppler - effect for moving media , or as generation of another , 6.4-kev emission line in cold cloud , density of which evolved with time . if the reason is doppler - effect , then this is the second time , after ss433 , when an x - ray emission generated by relativistic jet plasma is observed in our galaxy . in this case
the cloud is expanded under angle @xmath75 to the line of sight .
the appearance of the line near 8 kev is probably due to the emission of highly ionized ni or fe @xmath62 line , and , as far as we know , is detected for the first time for galactic sources .
authors thank e.churazov for helpfull advises and comments , s. sazonov , s. trudolyubov and a. finoguenov for extensive discussions of the presented results .
authors especially thank keith jahoda for the help in the analysis of spectral capabilities of pca .
also we thank y.ueda and h.inoue for the possibility to get to know asca results before its publication .
the research has made use of data obtained through the high energy astrophysics science archive research center online service , provided by the nasa / goddard space flight center .
this work was partly supported be rbrf grant 96 - 15 - 96343 .
bradt h. , rotshild r. , swank j. astron .
astrophys . suppl .
ser . , 1993 , vol .
1 , p. 355 - 360 .
clccccc + & obs .
i d & date & obs .
time & + & & & & pca&hexte@xmath76 + + 1 & 30171 - 01 - 01 - 00 & 01/04/98 & 01:52:48 - 03:52:32 & 4109 & 1443 + 2 & 30171 - 01 - 02 - 07s & 01/04/98 & 08:21:36 - 10:52:32 & 6176 & 4142 + 3 & 30409 - 01 - 03 - 00 & 02/04/98 & 06:40:32 - 10:49:36 & 9702 & 3345 + 4 & 30409 - 01 - 04 - 00 & 03/04/98 & 05:02:24 - 09:10:24 & 2740 & 892 + 5 & 30409 - 01 - 04 - 01 & 03/04/98 & 11:30:24 - 11:50:24 & 710 & 217 + 6 & 30409 - 01 - 05 - 00 & 04/04/98 & 06:41:20 - 09:27:28 & 6950 & 2283 + 7 & 30409 - 01 - 06 - 00 & 05/04/98 & 06:44:48 - 10:10:24 & 7590 & 2671 + 8 & 30409 - 01 - 07 - 00 & 06/04/98 & 03:21:36 - 04:30:56 & 1503 & 1016 + 9 & 30409 - 01 - 08 - 00 & 07/04/98 & 02:25:20 - 05:01:36 & 4181 & 1389 + 10 & 30409 - 01 - 08 - 01 & 07/04/98 & 18:02:08 - 18:22:56 & 743 & 191 + 11 & 30409 - 01 - 09 - 00 & 08/04/98 & 03:35:12 - 06:11:28 & 5830 & 2009 + 12 & 30409 - 01 - 10 - 00 & 09/04/98 & 06:33:20 - 07:32:32 & 3035 & 1059 + + cccccc + & mjd&@xmath77&@xmath78&flux ( 320 kev ) & @xmath12 + & & & kev&@xmath79 erg/@xmath80/s & @xmath81 atom/@xmath80 + + 1 & 50904.08&@xmath82&@xmath83&@xmath84&@xmath85 + 2 & 50904.35&@xmath86&@xmath87&@xmath88&@xmath89 + 3 & 50905.28&@xmath90&@xmath91&@xmath92&@xmath93 + 4 & 50906.21&@xmath94&@xmath95&@xmath96&@xmath97 + 5 & 50906.48&@xmath98&@xmath99&@xmath100&@xmath97 + 6 & 50907.28&@xmath101&@xmath102&@xmath103&@xmath97 + 7 & 50908.28&@xmath104&@xmath105&@xmath106&@xmath97 + 8 & 50909.14&@xmath107&@xmath108&@xmath109&@xmath97 + 9 & 50910.10&@xmath110&@xmath111&@xmath112&@xmath97 + 10 & 50910.75&@xmath113&@xmath114&@xmath115&@xmath97 + 11 & 50911.15&@xmath116&@xmath117&@xmath118&@xmath119 + 12 & 50912.27&@xmath120&@xmath121&@xmath122&@xmath97 + + parameter @xmath12 is presented without errors , because of the main contribution to the uncertainty of this parameter is not the statistical one , but rather the uncertainty of respose matrix on energy range less than 4 - 5 kev . as usual
the uncertainty of @xmath12 value is about @xmath123 ccccccccc + & & blend 6.7 kev & & & blend 8 kev & + + & e , kev & @xmath124 , kev & ew , ev & e , kev & @xmath124 , kev & ew , ev + + + 1&@xmath125&@xmath126&@xmath127&@xmath128&@xmath129&@xmath130 + 2&@xmath131&@xmath132&@xmath133&@xmath134&@xmath129&@xmath135 + 3&@xmath131&@xmath136&@xmath137&@xmath138&@xmath129&@xmath139 + 4&@xmath140&@xmath141&@xmath142&@xmath143&@xmath129&@xmath144 + 5&@xmath145&@xmath146&@xmath147&@xmath148&@xmath129&@xmath149 + 6&@xmath150&@xmath151&@xmath152&@xmath153&@xmath129&@xmath154 + 7&@xmath155&@xmath156&@xmath157 & 8.10@xmath158&@xmath129&@xmath159 + 8&@xmath160&@xmath161&@xmath162 & 8.10@xmath158&@xmath129&@xmath163 + 9&@xmath164&@xmath161&@xmath165 & 8.10@xmath158&@xmath129&@xmath166 + 10&@xmath167&@xmath168&@xmath169 & 8.10@xmath158&@xmath129&@xmath170 + 11&@xmath171&@xmath172&@xmath173 & 8.10@xmath158&@xmath129&@xmath174 + 12&@xmath175&@xmath176&@xmath177 & 8.10@xmath158&@xmath129&@xmath178
+ + + ccccccccc + & @xmath179 & @xmath180 & @xmath181 & red shift@xmath182 & @xmath180 & @xmath183 + & kev & bremsshtr . &
kev & & meka & ( 290 dof . )
+ + 1&@xmath184&@xmath185&@xmath186&@xmath187&@xmath188&324.68 + 2&@xmath189&@xmath190&@xmath191&@xmath187&@xmath192&298.46 + 3&@xmath193&@xmath194&@xmath195&@xmath196&@xmath197&280.74 + 4&@xmath198&@xmath199&@xmath200&@xmath201&@xmath202&241.21 + 5&@xmath203&@xmath204&@xmath205&@xmath206&@xmath207&258.72 + 6&@xmath208&@xmath209&@xmath210&@xmath211&@xmath212&269.72 + + + | we present the results of observations of xte j0421 + 560 x - ray transient with rossi x - ray timing explorer in the beginning of april 1998 .
lightcurve , obtained by asm all - sky monitor , shows unusually fast decrease in the object s luminosity .
spectra from series of observations by pca and hexte experiments were studied in detail .
two emission line regions with energies around 6.5 kev and around 8 kev have been mentioned in the spectra . the possibility of generation of the observed emission in relativistic plasma jet has been discussed .
some analogy is found with known galactic jet source ss433 . |
we start with the discussion of the recent _ ab initio _ variational calculations of light nuclei up to the mass number @xmath0 of the argonne - illinois group @xcite .
they take the two nucleon interaction obtained from the scattering data and add the three body interaction so as to reproduce the binding energies of the three nucleon systems .
they reproduce very nicely the light nuclei including the excited states .
we have now a technique to calculate many body quantum systems with mass of about 10 .
these calculations provide another surprising result that the pion matrix element is about @xmath1 of the whole attraction .
this importance of the pion degrees of freedom in nuclei should have experimental consequences .
for this we show in fig .
1 the recent gamow - teller distributions obtained by ( @xmath2he , t ) reaction at rcnp @xcite . due to the ultra high resolution , we find very small peaks at low excitation energy .
considering the very simple structure of the gamow - teller operator , @xmath3 , it is surprising to have so many peaks in the low excitation region .
we will show that this fragmentation of the gt strength is a natural consequence of the role of pion in finite nuclei .
we mention also that ratios of the longitudinal and transverse spin response functions are also the indication of the role of pion in finite nuclei .
these observations motivated us to examine the possibility of the finite pion mean field in finite nuclei @xcite .
we keep the pion interaction term in the lagrangian and introduce the finite pion mean field .
the introduction of finite pion mean field makes the breaking of parity and the mixing of protons and neutrons in the intrinsic single particle state .
we mention also that the pion source term contains the derivative of the spin - isospin density distribution and hence needs spatial variation .
we have calculated various n = z nuclei in the periodic table without the coulomb term with the inclusion of the pion - nucleon coupling .
we have found that the pion mean field becomes finite and the energy gain due to the pion term behaves as the nuclear surface , which indicate that pion contribution comes from the nuclear surface .
we have found also that the pionic phase transition is not so strong and need a special care in the treatment of the finite pion mean field .
this preliminary study of the role of pion in finite nuclei suggests us to look into the parity and the charge projection and also the variation of the symmetry projected wave function . in order to see the structure of the projected wave function , we write here the case of parity projection to simplify the notation .
we discuss here the role of the pion by performing the parity projection from the symmetry broken intrinsic state .
we write the single particle state with mixed parity in a simpler form as @xmath4 here , @xmath5 denotes a parity mixed single particle state expressed as a linear combination of @xmath6 , some parity state ( we call it as a normal parity state ) and @xmath7 , the opposite parity state ( abnormal parity state ) .
we write the intrinsic state with these single particle states up to the fermi surface and with all the magnetic sub - shells filled as @xmath8 this intrinsic state has the total spin 0 because all the magnetic sub - shells are filled , but the parity is mixed
. the first term , @xmath9 , in ( [ eq : state ] ) has the positive parity and corresponds to the ground state in the zeroth order .
the second term has the negative parity , since each normal parity state , @xmath10 , is replaced by an abnormal parity state , @xmath11 for all occupied @xmath10 .
hence , if we say the first term as the 0p-0h state , then the second term is a coherent 1p-1h state with @xmath12 spin parity .
the third term consists of 2p-2h states with a pair of 1p-1h states with @xmath12 spin parity and therefore has @xmath13 spin parity .
the next term has three 1p-1h states with @xmath12 spin parity and therefore has @xmath12 spin parity and so on .
hence the positive parity projection @xmath14 would provides the state with even number of 1p-1h states with @xmath12 spin parity .
this means that the positive parity projection provides 2p-2h states as the major correction terms . in this sense ,
the parity projected mean field theory with pion condensation is related strongly with the finding of kaiser et al .
, who claim that a large attraction arises from 2p-2h configurations due to the pion exchange interaction @xcite . the negative parity projection @xmath16 would provide the state with odd number of 1p-1h states with @xmath12 spin parity .
this is the brother state having the quantum number of @xmath12 to the @xmath13 ground state .
the ground state consists of highly correlated particle
hole states .
we perform the parity and charge projection calculations for the alpha nucleus @xcite .
the role of the symmetry projection is extremely important for the quantitative account of the pionic correlations for finite nuclei .
the chiral symmetry is the most important symmetry in the strong interaction .
we studied the role of chiral symmetry on the property of finite nuclei using the chiral sigma model @xcite .
the famous lagrangian is the one of gell - mann and levy , where the pion field appears symmetrically with the sigma field @xcite .
we start with the linear sigma model with the omega meson field , which is defined by the following lagrangian , @xmath18 the fields @xmath19 , @xmath20 and @xmath21 are the nucleon , sigma and the pion fields . @xmath22 and @xmath23 are the sigma model coupling constants . here
we have introduced the explicit chiral symmetry breaking term , @xmath24 , and in addition the mass generation term for the omega meson due to the sigma meson condensation as the case of the nucleon mass in the free space @xcite .
we call this model as the extended chiral sigma model hereafter . in a finite nuclear system
, it is believed to be essential to use the non - linear representation of the chiral symmetry .
this is because the pseudoscalar pion - nucleon coupling in the linear sigma model makes the coupling of positive and negative energy states extremely strong and we have to treat the negative energy states very carefully .
we can derive the non - linear sigma model by introducing new variables in the polar coordinates from the rect - angular coordinates and making a suitable transformation , @xmath25 , with @xmath26 .
we further implement the weinberg transformation for the nucleon field as @xmath27 .
after several steps , we obtain the sigma - omega model lagrangian in non - linear representation given as follows , @xmath28 where we set @xmath29 , @xmath30 , @xmath31 and @xmath32 .
the effective mass of the nucleon and omega meson are given by @xmath33 and @xmath34 , respectively .
we take the following masses and the pion decay constant as , @xmath35 , @xmath36 , @xmath37 , and @xmath38 .
then , the other parameters can be fixed automatically by the following relations , @xmath39 and @xmath40 .
the strength of the cubic and quadratic sigma meson self - interactions depends on the sigma meson mass through the following relation , @xmath41 , in the chiral sigma model .
the mass of the sigma meson , @xmath42 , and the coupling constant of omega and nucleon , @xmath43 , are the free parameters .
we calculate first infinite matter using this lagrangian in the mean field approximation without the pion mean field @xcite . the sigma meson mass and the omega nucleon coupling
are adjusted to reproduce the saturation property of nuclear matter .
we found that the incompressibility came out to be very large , k @xmath44 mev instead of k @xmath45 mev .
we found also the vector and scalar self - energies were about half as large as those in the case of the rmf(tm1 ) .
we discuss now the consequence of these properties of nuclear matter on the finite nuclei .
we discuss the properties of finite nuclei in terms of the extended chiral sigma model .
we show the results of binding energies per particle of n = z even - even mass nuclei from n = 16 up to n = 34 in fig .
we take all the parameters of the extended chiral sigma model as those of the nuclear matter except for @xmath43 = 7.176 instead of 7.033 for overall agreement with the rmf(tm1 ) results @xcite . for comparison , we calculate these nuclei within the rmf approximation without pairing nor deformation . the rmf(tm1 ) provides the magic numbers , which are seen as the binding energy per particle increases at n = z = 20 and 28 . on the other hand , the extended chiral sigma model without the pion mean field provides the magic number behavior only at n = z = 18 instead of n = z = 20 .
we show also the result of the finite pion mean field , which is achieved by making the single particle states have mixed parity with definite total spin .
the inclusion of the pion mean field provides the magic number at n = z = 28 . in order to see why the difference between the two models for the lagrangian arises
, we discuss the single particle levels for the two models . in the case of the tm1 parameter
set , the shell gaps appear clearly at n = 20 and 28 .
the magic number at n = 20 is due to the central potential , while the magic numbers at n = 28 comes from the spin - orbit splitting of the 0f - orbit .
this is definitely due to the fact that the vector potential and the scalar potential in nuclear matter are large so as to provide the large spin - orbit splitting . on the other hand ,
the single particle spectrum of the extended chiral sigma model is quite different from this case as seen in fig .
most remarkable structure is that the @xmath46 orbit is strongly pushed up . due to this reason the @xmath47 orbit becomes the magic shell at n = 18 and the magic number appears at n = 18 instead of n = 20 .
we see also not strong spin - orbit splitting and hence there appears no shell gap at n = 28
. the first discrepancy could be due to the large incompressibility as seen in the nuclear matter energy per particle .
the other is due to the relatively small vector and scalar potentials in nuclear matter @xcite .
we discuss the effect of the finite pion mean field on the single particle energies .
we show in fig .
4 the single particle spectra for various nuclei .
we state again that the single particle states have mixed parity with definite total spin indicated beside each level .
we see clearly the large energy differences between the spin - orbit partners to be produced by the finite pion mean field as the energy differences become maximum for nuclei at n = 28 .
the pion mean field makes coupling of different parity states with the same total spin .
the @xmath48 state repels each other with the @xmath49 state and therefore the @xmath48 state is pushed down and the @xmath49 state is pushed up .
the next partner is @xmath50 and @xmath47 .
the @xmath50 state is pushed down , while the @xmath47 state is pushed up .
the next partner is @xmath51 and @xmath52 .
the @xmath51 state is pushed down , while the @xmath52 state is pushed up .
this pion mean field effect continues to higher spin partners .
this coupling of the different parity states with the same total spin due to the finite pion mean field causes the splittings of the spin - orbit partners as seen clearly for the @xmath53 spin - orbit partner , @xmath54 spin - orbit partner and @xmath55 spin - orbit partner in @xmath56ni .
it is extremely interesting to see that the appearance of the energy splitting between the spin - orbit partners for the case of the finite pion mean field is caused by completely a different mechanism from the case of the spin - orbit interaction .
we showed the recently observed high resolution gamow - teller distributions of @xmath57zr(@xmath2he , t ) reaction in fig.1 .
the finite pion mean field makes the intrinsic single particle states have mixed parities .
hence , in addition to the natural shell model type gt excitation , there should appear many tiny gt peaks , whose excitation energies are related with the pion mean field strength .
we are working on the parity and the charge projection for the gt excitation spectra .
we discuss qualitatively also other observables to be explained by the finite pion mean field .
there is the famous missing single particle strength problem , in which the s - wave component is largely missing .
the experiment is the electron scattering with @xmath58pb and @xmath59tl .
the difference between these two charge distributions is compared with the expected single particle @xmath60 wave function .
particularly , the strength at the nuclear center expected from the @xmath60 wave function is missing by almost 30% .
this experimental data is strongly related with the parity mixing of the intrinsic single particle state , where @xmath61 state should have an admixture from @xmath62 state .
the magnetic moments are largely deviated from the schmit values . if the intrinsic single particle states have some total spin with mixed parity , then the magnetic moment of that orbit ought to have a value in between the schmit values .
there are many other observables to be explained qualitatively by the finite pion mean field causing parity mixing .
we have discussed the role of chiral symmetry in terms of the chiral sigma model for the description of finite nuclei . we have applied the chiral sigma model with finite pion mean field to finite nuclei and found that the nucleon mass is reduced by about 20% .
the single particle levels reflected the bad feature of the nuclear matter property as having a too strong incompressibility .
we could reproduce the j - j magic number due to the finite pion mean field , which provides a mechanism of providing the splitting between spin - orbit partners .
all the observables related with the spin and isospin should be related with the finite pion mean field
. we have to work out the projection of the parity and the charge numbers for quantitative description of finite nuclei . | we introduce the chiral symmetry in the description of finite nuclei with the hope to unify the hadron and nuclear physics .
this trial becomes possible by introducing finite pion mean field , which contributes largely to enhance the spin - orbit splitting so as to provide the j - j magic number .
we discuss the qualitative consequence of the finite pion mean field for various observables . |
one of the oldest theorems in graph theory is attributed to euler , and it says that a ( undirected ) graph admits an _ euler circuit _ , i.e. , a closed walk visiting every edge exactly once , if and only if a graph is connected and all its vertices are of even degrees . respectively , a directed graph has a _ directed euler circuit _ if and only if a graph is ( weakly ) connected and for each vertex , its in - degree is equal to its out - degree . while checking
if a given directed or undirected graph is eulerian is easily done in polynomial time , the problem of finding @xmath1 edges ( arcs ) in a graph to form an eulerian subgraph is np - hard .
we refer to the book of fleischner @xcite for a thorough study of eulerian graphs and related topics . in @xcite , cai and
yang initiated the study of parameterized complexity of subgraph problems motivated by eulerian graphs .
particularly , they considered the following parameterized subgraph and induced subgraph problems : and the decision versions of both @xmath1-circuit and euler @xmath1-subgraph are known to be the @xmath4-complete @xcite .
cai and yang in @xcite proved that @xmath1-circuit on undirected graphs is @xmath5 . on the other hand ,
the authors have shown in @xcite that euler @xmath1-subgraph is @xmath6$]-hard .
the variant of the problem @xmath7-circuit , where one asks to remove at most @xmath1 edges to obtain an eulerian subgraph was shown to be @xmath5 by cygan et al .
@xcite on directed and undirected graphs .
the problem of removing at most @xmath1 vertices to obtain an induced eulerian subgraph , namely euler @xmath8-subgraph , was shown to be @xmath6$]-hard by cai and yang for undirected graphs @xcite and by cygan et al .
for directed graphs @xcite .
dorn et al . in @xcite
provided fpt algorithms for the weighted version of eulerian extension . in this work
we extend the set of results on the parameterized complexity of eulerian subgraph problems by considering the problems of finding an ( induced ) eulerian subgraph with _ at least _
@xmath1 ( vertices ) edges .
we consider the following problems : and the decision version of long circuit was shown to be @xmath4-complete by cygan et al . in @xcite and it is not difficult to see that same is true for large euler subgraph . let us note that by plugging - in these observations into the framework of bodlaender et al .
@xcite , it is easy to conclude that on undirected graphs both problems have no polynomial kernels unless @xmath9 .
however , the parameterized complexity of these problems appears to be much more interesting . * our results . *
we show that large euler subgraph behaves differently for directed and undirected cases . for undirected graphs , we prove that the problem is @xmath5 .
we find this result surprising , because very closely related euler @xmath1-subgraph is known to be @xmath6$]-hard @xcite . the proof is based on a structural result interesting in its own .
roughly speaking , we show that large treewidth certifies containment of a large induced euler subgraph . for directed graphs ,
large euler subgraph is @xmath4-complete for each @xmath10 , and this bound is tight the problem is polynomial - time solvable for each @xmath3 .
we also prove that euler @xmath1-subgraph is @xmath6$]-hard for directed graphs .
long circuit is proved to be @xmath5 for directed and undirected graphs .
our algorithm is based on the results by gabow and nie @xcite about the parameterized complexity of finding long cycles .
the known and new results about euler subgraph problems are summarized in table [ tabl : compl ] .
.parameterized complexity of euler subgraph problems .
[ cols="^,^,^",options="header " , ] this paper is organised as follows .
section [ sec : defs ] contains basic definitions and preliminaries . in section
[ sec : vertex - undir ] we show that large euler subgraph is fpt on undirected graphs . in section [ sec
: vertex - dir ] we prove that on directed graphs , euler @xmath1-subgraph is w[1]-hard while large euler subgraph is np - complete for each @xmath10 . in section
[ sec : edge ] we treat long curcuit and show that it is fpt on directed and undirected graphs .
* graphs . *
we consider finite directed and undirected graphs without loops or multiple edges .
the vertex set of a ( directed ) graph @xmath0 is denoted by @xmath11 , the edge set of an undirected graph and the arc set of a directed graph @xmath0 is denoted by @xmath12 . to distinguish edges and arcs , the edge with two end - vertices @xmath13
is denoted by @xmath14 , and we write @xmath15 for the corresponding arc . for a set of vertices @xmath16 , @xmath17 $ ] denotes the subgraph of @xmath0 induced by @xmath18 , and by @xmath19 we denote the graph obtained form @xmath0 by the removal of all the vertices of @xmath18 , i.e. , the subgraph of @xmath0 induced by @xmath20 .
let @xmath0 be an undirected graph .
for a vertex @xmath21 , we denote by @xmath22 its _ ( open ) neighborhood _ , that is , the set of vertices which are adjacent to @xmath21 . the _ degree _ of a vertex @xmath21 is denoted by @xmath23 , and @xmath24 is the maximum degree of @xmath0 .
let now @xmath0 be a directed graph .
for a vertex @xmath25 , we say that @xmath26 is an _ in - neighbor _ of @xmath21 if @xmath27 . the set of all in - neighbors of @xmath21
is denoted by @xmath28 .
the _ in - degree _ @xmath29 .
respectively , @xmath26 is an _ out - neighbor _ of @xmath21 if @xmath30 , the set of all out - neighbors of @xmath21 is denoted by @xmath31 , and the _ out - degree _ @xmath32 . for a ( directed ) graph @xmath0 , a ( directed ) _ trail _ of _ length _
@xmath1 is a sequence @xmath33 of vertices and edges ( arcs resp . ) of @xmath0 such that @xmath34 , @xmath35 , the edges ( arcs resp . ) @xmath36 are pairwise distinct , and for @xmath37 , @xmath38 ( @xmath39 resp . ) .
a trail is said to be _ closed _ if @xmath40 .
a closed ( directed ) trail is called a ( directed ) _ circuit _ , and it is a ( directed ) _ cycle _ if all its vertices except @xmath40 are distinct . clearly , any cycle is a subgraph of @xmath0 , and it is said that @xmath41 is an _ induced cycle _ of @xmath0 if @xmath42 $ ] .
a ( directed ) path is a trail such that all its vertices are distinct . for a ( directed ) walk ( trail , path resp . )
@xmath33 , @xmath43 and @xmath44 are its _ end - vertices _ , and @xmath45 are its _ internal _ vertices . for a a ( directed ) walk ( trail , path resp . ) with end - vertices @xmath26 and @xmath21 , we say that it is an _
( _ trail _ , _ path _ resp . ) .
we omit the word `` directed '' if it does not create a confusion .
also we write a trail as a sequence of its vertices @xmath46 .
a connected ( directed ) graph @xmath0 is an _ euler _ ( or _ eulerian _ ) graph if it has a ( directed ) circuit that contains all edges ( arcs resp . ) of @xmath0 . by the famous result by euler ( see , e.g. , @xcite ) ,
a connected graph @xmath0 is an euler graph if and only if all its vertices have even degrees . respectively
, a connected directed graph @xmath0 is an euler directed graph if and only if for each vertex @xmath25 , @xmath47 .
* ramsey numbers .
* the _ ramsey number _
@xmath48 is the minimal integer @xmath49 such that any graph on @xmath49 vertices has either a clique of size @xmath50 or an independent set of size @xmath51 . by the famous paper by erds and szekeres @xcite , @xmath52
. * parameterized complexity .
* parameterized complexity is a two dimensional framework for studying the computational complexity of a problem .
one dimension is the input size @xmath49 and another one is a parameter @xmath1 .
it is said that a problem is _ fixed parameter tractable _ ( or @xmath5 ) , if it can be solved in time @xmath53 for some function @xmath54 , and it is said that a problem is in @xmath55 , if it can be solved in time @xmath56 for some function @xmath54 .
one of the basic assumptions of the parameterized complexity theory is the conjecture that the complexity class @xmath57}}\neq { \ensuremath{\operatorname{fpt}}}$ ] , and it is unlikely that a @xmath6$]-hard problem could be solved in @xmath5-time .
a problem is _ @xmath58-hard_(_complete _ )
if it is @xmath4-hard ( complete ) for some fixed value of the parameter @xmath1 . clearly , a @xmath58-hard problem is not in @xmath55 unless @xmath59@xmath60@xmath4 .
we refer to the books of downey and fellows @xcite , flum and grohe @xcite , and niedermeier @xcite for detailed introductions to parameterized complexity .
* treewidth . * a _ tree decomposition _ of a graph
@xmath0 is a pair @xmath61 where @xmath62 is a tree and @xmath63 is a collection of subsets ( called _ bags _ ) of @xmath11 such that : 1 .
@xmath64 , 2 .
for each edge @xmath65 , @xmath66 for some @xmath67 , and 3 . for each @xmath68 the set @xmath69 induces a connected subtree of @xmath62
. the _ width _ of a tree decomposition @xmath70 is @xmath71 .
treewidth _ of a graph @xmath0 ( denoted as @xmath72 ) is the minimum width over all tree decompositions of @xmath0 .
recall that our aim is to prove that long circuit is @xmath5 for directed and undirected graphs and large euler subgraph is @xmath5 for undirected graphs .
hence , we should argue that these problems are @xmath4-hard .
notice that for exact variants of these problems , i.e. , @xmath1-circuit and euler @xmath1-subgraph for undirected graphs , the @xmath4-hardness was proved by cai and yang @xcite , and for long circuit ( for directed and undirected cases ) , it was shown by cygan et al . in @xcite .
[ prop : npc ] large euler subgraph is @xmath4-complete for undirected graphs when @xmath1 is a part of the input . let @xmath0 be a @xmath49-vertex cubic graph
. denote by @xmath73 the graph obtained by subdividing each edge of @xmath0 .
it is straightforward to see that @xmath73 has an induced euler subgraph with at least @xmath74 vertices if and only if @xmath0 is hamiltonian .
as the hamiltonian cycle is known to be @xmath4-complete for cubic planar graphs @xcite , the result follows .
finally , we observe that while we obtain @xmath5 results , it is unlikely that these problems have polynomial kernels . let @xmath0 be a ( directed ) graph with @xmath75 connected components @xmath76 , and let @xmath1 be a positive integer .
notice that @xmath0 has a circuit with at least @xmath1 edges ( arcs resp . ) if and only if @xmath77 has a circuit with at least @xmath1 edges ( arcs resp . ) for some @xmath78 .
also @xmath0 has an induced euler subgraph with at least @xmath1 vertices if and only if @xmath77 has an induced euler subgraph with at least @xmath1 vertices for some @xmath78 . by the results of bodlaender
@xcite , this observation together with proposition [ prop : npc ] and the np - hardness of long circuit @xcite immediately imply the following proposition .
[ prop : kernel ] long circuit for directed and undirected graphs and large euler subgraph for undirected graphs have no polynomial kernels unless @xmath9 .
in this section we show that large euler subgraph is @xmath5 for undirected graphs . using ramsey arguments
, we prove that if a graph @xmath0 has sufficiently large treewidth , then @xmath0 has an induced euler subgraph on at least @xmath1 vertices .
then if the input graph has large treewidth , we have a yes - answer .
otherwise , we use the fact that large euler subgraph can be solved in @xmath5 time for graphs of bounded treewidth . all considered here graphs are undirected . for a given positive integer @xmath1
, we define the function @xmath79 for integers @xmath80 recursively as follows : * @xmath81 , * @xmath82 for @xmath83 .
we need the following two lemmas .
[ lem : connect ] let @xmath0 be a graph , and suppose that @xmath84 are distinct vertices joined by at least @xmath79 internally vertex - disjoint paths of length at most @xmath85 in @xmath0 for some @xmath80 .
then @xmath0 has an induced euler subgraph on at least @xmath1 vertices .
consider the minimum value of @xmath85 such that @xmath0 has @xmath79 internally vertex disjoint @xmath86-paths .
we have at least @xmath87 such paths @xmath88 that are distinct from the path @xmath84 .
we assume that each path @xmath89 has no chords that either join two internal vertices or an internal vertex and one of the end - vertices , i.e. , each internal vertex is adjacent in @xmath90 $ ] only to its two neighbors in @xmath89 .
otherwise , we can replace @xmath89 by a shorter path with all vertices in @xmath91 distinct from the path @xmath84 .
we consider two cases .
* case 1 . @xmath92 . * the paths @xmath88 are of length two and therefore have exactly one internal vertex .
assume that @xmath93 are internal vertices of these paths .
because @xmath94 , the graph @xmath95 $ ] either has a clique @xmath96 of size @xmath1 or an independent set @xmath97 of size at least @xmath98 .
suppose that @xmath0 has a clique @xmath96 .
if @xmath1 is odd , then @xmath99 $ ] is an induced euler subgraph on @xmath1 vertices .
if @xmath1 is even , then @xmath100 $ ] is an induced euler subgraph on @xmath101 vertices .
assume now that that @xmath102 is an independent set of size @xmath98 .
let @xmath103 .
if @xmath104 and @xmath1 is even or @xmath105 and @xmath1 is odd , then @xmath106 $ ] is an induced euler subgraph on @xmath101 vertices .
else if @xmath105 and @xmath1 is even or @xmath104 and @xmath1 is odd , then @xmath107 $ ] is an induced euler subgraph on @xmath1 vertices .
@xmath108 . *
we say that paths @xmath89 and @xmath109 are _ adjacent _ if they have adjacent internal vertices .
let @xmath110 .
suppose that there is an internal vertex @xmath21 of one of the paths @xmath88 that is adjacent to at least @xmath111 internal vertices of some other distinct @xmath111 paths .
then there are @xmath110 paths @xmath112 that have respective internal vertices @xmath113 such that i ) @xmath21 is adjacent to @xmath113 and ii ) either each @xmath114 is at distance at most @xmath115 from @xmath51 in @xmath116 for all @xmath117 or each @xmath114 is at distance at most @xmath115 from @xmath75 in @xmath116 for all @xmath117 .
but then either the vertices @xmath118 or @xmath119 are joined by at least @xmath120 internally vertex - disjoint paths of length at most @xmath121 .
this contradicts our choice of @xmath85 .
hence , for each @xmath122 , any internal vertex of @xmath89 has adjacent internal vertices in at most @xmath123 other paths , and @xmath89 is adjacent to at most @xmath124 other paths . as @xmath125 , there are @xmath98 distinct paths @xmath126 that are pairwise non - adjacent , i.e. , they have no adjacent internal vertices .
let @xmath127 $ ] and @xmath128 $ ] .
notice that by our choice of the paths , @xmath129 and @xmath130 if @xmath105 , and @xmath131 ( @xmath132 resp . ) can be obtained from @xmath133 ( @xmath134 resp . ) by the removal of @xmath135 if @xmath84 are adjacent .
if @xmath104 and @xmath1 is even or @xmath105 and @xmath1 is odd , then @xmath133 is an induced euler subgraph on at least @xmath101 vertices .
else if @xmath105 and @xmath1 is even or @xmath104 and @xmath1 is odd , then @xmath134 is an induced euler subgraph on at least @xmath1 vertices . for @xmath10 ,
let @xmath136 [ lem : deg ] for @xmath10 , any 2-connected graph @xmath0 with @xmath137 has an induced euler subgraph on at least @xmath1 vertices .
let @xmath0 be a 2-connected graph and let @xmath26 be a vertex of @xmath0 with @xmath138 .
as @xmath0 is 2-connected , @xmath139 is connected .
let @xmath21 be an arbitrary vertex of @xmath140 .
denote by @xmath62 a tree of shortest paths from @xmath21 to all other vertices of @xmath140 in @xmath73 .
* claim a. * _ if there is a @xmath141-path @xmath142 of length at least @xmath143 in @xmath62 for some @xmath144 , then @xmath0 has an induced euler subgraph on at least @xmath1 vertices .
_ denote by @xmath145 the vertices of @xmath140 in @xmath142 in the path order , @xmath146 and @xmath147 .
let @xmath148 be the union of @xmath149-subpaths of @xmath142 , let @xmath150 be the union of @xmath151-subpaths of @xmath142 , and let @xmath152 be the union of @xmath153-subpaths of @xmath142 .
notice that some subpaths can be empty depending whether @xmath50 modulo 3 is 0 , 1 or 2 .
observe that @xmath154 are edge - disjoint induced subgraphs of @xmath0 . since @xmath155 , there is @xmath156 for @xmath157 with at least @xmath158 edges .
then @xmath156 has at least @xmath98 vertices .
let @xmath159 $ ] . by the definition of @xmath156 , @xmath133 is a union of induced cycles with one common vertex @xmath26 such that for different cycles @xmath160 in the union , @xmath161 and @xmath162 whenever @xmath163 and @xmath164 .
hence , @xmath133 is an euler graph with at least @xmath1 vertices .
from now we assume that all @xmath141-paths in @xmath62 have length at most @xmath165 for @xmath144 .
* claim b. * _ if there is a vertex @xmath166 with @xmath167 , then @xmath0 has an induced euler subgraph on at least @xmath1 vertices .
_ recall that @xmath62 is a tree of shortest paths from @xmath21 .
we assume that @xmath62 is rooted in @xmath21 .
then the roof defines the parent - child relation on @xmath62 .
let @xmath168 be a parent of @xmath169 ( if exists ) and let @xmath170 be the children of @xmath169 .
if @xmath169 has no parent , then @xmath171 and @xmath172 .
otherwise , @xmath173 .
let @xmath174 . for each @xmath175 ,
let @xmath176 be a descendant of @xmath177 in @xmath62 .
denote by @xmath89 the unique @xmath178-path in @xmath62 for @xmath179 . as all @xmath141-paths in @xmath62 have length at most @xmath165 for @xmath144 , the paths @xmath180 have length at most @xmath165 .
notice that these paths have no common vertices except @xmath169 .
observe now that @xmath181 are adjacent to @xmath26 in @xmath0 .
therefore , we have at least @xmath182 internally vertex - disjoint @xmath183-paths in @xmath0 . by lemma [ lem :
connect ] , it implies that @xmath0 has an induced euler subgraph on at least @xmath1 vertices . to complete the proof of the lemma , it remains to observe that if @xmath184 and all @xmath141-paths in @xmath62 have length at most @xmath165 for @xmath144 , then @xmath185 kosowski et al .
@xcite obtained the following bound for treewidth .
[ thm : bod ] let @xmath0 be a graph without induced cycles with at least @xmath186 vertices and let @xmath187 .
then @xmath188 .
this theorem together with lemma [ lem : deg ] immediately imply the next lemma .
[ lem : tw - bound ] let @xmath0 be a graph and let @xmath10 .
if @xmath189 , then @xmath0 has an induced euler subgraph on at least @xmath1 vertices .
suppose that @xmath189 .
then @xmath0 has a 2-connected component @xmath73 with @xmath190 .
if @xmath191 , then @xmath73 has an induced euler subgraph on at least @xmath1 vertices by lemma [ lem : deg ] .
otherwise , by theorem [ thm : bod ] , @xmath73 has an induced cycle on at least @xmath1 vertices , i.e. , an induced euler subgraph .
now we show that large euler subgraph is @xmath5 for graphs of bounded treewidth .
[ lem : cmsol ] for any positive integer @xmath75 , large euler subgraph can be solved in linear time for graphs of treewidth at most @xmath75 .
recall that the syntax of the _ monadic second - order logic _ ( msol ) of graphs includes logical connectives @xmath192 , @xmath193 , @xmath194 , variables for vertices , edges , sets of vertices and edges , and quantifiers @xmath195 , @xmath196 that can be applied to these variables .
besides the standard relations @xmath60 , @xmath197 , @xmath198 , the syntax includes the relation @xmath199 for two vertex variables , which expresses whether two vertex @xmath26 and @xmath21 are adjacent , and for a vertex variable @xmath21 and an edge variable @xmath200 , we have the relation @xmath201 which expresses that @xmath21 is incident with @xmath200 . the _ counting monadic first - order logic _
( cmsol ) is an extension of msol with the additional predicate @xmath202 which expresses whether the cardinality of a set @xmath203 is @xmath204 modulo @xmath205 . by the celebrated courcelle s theorem @xcite ,
any problem that can be expressed in cmsol can be solved in linear time for graphs of bounded treewidth . moreover
, this result holds also for optimization problems .
observe that to solve large euler subgraph for a graph @xmath0 , it is sufficient to find a subset of vertices @xmath206 of maximum size such that @xmath206 induces an euler graph
. clearly , @xmath206 induces an euler graph if and only if i ) @xmath207 $ ] is connected and ii ) each vertex of @xmath208 $ ] has even degree . the both properties can be expressed in cmsol .
the standard way to express connectivity is to notice that @xmath207 $ ] is connected if and only if for any @xmath209 , @xmath210 and @xmath211 , there is an edge @xmath212 such that @xmath213 and @xmath214 .
then we have to express the property that for any @xmath215 , @xmath216}(u)=|n_{g[u]}(u)|$ ] is even . to do it
, it is sufficient to observe that @xmath217}(u)$ ] if and only if @xmath218 such that i ) for any @xmath219 , @xmath220 , and ii ) for any @xmath221 such that @xmath220 , @xmath219 .
since we can express in cmsol whether @xmath222}(u)|$ ] is even , the claim follows .
now we can prove the main result of this section .
[ thm : fpt - vertex ] for any positive integer @xmath1 , large euler subgraph can be solved in linear time for undirected graphs .
clearly , we can assume that @xmath186 , as any euler graph has at least three vertices .
if @xmath223 , then we can find any shortest cycle in the input graph @xmath0 .
it is straightforward to see that if @xmath0 has no cycles , then we have no euler subgraph , and any induced cycle is an induced euler subgraph on at least three vertices .
hence , it can be assumed that @xmath10 .
we check in linear time whether @xmath189 using the bodlaender s algorithm @xcite .
if it is so , we solve our problem using lemma [ lem : cmsol ] .
otherwise , by lemma [ lem : tw - bound ] , we conclude that @xmath0 has an induced euler subgraph on at least @xmath1 vertices and return a yes - answer .
notice , that the proof of theorem [ thm : fpt - vertex ] is not constructive . in the conclusion of this section
we sketch the algorithm that produces an induced euler subgraph on at least @xmath224 vertices if it exists .
first , for each @xmath80 , we can test existence of two vertices @xmath84 such that the input graph @xmath0 has at least @xmath79 internally vertex - disjoint @xmath86-paths of length at most @xmath85 in @xmath5 time with the parameter @xmath85 using the color coding technique @xcite .
if we find such a structure for @xmath225 , we find an induced euler subgraph with at least @xmath1 vertices that is either a clique or a union of @xmath86-paths as it is explained in the proof of lemma [ lem : connect ] . otherwise , we find all 2-connected components . if there is a 2-connected component @xmath73 with a vertex @xmath26 with @xmath226 , then we find an induced euler subgraph with at least @xmath1 vertices that is a union of cycles with the common vertex @xmath26 using the arguments form the proof of lemma [ lem : deg ] .
if all 2-connected components have bounded maximum degrees , we use the algorithm of kosowski et al .
@xcite that in polynomial time either finds an induced cycle on at least @xmath1 vertices or constructs a tree decomposition of width at most @xmath227 .
in the fist case we have an induced euler subgraph on at least @xmath1 vertices .
in the second case the treewidth is bounded , and large euler subgraph is solved by a dynamic programming algorithm instead of applying lemma [ lem : cmsol ] . in this section
we show that euler @xmath1-subgraph and large euler subgraph are hard for directed graphs .
first , we consider euler @xmath1-subgraph .
it is straightforward to see that this problem is in @xmath55 , since we can check for every subset of @xmath1 vertices , whether it induces an euler subgraph .
we prove that this problem can not be solved in @xmath228 time unless @xmath229}}$ ] .
[ thm : w - hard ] the euler @xmath1-subgraph is @xmath6$]-hard for directed graphs .
we reduce the multicolored @xmath1-clique problem that is known to be @xmath6$]-hard @xcite : let @xmath230 be an instance of multicolored @xmath1-clique .
we construct the directed graph @xmath133 as follows .
* construct the copies of @xmath231 .
* for @xmath232 and for each @xmath233 and @xmath234 , if @xmath235 , then construct an arc @xmath15 for the copies of @xmath26 and @xmath21 in @xmath133 .
* for each @xmath37 , construct two vertices @xmath236 , then join @xmath177 by arcs with all vertices of @xmath237 and join every vertex of @xmath237 with @xmath238 by an arc .
* construct arcs @xmath239 .
we set @xmath240 .
we claim that @xmath0 has a clique with @xmath1 vertices if and only if @xmath133 has an induced euler subgraph with at least @xmath241 vertices .
let @xmath242 be a clique in @xmath0 , @xmath243 for @xmath37 .
observe that @xmath244 are pairwise non - adjacent in @xmath133 .
hence , the set of vertices @xmath245 induces a cycle in @xmath133 .
hence , we have an induced euler subgraph with at least @xmath241 vertices .
suppose now that @xmath133 has an induced euler subgraph @xmath41 with at least @xmath241 vertices .
observe that every directed cycle in @xmath0 contains the arc @xmath246 , because if we delete this arc , we obtain a directed acyclic graph . since any euler directed graph is a union of arc - disjoint directed cycles ( see , e.g. , @xcite ) , @xmath41 is an induced directed cycle .
moreover , for each @xmath37 , @xmath41 contains at most one vertex of @xmath237 .
indeed , assume that two vertices @xmath13 of @xmath41 are in the same set @xmath237 .
then the @xmath15-paths and the @xmath247-path in @xmath41 should contain @xmath246 , but this is impossible . because @xmath41 has @xmath248 vertices , we conclude that @xmath41 contains exactly one vertex @xmath249 from each @xmath237 for @xmath37 , and @xmath250 for @xmath37
. then @xmath251 $ ] . since @xmath41 is an induced cycle , @xmath244 are pairwise non - adjacent in @xmath133 .
then @xmath252 is a clique in @xmath0 . for large euler subgraph for directed graphs ,
we prove that this problem is @xmath58-complete .
[ thm : paranp ] for any @xmath10 , large euler subgraph is @xmath4-complete for directed graphs .
we reduce to the 3-satisfiability problem .
it is known @xcite that this problem is @xmath4-complete even if each variable is used in exactly 4 clauses : 2 times in positive and 2 times in negations .
suppose that boolean variables @xmath253 and clauses @xmath254 compose an instance of 3-satisfiability such that each variable is used in exactly 4 clauses : 2 times in positive and 2 times in negations .
let @xmath255 .
without loss of generality we assume that @xmath256 .
we construct the directed graph @xmath0 as follows ( see fig .
[ fig : npc ] ) . * for each @xmath257 , construct vertices @xmath258 and arcs @xmath259 , @xmath260 .
* for each @xmath261 , construct @xmath262 . * for each @xmath263 , construct vertices @xmath264 and arcs @xmath265 , @xmath266 , @xmath267 , @xmath268 , @xmath269 , @xmath270 , @xmath271 , @xmath272 , @xmath273 .
* for each @xmath274 , construct @xmath275 . *
construct @xmath276 .
* for each @xmath277 , let @xmath278 . for @xmath279 ,
* * if @xmath280 for some @xmath281 and @xmath282 is the fist occurrence of the literal @xmath177 in @xmath283 , then construct @xmath284 ; * * if @xmath280 for some @xmath281 and @xmath282 is the second occurrence of the literal @xmath177 in @xmath283 , then construct @xmath285 ; * * if @xmath286 for some @xmath281 and @xmath282 is the fist occurrence of the literal @xmath287 in @xmath283 , then construct @xmath288 ; * * if @xmath286 for some @xmath281 and @xmath282 is the second occurrence of the literal @xmath287 in @xmath283 , then construct @xmath289 .
we claim that @xmath283 can be satisfied if and only if @xmath0 has an induced euler subgraph with at least @xmath1 vertices . for @xmath257 ,
let @xmath290 and @xmath291 . for @xmath292 and @xmath279 , @xmath293 .
let also @xmath294 and @xmath295 .
suppose that the variables @xmath253 have values such that @xmath283 is satisfied .
we construct the set of vertices @xmath206 as follows : * for @xmath257 , if @xmath296 , then the vertices of the set @xmath89 are included in @xmath206 , and @xmath297 is included otherwise ; * for @xmath263 , if @xmath278 , then we choose a literal @xmath298 in @xmath299 and include @xmath300 in @xmath206 .
observe that @xmath206 induces a cycle in @xmath0 and @xmath301 .
then @xmath207 $ ] is an induced euler subgraph on at least @xmath1 vertices .
assume now that a set @xmath302 induces an euler graph @xmath303 $ ] and @xmath304 .
since @xmath10 , @xmath305 because @xmath306 has no cycles except vertex - disjoint triangles .
suppose that some @xmath307 .
if @xmath308 , then @xmath309 , and if @xmath310 , then @xmath311 .
also exactly one of the vertices @xmath312 is in @xmath206 .
. then @xmath314 .
further , either @xmath315 or some vertex @xmath316 or @xmath317 .
but if @xmath316 or @xmath317 , @xmath318 .
hence , @xmath315 . if @xmath319 , then @xmath320 , and if @xmath321 , then @xmath322 .
suppose now that some @xmath315 .
if @xmath319 , then @xmath320 , and if @xmath321 , then @xmath322 .
also exactly one of the vertices @xmath323 is in @xmath206 .
. then @xmath313 .
further , either @xmath307 or some vertex @xmath316 or @xmath317 .
but if @xmath316 or @xmath317 , @xmath324
. hence , @xmath307 . if @xmath310 , then @xmath311 .
we have that for each @xmath277 , exactly one @xmath325 for @xmath279 , and for @xmath326 , @xmath327 .
also we have that @xmath320 .
now we consequently consider @xmath328 .
we already know that @xmath320 .
assume inductively that @xmath329 .
then exactly one of the vertices @xmath330 is in @xmath206 .
assume that @xmath331 is in @xmath206 as another case is symmetric . then either @xmath332 or some vertex @xmath333 should be in @xmath206 .
but if @xmath313 , then because @xmath307 , @xmath334 and this is impossible for an euler graph
. then @xmath335 . by the same arguments
we show that @xmath336 . if @xmath337 , then @xmath338 , and we can proceed with our inductive arguments .
we conclude that for each @xmath339 , either @xmath340 or @xmath341 , and if @xmath340 ( @xmath341 resp . ) , then @xmath342 ( @xmath343 resp . ) . moreover , if @xmath340 ( @xmath341 resp . ) and @xmath344 , then @xmath133 has no arcs between the vertices of these two sets .
we define the truth assignment for the variables @xmath253 as follows : for each @xmath339 , if @xmath340 , then @xmath345 , and @xmath296 otherwise .
we claim that @xmath283 is satisfied by this assignment . to show it , consider a clause @xmath278 . we know that there is @xmath279 such that @xmath344 .
assume that @xmath280 ( the case @xmath286 is symmetric ) .
then @xmath346 is joined by arcs in @xmath0 with @xmath89 .
it follows that @xmath89 was not included in @xmath206 , i.e. , @xmath341 and @xmath296 .
hence , @xmath298 .
we proved that large euler subgraph is @xmath4-complete for directed graphs for @xmath10 . in the conclusion of this section
we observe that the bound @xmath10 is tight unless @xmath59@xmath60@xmath4 .
[ prop : poly ] large euler subgraph can be solved in polynomial time for @xmath3 . for @xmath347 ,
the problem is trivial . if @xmath348 , then any shortest cycle @xmath41 in a directed graph @xmath0 is an induced euler subgraph of @xmath0 with at least two vertices , and @xmath0 has no induced euler subgraph if @xmath0 is a directed acyclic graph .
hence , it remains to consider @xmath223 .
suppose that @xmath133 is an induced euler subgraph of a directed graph @xmath0 , and @xmath133 has at least three vertices .
denote by @xmath134 the graph obtained from @xmath133 by the deletion of all pairs of opposite arcs , i.e. , for each pair of vertices @xmath349 such that @xmath350 , we delete @xmath351 .
clearly , for any @xmath352 , @xmath353 . if @xmath134 is empty , then
because @xmath354 , there are three distinct vertices @xmath355 such that @xmath356 and either @xmath357 or @xmath358 . then @xmath359 $ ] is an euler subgraph of @xmath0 .
if @xmath134 is non - empty , then @xmath134 has a shortest cycle @xmath41 . because @xmath134 has no cycles of length two , @xmath360 $ ] is an induced euler subgraph with at least three vertices .
we conclude that large euler subgraph can be solved for @xmath223 as follows . if @xmath0 has three distinct vertices @xmath361 such that @xmath362 and either @xmath363 or @xmath364 , then @xmath359 $ ] is an euler subgraph of @xmath0 . otherwise , let @xmath73 be the graph obtained from @xmath0 by the deletion of all pairs of opposite arcs . we find a shortest cycle @xmath41 in @xmath73 , and we have that @xmath360 $ ] is an induced euler subgraph with at least three vertices
. finally , if @xmath73 is a directed acyclic graph , then we return a no - answer .
in this section we show that long circuit problem is @xmath5 for directed and undirected graphs . clearly , we can solve this problem in @xmath5 time for undirected graphs applying the algorithm by cai and yang @xcite for @xmath50-circuit for each @xmath365 .
for directed graphs , we can use the same approach based on the color coding technique introduced by alon , yuster and zwick @xcite . for completeness , we sketch the proof here . [
lem : atmost ] the at least @xmath1 and at most @xmath241-circuit problem can be solved in @xmath366 expected time and in @xmath367 worst - case time for ( directed ) graphs with @xmath49 vertices and @xmath368 edges ( arcs ) . as the algorithms for directed and undirected graphs are basically the same , we consider here the directed case . for simplicity ,
we solve the decision problem , but the algorithm can be easily modified to obtain a circuit of prescribed size if it exists . first , we describe the randomized algorithm .
we color the arcs of @xmath0 by @xmath241 colors @xmath369 uniformly at random independently from each other .
now we are looking for a _ colorful _ circuit in @xmath0 that has at least @xmath1 arcs , i.e. , for a circuit such that all arcs are colored by distinct colors . to do it
, we apply the dynamic programming across subsets .
we choose an initial vertex @xmath26 and try to construct a circuit that includes @xmath26 . for a set of colors
@xmath370 , denote by @xmath371 the set of vertices @xmath25 such that there is a @xmath15-trail with @xmath372 edges colored by distinct colors from @xmath203 .
it is straightforward to see that @xmath373 . for @xmath210 , @xmath374 if and only if @xmath21 has an in - neighbor @xmath375 such that @xmath376 is colored by a color @xmath377 and @xmath378 .
we consequently construct the sets @xmath371 for @xmath203 with @xmath379 elements .
we stop and return a yes - answer if @xmath380 for some @xmath203 of size at least @xmath1 .
notice that the sets @xmath371 can be constructed in time @xmath381 .
since we try all possibilities to select @xmath26 , the running time is @xmath382 .
now we observe that for any positive number @xmath383 , there is a constant @xmath384 such that after running our randomized algorithm @xmath385 times , we either get a yes - answer or can claim that with probability @xmath204 @xmath0 has no directed circuit with at least @xmath1 and at most @xmath241 arcs
. this algorithm can be derandomized by the technique proposed by alon , yuster and zwick @xcite . to do it , we replace random colorings by a family of at most @xmath386 hash functions that can be constructed in time @xmath387 .
let us recall that a fundamental cycle in undirected graph is formed from a spanning tree and a nontree edge .
for undirected graphs , it is slightly more convenient to use the structural result by gabow and nie .
[ thm : gabow - undir ] in a connected undirected graph having a cycle with @xmath1 edges , either every depth - first search tree has a fundamental cycle with at least @xmath1 edges or some cycle with at least @xmath1 edges has at most @xmath393 edges .
[ lem : bound - length ] let @xmath0 be a ( directed ) graph without cycles of length at least @xmath1 . if @xmath0 has a circuit with at least @xmath1 edges ( arcs resp . )
, then @xmath0 has a circuit with at least @xmath1 and at most @xmath394 edges ( arcs resp . ) .
let @xmath41 be a circuit in @xmath0 .
it is well - known ( see , e.g. , @xcite ) that @xmath41 is a union of edge - disjoint ( arc - disjoint ) cycles @xmath395 .
moreover , it can be assumed that for any @xmath175 , the circuit @xmath396 is connected .
suppose now that @xmath41 is a circuit with at least @xmath1 edges ( arcs resp . )
that has minimum length
. then the circuit @xmath397 has at most @xmath98 edges ( arcs resp . ) .
since @xmath0 has no cycles of length at least @xmath1 , @xmath398 has at most @xmath98 edges ( arcs resp . ) .
thus @xmath41 has at most @xmath394 edges ( arcs resp . ) .
[ thm : fpt - edge ] the long circuit problem can be solved in @xmath391 expected time and in @xmath392 worst - case time for directed graphs with @xmath49 vertices and @xmath368 arcs , and in @xmath389 expected time and in @xmath390 worst - case time for undirected graphs with @xmath49 vertices and @xmath368 edges .
first , we consider directed graphs .
let @xmath0 be a directed graph . by theorem [ thm : gabow - dir ]
, we can check whether @xmath0 has a cycle with at least @xmath1 arcs .
if we find such a cycle @xmath41 , then @xmath41 is a circuit with at least @xmath1 arcs , and we have a yes - answer .
otherwise , we conclude that each cycle in @xmath0 is of length at most @xmath98 .
then by lemma [ lem : bound - length ] , if @xmath0 has a circuit with at least @xmath1 arcs , then it has a circuit with at least @xmath1 and at most @xmath394 arcs .
we find such a circuit ( if it exist ) by solving at least @xmath1 and at most @xmath241-circuit for @xmath399 by making use of lemma [ lem : atmost ] .
combining the running times , we have that long circuit can be solved in @xmath391 expected time and in @xmath392 worst - case time . for the undirected case ,
we assume that the input graph @xmath0 is connected , as otherwise we can solve the problem for each component .
we choose a vertex @xmath21 arbitrarily and perform the depth - first search from @xmath21 . in this way we find the fundamental cycles for the dfs - tree rooted in @xmath21 , and check whether there is a fundamental cycle of length at least @xmath1 .
if we have such a cycle , then it is a circuit with at least @xmath1 edges , and we have a yes - answer .
otherwise , by theorem [ thm : gabow - undir ] , either @xmath0 has no cycles of length at least @xmath1 or @xmath0 has a cycle with at least @xmath1 and at most @xmath393 edges .
if @xmath0 has no cycles with at least @xmath1 edges , then by lemma [ lem : bound - length ] , if @xmath0 has a circuit with at least @xmath1 edges , it contains a circuit with at least @xmath1 and at most @xmath394 edges .
we conclude that if the constructed fundamental cycles have lengths at most @xmath98 , then @xmath0 either has a circuit with at least @xmath1 and at most @xmath394 edges or has no circuit with at least @xmath1 edges .
we check whether @xmath0 has a circuit with at least @xmath1 and at most @xmath394 edges by solving at least @xmath1 and at most @xmath241-circuit for @xmath399 using lemma [ lem : atmost ] .
since the depth - first search runs in linear time , we have that on undirected graphs long circuit can be solved in @xmath389 expected time and in @xmath390 worst - case time . | an undirected graph is eulerian if it is connected and all its vertices are of even degree .
similarly , a directed graph is eulerian , if for each vertex its in - degree is equal to its out - degree .
it is well known that eulerian graphs can be recognized in polynomial time while the problems of finding a maximum eulerian subgraph or a maximum induced eulerian subgraph are np - hard . in this paper
, we study the parameterized complexity of the following euler subgraph problems : * large euler subgraph : for a given graph @xmath0 and integer parameter @xmath1 , does @xmath0 contain an induced eulerian subgraph with at least @xmath1 vertices ? * long circuit : for a given graph @xmath0 and integer parameter @xmath1 , does @xmath0 contain an eulerian subgraph with at least @xmath1 edges ?
our main algorithmic result is that large euler subgraph is fixed parameter tractable ( fpt ) on undirected graphs .
we find this a bit surprising because the problem of finding an induced eulerian subgraph with exactly @xmath1 vertices is known to be w[1]-hard .
the complexity of the problem changes drastically on directed graphs . on directed graphs we obtained the following complexity dichotomy : large euler subgraph is np - hard for every fixed @xmath2 and is solvable in polynomial time for @xmath3 . for long circuit ,
we prove that the problem is fpt on directed and undirected graphs . |
multi - object tracking refers to the problem of jointly estimating the number of objects and their trajectories from sensor data .
driven by aerospace applications in the 1960 s , today multi - object tracking lies at the heart of a diverse range of application areas , see for example the texts bsf88,bp99 , mah07 , mallickbook12 , mahler2014advances .
the most popular approaches to multi - object tracking are the joint probabilistic data association filter @xcite , multiple hypothesis tracking bp99 , and more recently , random finite set ( rfs ) @xcite .
the rfs approach has attracted significant attention as a general systematic treatment of multi - object systems and provides the foundation for the development of novel filters such as the probability hypothesis density ( phd ) filter @xcite , cardinalized phd ( cphd ) filter @xcite , and multi - bernoulli filters mah07,vvc09,vvps10 .
while these filters were not designed to estimate the trajectories of objects , they have been successfully deployed in many applications including radar / sonar @xcite , tobiaslanterman08 , @xcite , computer vision mtc_csvt08,hvvs_pr12,hvv_tsp13 , cell biology @xcite , autonomous vehicle @xcite , automotive safety @xcite , sensor scheduling rv10,rvc11,hv14sencon , gostar13 and sensor network zhang_tac11,bcf_stsp13,ucj_stsp13 .
the introduction of the generalized labeled multi - bernoulli ( glmb ) rfs in @xcite has led to the development of the first tractable rfs - based multi - object tracker - the @xmath0-glmb filter .
the @xmath1-glmb filter is attractive in that it exploits the conjugacy of the glmb family to propagate forward in time the ( labeled ) multi - object filtering density exactly @xcite .
each iteration of this filter involves an update operation and a prediction operation , both of which result in weighted sums of multi - target exponentials with intractably large number of terms .
the first implementation of the @xmath0-glmb filter truncate these sums by using the @xmath2-shortest path and ranked assignment algorithms , respectively , in the prediction and update to determine the most significant components @xcite . while the original two - staged implementation is intuitive and highly parallelizable , it is structurally inefficient as it requires many intermediate truncations of the @xmath0-glmb densities . specifically , in the update , truncation is performed by solving a ranked assignment problem for each predicted @xmath0-glmb component . since truncation of the predicted @xmath0-glmb sum is performed separately from the update , in general , a significant portion of the predicted components would generate updated components with negligible weights .
hence , computations are wasted in solving a large number of ranked assignment problems , each of which has cubic complexity in the number of measurements . in this paper , we present a new implementation by formulating a joint prediction and update that eliminates inefficient truncation procedures in the original approach .
the key innovation is the exploitation of the direct relationship between the components of the @xmath0-glmb filtering densities at consecutive iterations to circumvent solving a ranked assignment problem for each predicted component .
in contrast to the original implementation , the proposed joint implementation only requires one truncation per component in the filtering density .
naturally , the joint prediction and update allows truncation of the @xmath0-glmb filtering density ( without explicitly enumerating all the components ) using the ranked assignment algorithm @xcite , @xcite , @xcite .
more importantly , it admits a very efficient approximation of the @xmath0-glmb filtering density based on markov chain monte carlo methods .
the key innovation is the use of gibbs sampling to generate significant updated @xmath0-glmb components , instead of deterministically generating them in order of non - increasing weights .
the advantages of the proposed stochastic solution compared to the rank assignment algorithm are two - fold .
first , it eliminates unnecessary computations incurred by sorting the components , and reduces the complexity from cubic to linear in the number of measurements .
second , it automatically adjusts the number of significant components generated by exploiting the statistical characteristics of the component weights .
the paper is organized as follows .
background on labeled rfs and the @xmath3-glmb filter is provided in section [ sec : bg ] .
section sec : fast_impl presents the joint prediction and update formulation and the gibbs sampler based implementation of the @xmath0-glmb filter .
numerical results are presented in section [ sec : sim ] and concluding remarks are given in section [ sec : sum ] .
this section summarizes the labeled rfs and the glmb filter implementation .
we refer the reader to the original work @xcite for detailed expositions . for the rest of the paper , single - object states
are represented by lowercase letters , e.g. @xmath4 , @xmath5 while multi - object states are represented by uppercase letters , e.g. @xmath6 , @xmath7 , symbols for labeled states and their distributions are bolded to distinguish them from unlabeled ones , e.g. @xmath5 , @xmath7 , @xmath8 , etc , spaces are represented by blackboard bold e.g. @xmath9 , @xmath10 , @xmath11 , @xmath12 , etc , and the class of finite subsets of a space @xmath9 is denoted by @xmath13 .
we use the standard inner product notation @xmath14 and the following multi - object exponential notation @xmath15 , where @xmath16 is a real - valued function , with @xmath17 by convention .
we denote a generalization of the kronecker delta that takes arbitrary arguments such as sets , vectors , etc , by @xmath18and the inclusion function , a generalization of the indicator function , by@xmath19we also write @xmath20 in place of @xmath21 when @xmath6 = @xmath22 .
a labeled rfs is simply a finite set - valued random variable where each single - object dynamical state is augmented with a unique label that can be stated concisely as follows a labeled rfs with state space @xmath9 and ( discrete ) label space @xmath11 is an rfs on @xmath23 such that each realization has distinct labels .
let @xmath24 be the projection @xmath25 , then a finite subset set @xmath7 of @xmath23 has distinct labels if and only if @xmath7 and its labels @xmath26 have the same cardinality , i.e. @xmath27 . the function @xmath28 @xmath29 is called the _ distinct label indicator_. the set integral defined for any function @xmath30 is given by@xmath31 where the integral of a function @xmath32 is : @xmath33 the notion of labeled rfs enables the incorporation of individual object identity into multi - object system and the bayes filter to be used as a tracker of these multi - object states .
suppose that at time @xmath34 , there are @xmath35 target states @xmath36 , each taking values in the ( labeled ) state space @xmath37 . in the random finite set formulation
the set of targets is treated as the _ multi - object state _
@xmath38 each state @xmath39 either survives with probability @xmath40 and evolves to a new state @xmath41 or dies with probability @xmath42 .
the dynamics of the survived targets are encapsulated in the multi - object transition density @xmath43 ) . for a given multi - object state @xmath44 , each state @xmath45 at time @xmath34
is either detected with probability @xmath46 and generates an observation @xmath47 with likelihood @xmath48 or missed with probability @xmath49 .
the _ multi - object observation _ at time @xmath34 , @xmath50 , is the superposition of the observations from detected states and poisson clutters with intensity @xmath51 . assuming that , conditional on @xmath52 , detections are independent , and that clutter is independent of the detections and is distributed as a poisson rfs , the multi - object likelihood is given by voglmb13,vvp_glmb13 @xmath53^{\mathbf{x}_k } \label{eq : rfsmeaslikelihood0}\]]where @xmath54 is a function such that @xmath55 implies @xmath56 , and @xmath57 @xmath58 is called an _ association map _ since it provides the mapping between tracks and observations , i.e. which track generates which observation , with undetected tracks assigned to 0 .
the condition @xmath55 implies @xmath56 ensures that a track can generate at most one measurement at a point of time . the image of a set @xmath59 through the map @xmath60 is denoted by @xmath61 , i.e. @xmath62 while the notation @xmath63 is used to denote the collection of all eligible association maps on domain @xmath64 , i.e. @xmath65 if the clutter is distributed as an iid cluster rfs , i.e. @xmath66^k$ ] , the multi - object likelihood is @xmath67 where @xmath68 is the association map from @xmath69 to @xmath70 and given a multi - object system as described above , the objective is to find the _ multi - object filtering density _ , denoted by latexmath:[$\mathbf{\pi } _ { k+1}(\mathbf{x }
_ { } measurements upto time @xmath34 ] , which captures all information on the number of targets and individual target states at time @xmath72 . in multi - object baysian filtering , the multi - object filtering density is computed recursively in time according to the following prediction and update , commonly referred to as _ multi - object bayes recursion _ @xcite @xmath73 note , however , that the bayes filter is intractable since the set integrals in - have no analytic solution in general .
the @xmath1-glmb rfs , a special class of labeled rfs , provides an exact solution to - .
this is because the @xmath1-glmb rfs is closed under the multi - object chapman - kolmogorov equation with respect to the multi - object transition kernel and is conjugate with respect to the multi - object likelihood function @xcite .
a @xmath1-glmb rfs is a labeled rfs with state space @xmath9 and ( discrete ) label space @xmath11 , distributed according to @xmath74 ^{\mathbf{x}},\]]where @xmath75 is a discrete space while @xmath76 and @xmath77 satisfy @xmath78 the @xmath1-glmb density is essentially a mixture of multi - object exponentials , in which each components is identified by a pair @xmath79 .
each @xmath80 is a set of tracks labels while @xmath81 represents a history of association maps @xmath82 .
the pair @xmath79 can be interpreted as the hypothesis that the set of tracks @xmath64 has a history of @xmath83 association maps and corresponding kinematic state densities @xmath84 .
the weight @xmath85 , therefore , can be considered as the probability of the hypothesis @xmath86 .
denote the collection of all label sets with @xmath87 unique elements by @xmath88 , the cardinality distribution of a @xmath1-glmb rfs is given by @xmath89 a @xmath0-glmb is completely characterized by the set of parameters @xmath90 . for implementation
it is convenient to consider the set of parameters as an enumeration of all @xmath0-glmb components ( with positive weight ) together with their associated weights and track densities @xmath91 , as shown in figure [ fig : paramtable ] , where @xmath92 and @xmath93 .
-glmb parameter set with each component indexed by an integer @xmath16 .
the hypothesis for component @xmath16 is @xmath94 while its weight and associated track densities are @xmath95 and @xmath96 , @xmath97 . ] given a @xmath1-glmb initial density , all subsequent multi - object densities are @xmath1-glmbs and can be computed exactly by a tractable filter called the @xmath0-glmb filter .
the @xmath0-glmb filter recursively propagates a @xmath0-glmb density forward in time via the bayes recursion equations and .
closed form solutions to the prediction and update of the @xmath0-glmb filter are given in the following propositions @xcite .
[ prop_ck_strong ] if the multi - target posterior at time @xmath98 is a @xmath1-glmb of the form , i.e. @xmath99^{\!\mathbf{x}_{}}\!\!\!,\end{aligned}\ ] ] and the birth density @xmath100 is defined on @xmath101 according to @xmath102^{\!\mathbf{y}},\ ] ] then the multi - target prediction density to the next time is a @xmath0-glmb given by @xmath103^{\!\mathbf{x}_{}}\!\!\!,\end{aligned}\ ] ] where@xmath104^{\!l}\!\!\!\sum_{i_{_{\!}k_{\!}-_{\!}1_{\!}}\supseteq_{_{\!}}l}\!\!\left[_{\!}1\!-\!\eta_{_{\!}s\!}^{_{\!}(_{\!}\xi_{_{\!}k_{\!}-_{\!}1\!})\!}\right]^{\!i_{_{\!}k_{\!}-_{\!}1\!}-l}\label{eq : propckstrongws } \\
\!\!\!\!\!\eta_{_{\!}s\!}^{_{\!}(_{\!}\xi_{_{\!}k_{\!}-_{\!}1\!})\!}(_{\!}\ell)\!\!\!\!\!&=&\!\!\!\!\!\left\langle p_{_{\!}s}(\cdot , \ell ) , p_{_{\!}k_{\!}-_{\!}1}^{_{\!}(_{\!}\xi_{_{\!}k_{\!}-_{\!}1\!})\!}(_{\!}\cdot , \ell)\right\rangle \label{eq : propckstrong_eta } \\ \!\!\!\!\!p_{_{\!}k_{_{\!}}|_{_{\!}}k_{\!}-_{\!}1_{\!}}^{_{\!}(_{\!}\xi_{_{\!}k_{\!}-_{\!}1\!})_{\!}}(_{\!}x,\ell_{_{\!}})\!\!\!\!\!&=&\!\!\!\!\!1_{\mathbb{l}_{\!}}(_{\!}\ell_{_{\!}})p_{_{\!}s}^{_{\!}(_{\!}\xi_{_{\!}k_{\!}-_{\!}1\!})_{\!}}(_{\!}x,\ell_{_{\!}})+1_{_{\!}\mathbb{b}_{\!}}(_{\!}\ell_{_{\!}})p_{_{\!}b_{_{\!}}}(_{\!}x,\ell_{_{\!}})\label{eq : propckstrongpp } \\
\!\!\!\!\!p_{_{\!}s}^{_{\!}(_{\!}\xi_{_{\!}k_{\!}-_{\!}1\!})_{\!}}(_{\!}x,\ell_{_{\!}})\!\!\!\!\!&=&\!\!\!\!\!\frac{\left\langle p_{_{\!}s}(\cdot , \ell)f_{_{\!}k_{_{\!}}|_{_{\!}}k_{\!}-_{\!}1_{\!}}(x|\cdot , \ell ) , p_{_{\!}k_{\!}-_{\!}1}^{_{\!}(_{\!}\xi_{_{\!}k_{\!}-_{\!}1\!})\!}(\cdot , \ell ) \right\rangle}{\eta_{_{\!}s\!}^{_{\!}(_{\!}\xi_{_{\!}k_{\!}-_{\!}1\!})\!}(\ell ) } \label{eq : propckstrongps}\end{aligned}\ ] ] [ propbayes_strong ] given the prediction density in , the multi - target posterior is a @xmath0-glmb given by@xmath105 ^{\!\mathbf{x}_{}}\!\!\ ! , \label{eq : propbayes_strong0}\end{aligned}\ ] ] where @xmath106^{\!i_{_{\!}k } } , \label{eq : propbayes_strong1 } \\
\eta_{_{\!}z_{_{\!}k}}^{_{\!}(_{\!}\theta_{_{\!}k\!})}(\ell)\!\!\ ! & = & \!\!\!\left\langle p_{_{\!}k_{_{\!}}|_{_{\!}}k_{\!}-_{\!}1{\!}}^{_{\!}(_{\!}\xi_{_{\!}k_{\!}-_{\!}1\!})\!}(\cdot , \ell ) , \psi_{_{\!}z_{_{\!}k}\!}(\cdot , \ell ; \theta_{_{\!}k\ ! } ) \right\rangle , \label{eq : propbayes_strong2 } \\
p_{k}^{_{\!}(_{\!}\xi_{_{\!}k_{\!}-_{\!}1\!},\theta_{_{\!}k\!})}(x,\ell |z_k)\!\!\ ! & = & \!\!\!\frac{p_{_{\!}k_{_{\!}}|_{_{\!}}k_{\!}-_{\!}1{\!}}^{_{\!}(_{\!}\xi_{_{\!}k_{\!}-_{\!}1\!})\!}(x,\ell)\psi_{_{\!}z_{_{\!}k}\!}(x,\ell;\theta_{_{\!}k\!})}{\eta_{_{\!}z_{_{\!}k}}^{_{\!}(_{\!}\theta_{_{\!}k\!})}(\ell ) } .
\label{eq : propbayes_strong3}\end{aligned}\ ] ] the propagations of @xmath1-glmb components through prediction and update are illustrated in fig .
[ fig : pred_propag ] and fig .
[ fig : upd_propag ] , respectively .
it is clear that the the number of components grows exponentially with time .
specifically , a component in the filtering density at time @xmath98 generates a large number of predicted components , of which each one in turn produces a new set of multiple @xmath1-glmb components in the filtering density at time @xmath34 .
hence , it is necessary to reduce the number of @xmath1-glmb components in both prediction and update densities at every time step .
-glmb prediction @xcite : component @xmath16 in the prior generates a large set of predicted components with @xmath107 , i.e. @xmath108 , , @xmath109 , and @xmath110 . ] -glmb update @xcite : component @xmath16 in the predicted density generates a ( large ) set of update components with @xmath111 and weights @xmath112, ,@xmath113 . ]
the simplest way to truncate a @xmath1-glmb density is discarding components with smallest weights .
the following proposition asserts that this strategy minimizes the @xmath114-distance between the true density and the truncated one @xcite [ prop_l1_error]let @xmath115 denote the @xmath114-norm of @xmath30 , and for a given @xmath116 let @xmath117 ^{\mathbf{x}}\]]be an unnormalized @xmath0-glmb density .
if @xmath118 then@xmath119
in this section , we briefly review the original implementation of the @xmath1-glmb filter in subsection [ subsec : orig_scheme ] and propose a new implementation strategy with joint prediction and update in subsection [ subsec : new_scheme ] .
the first implementation of the @xmath1-glmb filter , detailed in @xcite , recursively calculates the filtering density by sequentially computing the predicted and update densities at each iteration based on proposition [ prop_ck_strong ] and proposition [ propbayes_strong ] . since direct implementation of equations and
is difficult due to the sum over supersets in , the predicted and update densities are rewritten as and , respectively , with @xmath120^{\!i_{_{\!}k_{\!}-_{\!}1\!}-_{\!}l\!}\!\left[_{\!}\eta_{_{\!}s}^{_{\!}(_{\!}\xi_{_{\!}k_{\!}-_{\!}1\!})\!}\right]^{\!l\!}.\ ] ] c _ _ k__|__k_-_1(___|__z__k_-_1)=_(__)_i__k_-_1,__k_-_1 , l , j 1___(_i__k_-_1)(_l_)1___(___)(_j)__k_-_1^_(_i__k_-_1,__k_-_1)__s^_(_i__k_-_1,__k_-_1)(_l_)__b(_j)___lj__((__))^_[eq : propck_strong3 ] + _ _ k(__z__k)= [ eq : propbayes_strong4 ] in the prediction stage , each component @xmath121 with weight @xmath122 generates a set of prediction components @xmath123 with weight @xmath124 where @xmath125 and
@xmath126 represent two disjoint label sets for survival and birth tracks , respectively .
since the weight of the prediction component can be factorized into two factors , @xmath127 and @xmath128 , which depend on two mutually exclusive sets ; truncating the predicted density is performed by solving two separate @xmath2-shortest path problems for each set of tracks .
this is because running only one instance of the @xmath2-shortest path based on the augmented set of existing and birth tracks generally favours the selection of survival tracks over new births and typically results in poor track initiation . in the update stage ,
each prediction component @xmath123 generates a ( large ) set of update components @xmath129 .
these update components are truncated without having to exhaustively compute all the components by solving a ranked assignment problem .
although the original two - staged implementation is intuitive and highly parallelizable , it is has several drawbacks .
first , since truncation of the predicted @xmath0-glmb density is performed separately from the update based purely on _ a priori _ knowledge ( e.g. survival and birth probabilities ) , in general , a significant portion of the predicted components would generate updated components with negligible weights .
hence , computations are wasted in solving a large number of ranked assignment problems , each of which has cubic complexity in the number of measurements .
second , it would be very difficult to determine the final approximation error of the truncated filtering density as the implementation involves least three separate truncating processes : one for existing tracks , one for birth tracks , and one for predicted tracks . in the following subsections
, we will introduce the joint prediction and update as a better alternative to the original two - staged approach .
the joint strategy eliminates the need for separate prediction truncating procedures , thus involves only one truncation per iteration .
consequently , the new implementation yields considerable computational savings while preserving the filtering performance as well as the parallelizability of the original implementation . instead of computing the filtering density in two steps ,
the new strategy aims to generate the components of the filtering density in one combined step by formulating a direct relationship between the component of the current filtering density with those of the previous density .
specifically , we will derive a new formulation for @xmath130 that does not involve prediction induced variables @xmath125 and @xmath126 .
this can be done via an _ extended association map _ , denoted by @xmath131 , and defined as follows the extended association map is a function @xmath132 such that @xmath133 for @xmath134 implies @xmath56 .
the new map , in essence , only extends the original map to include a new association , @xmath135 .
in particular , @xmath131 is identical to @xmath136 except for non - survival and unconfirmed birth tracks , i.e. @xmath137 the image of a set @xmath138 through the extended map @xmath139 and the collection of all eligible extended association maps on domain @xmath64 are denoted by @xmath140 and @xmath141 , respectively .
based on the notion of extended association map , the following proposition establishes the direct relationship between two consecutive filtering densities at time @xmath34 and @xmath98 . for simplicity
, we assume that target births are modeled by ( labeled ) multi - bernoulli rfs s , i.e. @xmath142^{\mathbb{b}-j}[r(\cdot)]^j$ ] with @xmath143 denotes the existence probability of track @xmath144 .
[ prop_joint ] if the multi - target posterior at time @xmath98 is a @xmath0-glmb of the form and the set of targets born at the next time is distributed as a labeled multi - bernoulli rfs , then the multi - target posterior at the next time is a @xmath0-glmb given by @xmath145^{\!\mathbf{x}_{}}\end{aligned}\ ] ] where @xmath146 and @xmath147^{\!i_{_{\!}k_{\!}-_{\!}1\!}\cup_{_{\ ! } } \mathbb{b}}\\ & \!\gamma_{_{\!}z_{_{\!}k}}^{_{\!}(_{\!}\tilde{\theta}_{_{\!}k\!}(_{\!}\ell^{_{\!}})_{\!})\!}(_{\!}\ell^{_{\!}})\!=\!\begin{cases } 1_{\!}-_{\!}{\eta}_{_{\!}s}^{_{\!}(_{\!}\xi_{k\!-\!1\!}^{})}\!(_{\!}\ell^{_{\ ! } } ) & \!\forall\ell\!\in\ ! i_{_{\!}k_{\!}-_{\!}1}\!\ ! : \tilde{\theta}_{_{\!}k_{\!}}(_{\!}\ell^{_{\!}})\!=\!|z_{\!k}|\!+\!1 , \\
{ \eta}_{_{\!}s}^{_{\!}(_{\!}\xi_{k\!-\!1\!}^{})}\!(_{\!}\ell^{_{\!}}){\eta}_{_{\!}z_{_{\!}k}}^{_{\!}(_{\!}{\theta}_{_{\!}k\!})\!}(_{\!}\ell^{_{\ ! } } ) & \!\forall\ell\!\in\!i_{_{\!}k_{\!}-_{\!}1}\!\ ! : \tilde{\theta}_{_{\!}k_{\!}}(_{\!}\ell^{_{\!}})\!<\!|z_{\!k}|\!+\!1 , \\ 1_{\!}-_{\!}r(_{\!}\ell^{_{\ ! } } ) & \!\forall\ell\!\in\!\mathbb{b}\ ! : \tilde{\theta}_{_{\!}k_{\!}}(_{\!}\ell^{_{\!}})\!=\!|z_{\!k}|\!+\!1 , \\ r(_{\!}\ell^{_{\!}}){\eta}_{_{\!}z_{_{\!}k}}^{_{\!}(_{\!}{\theta}_{_{\!}k\!})\!}(_{\!}\ell^{_{\ ! } } ) & \!\forall\ell\!\in\!\mathbb{b}\ ! : \tilde{\theta}_{_{\!}k_{\!}}(_{\!}\ell^{_{\!}})\!<\!|z_{\!k}|\!+\!1
. \label{eq : gamma_def } \end{cases}\end{aligned}\ ] ] we now proceed to detail an efficient implementation of the @xmath1-glmb filter based on the result in proposition [ prop_joint ] .
let the sets of existing tracks , birth tracks , and measurements be enumerated by @xmath148 , @xmath149 , and @xmath150 , respectively .
given the @xmath16-th component @xmath151 , the objective is to find @xmath152 extended associations @xmath153 that produce highest update weights .
denote by @xmath154 a @xmath155 matrix whose entries are either @xmath156 or @xmath157 such that the sum of each row is exactly @xmath156 while the sum of each column is at most @xmath156 .
the matrix @xmath154 is called an _
assignment matrix _ since it represents a valid extended association map @xmath158 .
hence , finding the desirable extended associations is equivalently translated to finding the corresponding assignment matrices . the simplest way to determine these matrices without exhaustingly computing
all of the update weights is to assign an appropriate cost for each assignment matrix and then rank these matrices in non - decreasing order of their costs via murty s algorithm .
let @xmath159 be a matrix whose @xmath160 entry , with @xmath161 and @xmath162 , is defined as follows . @xmath163 as depicted in fig .
[ fig : contr_mat ] , the matrix @xmath159 contains all possible values of @xmath164 that any valid extended associations @xmath165 can generate .
intuitively speaking , each entry of @xmath159 is an indicator of how likely an extended association is assigned to a track ; hence , the matrix @xmath166 is called the cost matrix and cost function for a particular assignment matrix @xmath154 is given by @xmath167 it is straightforward to show that the ( unnormalized ) weight of the @xmath168-generated component in the filtering density at time @xmath34 is @xmath169.\ ] ] based on and , an implementation of the @xmath1-glmb filter with joint prediction and update is given in algorithm [ alg : joint_murty ] , where the subroutine @xmath170 uses murty s algorithm @xcite to generate a sequence of @xmath152 assignment matrices that yield lowest costs ( or equivalently , highest update weights ) without exhaustively navigating the whole assignment space . @xmath171 @xmath172 compute @xmath173 according to @xmath174 @xmath175 compute @xmath176 according to compute @xmath177 according to normalize weights @xmath178 similar to the original implementation , the joint prediction and update also operates independently on each components in the filtering density , thereby is highly parallelizable . in multi - target tracking ,
truncating procedures based on the original murty s algorithm with complexity @xmath179 , where @xmath180 is the number of assignments and @xmath181 is the number of the measurements , have been proposed in danchicknewnam93 , @xcite , @xcite .
more efficient algorithms with @xmath182 complexity have been proposed in @xcite , pascoaletal03 , @xcite , with the latter showing better efficiency for large @xmath183 .
the main drawback of these approaches is that a significant amount of computation is used to sort the data associations in a particular order despite the fact that order is effectively discarded after the update .
furthermore , the number of desired components must be predetermined , generally by a large enough number to capture all important associations .
if the number of significant components in the filtering density is much smaller than the chosen threshold , many insignificant components are generated that waste a lot of computation at the next iteration .
conversely if the number of significant component exceeds the chosen threshold , the filtering performance will likely degrades in subsequent iterations .
nonetheless a deterministic polynomial time solution is thus appealing in the sense that convergence and reproducibility is guaranteed without having to enumerate all possible solutions . in this subsection
, we propose an alternative to the ranked assignment based solution .
our proposed solution is based on stochastic simulation or markov chain monte carlo methods via a gibbs sampler which directly address the above mentioned drawbacks .
conceptually , instead of ranking extended associations in non - increasing order of their weights , each extended association is treated as a realization of a ( discrete ) random variable , where the probability of each extended association is proportional to the weight of its associated @xmath1-glmb component in the next filtering density .
candidate extended associations are then generated by sampling from this discrete distribution . extended associations with high weights are chosen more often than those with low weights in a statistically consistent manner .
consequently these samples of extended associations are more statistically diverse than those from obtained from a deterministic approach such as the ranked optimal assignment . using the same enumeration for tracks and measurement as in the previous section , a valid extended association map @xmath131 for each component
@xmath184 can be represented as a vector @xmath185^t \in \{0,1,\ldots , m\!+\!1\}^{p}$ ] .
the key idea of the stochastic based approach is that @xmath131 can be considered as realizations of a random variable in the space @xmath186 with the following distribution @xmath187where @xmath188,\label{eq : theta_req}\\ \omega _ { k}^{(i_{k-1}^{(h)},\xi_{k-1}^{(h)},\tilde{\theta})}&\propto \omega _ { k\!-\!1}^{(i_{k\!-\!1}^{(h)},\xi _ { k\!-\!1}^{(h)})}\!\left[\gamma_{_{\!}z_{_{\!}k}\!}^{_{\!}(_{\!}\tilde{\theta}(\cdot)_{\!})\!}(\cdot ) \right ] ^{\!i_{k\!-\!1}\cup \mathbb{b}},\label{eq : omeg_theta}\end{aligned}\]]with @xmath189 and @xmath190 is given in .
thus the probability of a valid extended association is proportional to the weight of the corresponding @xmath1-glmb component in the next filtering density while zero probability is allocated to extended associations which do not satisfy the constraint that each measurement is assigned to at most one track .
however , sampling directly from the distribution is very difficult since we can not exhaustively compute all of the values of @xmath191 .
a common solution to this kind of problem is to use markov chain monte carlo ( mcmc ) methods such as the gibbs sampler to obtain samples from without having to directly compute @xmath191 .
the gibbs sampler is a very efficient method to sample a difficult distribution if its conditional marginals can be computed in a simple closed form @xcite with proven convergence under generally standard assumptions @xcite .
the main theoretical contribution in this section is stated in the following proposition , which allows conditional marginals to be computed via the entries of the matrix @xmath159 .
[ marg_cond ] denote by @xmath192 the @xmath87-th element of @xmath131 , i.e. @xmath193 , and @xmath194 all the other elements except @xmath192 , i.e. @xmath195^t$ ] .
then , the conditional marginal @xmath196 is given by @xmath197\!\gamma_{_{\!}z_{_{\!}k}}^{_{\!}(_{\!}\tilde{\theta}_{_{\!}n\!})\!}(_{\!}\ell_{_{\!}n\!}).\ ] ] propostion [ marg_cond ] simply states that the conditional probability @xmath196 are zero when @xmath198 is inconsistent with @xmath199 , thus ensuring that each measurement can be assigned to at most on track .
the conditional probabilities are otherwise proportional to @xmath200 . with these conditionals the pseudo code for @xmath0-glmb filtering with the gibbs sampler
is given in algorithm [ alg : gibbs ] where the gibbs sampler is used directly in place of the ranked optimal assignment .
@xmath201 @xmath202 @xmath203 compute @xmath173 according to initialize @xmath204^t\gets\left[\tilde{\theta}_{_{\!}k}^{_{\!}(_{\!}h,0_{\!})\!}(_{\!}\ell_{_{\!}1\ ! } ) , \ldots,\tilde{\theta}_{_{\!}k}^{_{\!}(_{\!}h,0_{\!})\!}(_{\!}\ell_{_{\!}p\!})\right]^t$ ] compute @xmath205 according to @xmath206 @xmath207 @xmath208 @xmath209 $ ] @xmath210 in order to produce one sample , algorithm [ alg : gibbs ] requires a gibbs sequence of length @xmath211 , starting from an arbitrary initialization @xmath212^t$ ] , to be generated .
alternatively , we can sample a long gibbs sequence of length @xmath213 and then extract every @xmath211-th sample @xcite .
the length of the gibbs sequence , roughly speaking , depends on the convergence rate of the gibbs sampler and the distance from the initial point to the true sample space .
if we start with a good initialization right in the true sample space , we can use all the samples from the gibbs sequence @xcite . in practice ,
one example of good initialization that allows us to use all of the samples from the resulting gibbs sequence is the optimal assignment , which can be obtained via either munkres @xcite or jonker - volgenant algorithm @xcite .
otherwise , we can start with all zeros assignment ( i.e. all tracks are misdetected ) that is also valid sample and requires no additional computation . in terms of computational complexity ,
sampling from a discrete distribution is linear with the weight s length @xcite , therefore the total complexity of the gibbs sampling procedure presented in algorithm alg : gibbs is @xmath214 . in comparison ,
the fastest ranked optimal assignment algorithm is @xmath215 @xcite . for general multi - target tracking problems in practice
, we usually have @xmath216 , thus the gibbs sampling algorithm will generally be much faster than the ranked assignment given the same @xmath217 .
in this section we first compare the performance of the joint prediction and update approach with its traditional separated counterpart , both employ the ranked assignment algorithm for fair comparison .
then , we illustrate the superior performance of the gibbs sampler based truncation to the conventional ranked assignment via a difficult tracking scenario with low detection probability and very high clutter rate .
the first numerical example is based on a scenario adapted from vvp_glmb13 in which a varying number targets travel in straight paths and with different but constant velocities on the two dimensional region @xmath218
m \times [ -1000,1000]m $ ] .
the duration of the scenario is @xmath219 .
there is a crossing of 3 targets at the origin at time @xmath220 , and a crossing of two pairs of targets at position @xmath221 at time @xmath222 .
the region and tracks are shown in figure [ fig : traj ] . plane .
start / stop positions for each track are shown with @xmath223/@xmath224 . ]
the kinematic target state is a vector of planar position and velocity @xmath225^t$ ] .
measurements are noisy vectors of planar position only @xmath226^t$ ] .
the single - target state space model is linear gaussian according to transition density @xmath227 and likelihood @xmath228 with parameters @xmath229 where @xmath230 and @xmath231 denote the @xmath232 identity and zero matrices respectively , @xmath233 is the sampling period , @xmath234 and @xmath235 are the standard deviations of the process noise and measurement noise .
the survival probability is @xmath236 = 0.99 and the birth model is a labeled multi - bernoulli rfs with parameters @xmath237 where @xmath238 and @xmath239 with @xmath240^t , m^{\!(\!2\!)}_b= [ -100,0,-100,0]^t , m^{\!(\!3\!)}_b= [ 100,0,-100,0]^t , p_b = \text{diag } ( [ 10 , 10 , 10 , 10 ] ^t)^2 $ ] .
the detection probability is @xmath241 and clutter follows a poisson rfs with an average intensity of @xmath242 giving an average of @xmath243 false alarms per scan .
first , we compare the performance of the traditional separated and the proposed joint prediction and update approaches . for a fair comparison , both approaches are capped to the same maximum components .
results are shown over 100 monte carlo trials .
figures [ fig : card ] shows the mean and standard deviation of the estimated cardinality versus time .
figures fig : ospa_comb and [ fig : ospa_sep ] show the ospa distance @xcite and its localization and cardinality components for @xmath244 and @xmath245 .
it can be seen that both approaches estimate the cardinality equally well .
similarly , in terms of ospa distance , the performance of the two approach is virtually the same .
second , we demonstrate the fast implementation via the gibbs sampler . in this example , we keep all parameters the same as in the previous example except that the clutter rate is now increased to average @xmath246 false alarms per scan .
the performance of the gibbs sampler implementation is compared with that of a ranked assignment based implementation with the same maximum number of posterior hypotheses .
the average ospa distances over 100 monte carlo trials are presented in fig .
[ fig : ospa_100 ] . ) . ]
it is obvious that the gibbs sampler has a better ospa from around time @xmath247 onward .
the reason is in difficult scenario ( e.g. high clutter rate , low detection probability ) , if the number of existing targets are high the gibbs sampling technique is expected to pick up the new born target better than the ranked assignment algorithm given the same number of samples / hypotheses due to its randomized behaviour .
this is clearly illustrated in the cardinality statistics for both approaches in fig .
fig : card_gibbs and fig .
[ fig : card_murty ] . as expected , however , the joint approach averaged run time is significantly lower that that of the original approach .
reductions in execution time of 1 to 2 orders of magnitude are typical .
in this paper we propose a new implementation scheme for the @xmath1-glmb filter that allows joint prediction and update .
in contrast to the conventional two - staged implementation , the joint approach use _ a posteriori _ information to construct cost matrices for every individual track , thereby requires only one truncation in each iteration due to the elimination of inefficient intermediate steps . more importantly
, this joint strategy provides the platform for the development of an accelerated randomized truncation procedure that achieves superior performance as compared to that of its traditional deterministic counterpart .
the proposed method is also applicable to approximations of the @xmath1-glmb filter such as the labeled multi - bernoulli ( lmb ) filter @xcite .
m. tobias and a. d. lanterman , `` probability hypothesis density - based multitarget tracking with bistatic range and doppler observations , '' _ iee proceedings - radar , sonar and navigation _ , vol .
152 , no . 3 , pp .
195205 , june 2005 . d. e. clark and j. bell , `` bayesian multiple target tracking in forward scan sonar images using the phd filter , '' _ iee proceedings - radar , sonar and navigation _ , vol .
152 , no . 5 , pp .
327334 , october 2005 .
r. hoseinnezhad , b .-
vo , and b. t. vo , `` visual tracking in background subtracted image sequences via multi - bernoulli filtering , '' _ ieee trans .
signal process .
_ , vol .
61 , no . 2 ,
392397 , jan 2013 .
s. rezatofighi , s. gould , b. vo , b. vo , k. mele , and r. hartley , `` multi - target tracking with time - varying clutter rate and detection profile : application to time - lapse cell microscopy sequences , '' _ ieee trans . med .
_ , vol .
34 , no . 6 , pp .
13361348 , 2015 . c. lundquist , l. hammarstrand , and f. gustafsson , `` road intensity based mapping using radar measurements with a probability hypothesis density filter , '' _ ieee trans . signal process .
_ , vol .
59 , no . 4 , pp . 13971408 , april 2011 .
g. battistelli , l. chisci , s. morrocchi , f. papi , a. benavoli , a. di lallo , a. farina , and a. graziano , `` traffic intensity estimation via phd filtering , '' in _ proc .
2008 european radar conference ( eurad ) _ , oct 2008 , pp . 340343 .
a. gostar , r. hoseinnezhad , and a. bab - hadiashar , `` robust multi - bernoulli sensor selection for multi - target tracking in sensor networks , '' _ ieee signal process .
_ , vol . 20 , no . 12 , pp .
11671170 , dec 2013 .
g. battistelli , l. chisci , c. fantacci , a. farina , and a. graziano , `` consensus cphd filter for distributed multitarget tracking , '' _ ieee j. sel .
topics signal process . _ , vol . 7 , no . 3 , pp .
508520 , june 2013 .
m. pascoal , m. captivo , and j. clmaco , `` a note on a new variant of murty s ranking assignments algorithm , '' _
4or : quarterly journal of the belgian , french and italian operations research societies _ , vol . 1 , no . 3 , pp .
243255 , 2003 .
i. cox and s. hingorani , `` an efficient implementation of reid s multiple hypothesis tracking algorithm and its evaluation for the purpose of visual tracking , '' _ ieee trans . pattern anal .
_ , vol . 18 , no . 2 , pp .
138150 , 1996 .
a. frigessi , p. d. stefano , c .- r .
hwang , and s .- j .
sheu , `` convergence rates of the gibbs sampler , the metropolis algorithm and other single - site updating dynamics , '' _ journal of the royal statistical society .
series b ( methodological ) _ , vol .
55 , no . 1 ,
pp . 205219 , 1993 .
g. roberts and a. smith , `` simple conditions for the convergence of the gibbs sampler and metropolis - hastings algorithms , '' _ stochastic processes and their applications _ ,
49 , no . 2 ,
pp . 207216 , 1994 .
c. j. geyer and e. a. thompson , `` constrained monte carlo maximum likelihood for dependent data , '' _ journal of the royal statistical society .
series b ( methodological ) _ , vol .
54 , no . 3 , pp .
657699 , 1992 . | this paper proposes an efficient implementation of the generalized labeled multi - bernoulli ( glmb ) filter by combining the prediction and update into a single step .
in contrast to the original approach which involves separate truncations in the prediction and update steps , the proposed implementation requires only one single truncation for each iteration , which can be performed using a standard ranked optimal assignment algorithm . furthermore , we propose a new truncation technique based on markov chain monte carlo methods such as gibbs sampling , which drastically reduces the complexity of the filter .
the superior performance of the proposed approach is demonstrated through extensive numerical studies .
random finite sets , delta generalized labeled multi - bernoulli filter |
light harvesting complexes are the part of photosynthetic systems that channel energy from the antenna to the reaction centre .
one of the most studied complexes is the fenna - matthews - olson ( fmo ) complex @xcite which is part of the photosynthetic apparatus of green sulphur bacteria .
the observation of long lasting beatings , up to @xmath0 ps at @xmath1 k in time - resolved two - dimensional ( 2d ) spectra @xcite has generated enormous interest in theoretical modelling the process of energy transfer through light - harvesting complexes .
recently molecular - dynamics simulations started to develop a more microscopic picture of the dynamics @xcite , but an atomistic and fully quantum - mechanical model of the whole process reproducing the experimentally obtained spectra is not available .
therefore theoretical models treat the light - harvesting pigment - protein complex as an exciton system coupled to a bath which accounts for the vibrational modes of the pigments .
the choreography of the exciton - energy transfer can be elucidated with 2d echo - spectroscopy using three laser pulses which hit the sample within several femtoseconds time spacings .
the experimentally measured 2d echo - spectra of the complex show a variety of beating patterns with oscillation periods roughly consistent with the differences in eigenenergies of the excited system obtained from fits to the absorption and 2d echo - spectra @xcite . at long delay times between pump and probe pulses the system displays relaxation to the energetically lowest lying exciton eigenstates as expected from the interactions between the electronic degrees of freedom and the vibrational modes of the complex which drive the system into a state of thermal equilibrium @xcite
. one main goal of the research in this field is to clarify the role of a disordered and fluctuating environment in the energy - transfer efficiency .
theoretical quantum - transport studies show that the energy transfer in these systems results from a balance between coherent and dissipative dynamics @xcite .
calculations at room temperature also show that the optimal energy - transport efficiencies appear as a compromise between coherent dynamics and thermal fluctuations @xcite .
another open question in the theoretical study of these systems is the description of experimental observables , i.e. the absorption and the 2d echo - spectra .
most of the theoretical studies analyse the beatings in the population dynamics and compare them with the ones observed in the experimentally measured 2d echo - spectra . however , the 2d echo - spectrum reflects different information than the population dynamics and needs to be calculated and analysed separately in detail . in principle techniques such as
quantum state and process tomography made out of a sequence of 2d echo - spectra can be used to map out the complete density matrix @xcite .
in contrast to energy - transfer efficiency studies where an initial excitation enters the complex at specific sites close to the antenna , in 2d echo - spectra the whole complex is simultaneously excited .
also the two - exciton manifold yields prominent contributions to the signal , resulting in negative regions in the 2d echo - spectra . the non - pertubative calculation of 2d echo - spectra presents a considerable computational challenge due to the presence of two excitons giving raise to excited state absorption and the requirement to consider ensemble averages over differently orientated complexes with slightly varying energy levels .
previous calculations have employed markovian approximations @xcite or exclude the double exciton manifold @xcite .
in addition the systematic study of beatings in a series of 2d echo - spectra requires effective means to calculate a huge number of such spectra .
so far no theoretical method has been able to describe the long - lasting beatings in the time - resolved 2d spectra @xcite .
one possible explanation for the persistence of long coherence times has been the sluggish absorption of the reorganisation energy by the molecule , which requires theoretical descriptions that go beyond the markovian approximation and the rotating wave approximation @xcite .
the hierarchical equations of motions ( heom ) , first developed by tanimura and kubo @xcite and subsequently refined in @xcite , show oscillations in the dynamics of the exciton populations that persist even at temperature @xmath2 k @xcite .
the heom include the reorganisation process in a transparent way and are directly applicable for computations at physiological temperatures .
a recent calculation at temperature @xmath3 k of the 2d echo - spectra with the heom method has been performed by chen @xcite , which does not display the strong beatings observed experimentally .
chen conclude that the agreement between theory and experiments needs to be improved . here , we analyse under what conditions the theoretical calculations give better agreement with the experimental results .
we calculate the absorption and 2d echo - spectra in a temperature range between @xmath3 k to @xmath4 k using the heom and investigate the role of pigment and protein vibrations in the calculated spectra .
we systematically discuss the parameters needed to obtain results that agree better with the experimentally measured spectra .
we discuss the role of static disorder for the shapes of the peaks and the appearance of beatings in the time - resolved 2d echo - spectrum and their robustness against temperature changes and static disorder .
the paper is organised as follows : the exciton model and the heom are introduced in sec .
[ sec : model ] .
the absorption spectra and the line shapes obtained from the heom model of the fmo complex are analysed in sec .
[ sec:1d ] where we compare different methods for performing rotational averages . in sec .
[ sec:2d ] we discuss the 2d spectrum and analyse how disorder affects it .
furthermore we analyse the persistence of beatings in the spectra as a function of the delay time for several temperatures and with and without static disorder .
the fmo complex described by fenna , matthews and olson @xcite is a pigment - protein complex that channels the energy in the photosynthetic apparatus in green sulphur bacteria .
the fmo complex is arranged as a trimer with the different subunits interacting weakly with each other .
we thus restrict our study to a single subunit which contains eight bacteriochlorophyll molecules ( bchls ) @xcite .
the bchls pigments , which are wrapped in a protein environment , can be electronically excited and are coupled due to dipole - dipole interactions .
the bchls form a network that guides the energy from the antenna to the reaction centre .
since the eighth bchl is only loosely bound it usually detaches from the others when the system is isolated from its environment to perform experiments @xcite .
therefore , we do not take the eighth bchl into account and describe the energy transfer using the seven site frenkel exciton model @xcite .
the single exciton manifold is given by @xmath5 with @xmath6 .
the state @xmath7 corresponds to an electronic excitation of bchl@xmath8 with site energy @xmath9 .
dipole - dipole couplings between different chromophores yield the inter - site couplings @xmath10 , which lead to a delocalisation of the exciton over all bchls .
the site energies of the bchls @xmath11 consist of ` zero - phonon energies ' @xmath12 , and a term due to the reorganisation hamiltonian @xmath13 .
the reorganisation energy @xmath14 denotes the energy which is removed from the system to adjust the vibrational coordinates to the new equilibrium value @xcite .
here , @xmath15 denotes a dimensionless displacement of the @xmath16th oscillator measured in units of @xmath17 , where @xmath18 and @xmath19 are mass and frequency of the @xmath16th oscillator .
the vibrational modes of the bchl are modelled as a set of harmonic oscillators @xmath20 .
the vibrational environment couples linearly to the exciton system @xmath21 and induces fluctuations in the site energies .
we neglect correlations between the vibrations at different sites and model the coupling strength by a structureless spectral density @xmath22 where we assume identical couplings @xmath23 at each site .
this choice of spectral density facilitates the use of the heom method since the time - dependent bath correlations then assume an exponential form . for the fmo complex a variety of spectral densities are discussed in the literature , where some are a parametrisation of experimental fluorescence spectra @xcite , while others are extracted from molecular dynamics simulations @xcite .
we use the model hamiltonian for _ chlorobium tepidum _ given in ref . @xcite .
the hamiltonian originates from previous calculations in @xcite .
cho et al . @xcite changed the coupling constant between bchl 5 and bchl 6 and obtained new site energies from a fit to experimental results .
the hamiltonian of ref .
@xcite has been used in previous studies of 2d echo - spectra @xcite and facilitates comparisons between the different approaches . in the site basis
@xmath24 the matrix elements read @xmath25 in units of @xmath26 , where we have added an energy shift @xmath27 @xmath26 to the diagonal elements of the exciton hamiltonian @xmath28 to shift the positions of the eigenenergies to the frequency range of the figures in ref .
we set the parameters of the spectral density to @xmath29 @xmath26 and @xmath30 fs .
these parameters are chosen to provide best agreement to the smooth part of the spectral density in ref .
@xcite , where the low frequency component of the spectral density was fitted to fluorescence line narrowing spectra @xcite . since a single lorentz - drude peak is not sufficient to accurately reproduce the smooth part of the spectral density of ref .
@xcite , we choose the parameters to obtain agreement within a factor of @xmath31 to @xmath32 in the energy range from @xmath33 to @xmath34 @xmath26 , which corresponds to the typical difference between the exciton energies .
the heom approach rewrites time non - local effects into a hierarchy of coupled equations of motion for a set of auxiliary matrices @xmath35 .
the superscript @xmath36 is a vector of @xmath37 entries , where @xmath38 gives the order to which the bath of bchl@xmath8 is taken into account .
the reduced density matrix @xmath39 describing the exciton system is obtained after tracing out the vibrational degrees of freedom .
the time evolution of the reduced density matrix is given by @xmath40 with @xmath41 @xmath42 is the liouville operator of the coherent exciton dynamics and we define @xmath43 as the @xmath44th unit vector with @xmath6 entries . the operators @xmath45 and @xmath46 are defined by their action on a test operator @xmath47 @xmath48 where @xmath49 $ ] .
the auxiliary matrices are initialised to @xmath50 for @xmath51 and @xmath52 @xcite .
the hierarchy is truncated for sufficiently large @xmath53 , where the diagonal coupling in ( [ eq : dsigma ] ) becomes dominant . for the parameters used in the calculation , reorganisation energy @xmath29
@xmath26 and phonon relaxation time scale @xmath30 fs , we verified that the relative differences in the interesting energy region of the absorption spectrum using @xmath54 versus @xmath55 are below @xmath56 percent . in the 2d echo - spectrum we compared
the rephasing stimulated emission diagram for @xmath54 and @xmath55 and found a difference of less than @xmath57 percent of the amplitude .
we therefore consider a truncation at @xmath54 sufficient to reach converged results .
the derivation of the hierarchy stated in ( [ eq : drho ] ) and ( [ eq : dsigma ] ) relies on a high temperature approximation . for a temperature of @xmath3
k the high temperature approximation is not valid and low temperature correction terms have to be included @xcite . within the low temperature correction
we replace @xmath58 where @xmath59 is the first bosonic matsubara frequency following @xcite .
agreement with @xcite and proper thermalization to the boltzmann distribution @xcite confirms that the first order correction in ( [ eq : lowtemp ] ) is indeed sufficient .
a further improvement of the low temperature correction can be attained by including higher matsubara terms .
recently , also a pad spectrum decomposition has been proposed @xcite which shows faster convergence than the matsubara decomposition .
the numerical evaluation of the heom requires to propagate a large number of auxiliary matrices .
although the number of auxiliary matrices can be reduced by a scaled heom approach @xcite as well as by advanced filtering techniques @xcite , the heom are computationally very demanding , which sets limits to treatable system sizes . for 2d echo - spectroscopy one
has to propagate the set of auxiliary matrices not only for a huge number of time steps ( typically of the order of @xmath60 ) , but also the dimension of the auxiliary matrices gets enlarged since excited state absorption requires to extend the exciton hamiltonian to the two exciton manifold .
we overcome the computational limitation by using a gpu ( graphics processing unit ) implementation where the computation time goes down with the number of cores on a single gpu @xcite .
for example on a single nvidia c2050 graphics board we obtain a @xmath61 fold speedup compared to an equivalent single - core cpu implementation . with our efficient gpu algorithm the computation time for a 2d spectrum
is reduced to 50 minutes on a single gpu .
spectroscopy provides a tool that gives direct insight into the energy states of a quantum system .
absorption spectroscopy is used to identify the energy eigenstates and their interaction strengths with the radiation field , expressed in terms of dipole moments .
the energy eigenstates and the dipole moments are commonly identified with the peak position and the peak areas respectively .
the spectral peaks are broadened due to the interaction of the energy eigenstates with the remaining degrees of freedom of the system . for an ensemble of light - harvesting complexes the coupling of the electronic dynamics to the vibrational modes and the presence of fluctuations in the energy levels , due to e.g.
the motion of the surrounding proteins , broadens the absorption peaks . to calculate absorption - spectra we have to expand the system and add the ground state @xmath62 to our basis .
the hamiltonian is thus extended to an @xmath63 block matrix @xmath64 where we have chosen the zero exciton energy @xmath65 and @xmath28 is the @xmath66 matrix defined in ( [ eq : h ] ) .
using the same block structure as in ( [ eq:1d - h ] ) the dipole matrix that governs the photon - exciton conversion is constructed as @xmath67 where @xmath68 and the zero to single excitation dipole block @xmath69 is defined using the seven dipole moments @xmath70 that give the excitation of bchl@xmath8 by the laser pulse .
the absorption spectrum is calculated as @xmath71 where @xmath72 is the density operator for no excitation and we assume @xmath73-shaped laser pulses .
the different methods for performing the rotational average denoted by @xmath74 are detailed below .
the time evolution of the dipole operators in ( [ eq:1d - i ] ) is given by @xmath75 .
in the fmo complex the seven bchl molecules have a fixed orientation with respect to each other . within this configuration
each bchl has a dipole vector @xmath76 that characterises its interaction with light .
we consider the dipole vectors of the bchls to be along the connection line of two nitrogen atoms n@xmath77n@xmath78 @xcite whose positions we obtain from the protein data bank @xcite .
the norm of @xmath76 accounts for the strength of the dipole and we assume @xmath79 since all bchl molecules are identical . for the simulations we choose @xmath80 . given a polarisation direction of the laser @xmath81 , where @xmath82 and @xmath83 $ ] , the dipole moments in ( [ eq : mu_1ex ] ) read @xmath84 typical experiments
are performed using spectroscopy of a sample solution and many fmo complexes are illuminated at the same time .
each of them is randomly orientated with respect to the electric field in the laser pulse .
if we assume a coherent dynamics expression ( [ eq:1d - i ] ) is readily evaluated in eigenbasis as @xmath85 where @xmath86 is the @xmath87th eigenenergy and @xmath88 denotes the overlap of the @xmath44th state in site basis with the @xmath87th eigenstate .
the integrationally averaged dipole moments in eigenbasis are @xmath89 thus , we obtain the integrationally averaged squares of dipole moments in the eigenbasis @xmath90 , where the eigenenergies increase from left to right . for an appropriate @xmath91 ,
these values are in rough agreement with the averaged dipole moments obtained from a fit to the experimental spectrum @xmath92 debye@xmath93 @xcite .
however , we can also read the result as @xmath94 , where the three polarisation vectors @xmath95 are orthogonal to each other .
this now implies that the rotational average can be performed using three orthogonal directions for @xmath96 instead of sampling over the whole unit sphere .
note that this calculation is valid only if the interaction of system and bath does not interfere with the rotational average . in the following
we discuss how the line - shapes of the peaks depend on the theoretical method and approximations commonly used for performing the time - dependent propagation in ( [ eq:1d - i ] ) before we return to the different methods of performing the rotational average .
we calculate the absorption spectra of the fmo complex at the temperature @xmath1 k. we propagate the exciton - bath system during @xmath97 fs with a numerical time step of @xmath98 fs using the heom with reorganisation energy @xmath29 @xmath26 and a time scale for the bath correlations given by @xmath30 fs .
each heom propagation takes only 1 second on a gpu . in figure [ fig:1d - both ] ( a )
, we compare the results obtained for the absorption spectrum calculated using the heom and the secular born - markov approximation ( sbm ) following refs .
the line shapes for each eigenenergy can be calculated in the eigenbasis @xmath99 . in the sbm simulation
the complete spectrum is identical to the sum of the seven positive absorption peaks ( not shown ) . due to the rotating wave approximation
the terms @xmath100 for @xmath101 are zero . for the heom ,
these terms have non - vanishing contributions and the sum of the line shapes @xmath102 do no longer yield the spectrum .
thus we have to calculate the line shape of the @xmath87th exciton peak as @xmath103 .
including these terms leads to more complicated line shapes with shoulders and negative contributions as shown in figure [ fig:1d - both ] ( b ) for the peak of the second eigenstate .
we find that the more complicated line shapes arise from the non - secular approximations .
comparisons of different shapes of the spectrum obtained by different propagation methods and different spectral densities have been presented in @xcite for the b850 ring of the light - harvesting complex lh2 .
( [ eq:1d - i ] ) for the fmo complex calculated at @xmath1 k using the heom and the secular born - markov ( sbm ) approximation ( without lamb - shift ) for @xmath30 fs , @xmath29 @xmath26 and @xmath104 .
the vertical lines correspond to the eigenvalues of the exciton hamiltonian ( [ eq : h ] ) .
( b ) absorption lines of the second eigenstate calculated using the heom for different values of @xmath105 in the spectral density ( [ eq : spectral - density ] ) with fixed @xmath29 @xmath26 . ]
we observe in figure [ fig:1d - both ] ( a ) that the energies for the heom peaks are shifted compared to the peaks obtained in the sbm approximation . since we do not take into account the time - dependent lamb shift in the sbm approximation ( see eq .
( 8) in ref .
@xcite ) the corresponding sbm peaks are located at the exciton energies of hamiltonian ( [ eq : h ] ) , marked by the vertical lines .
note that the eigenenergies of hamiltonian ( [ eq : h ] ) include the reorganisation energy @xmath106 added to the `` bare exciton - energies '' .
due to the dissipation of the reorganisation energy within the vibrational environment , taken into account by the heom @xcite , the peaks of the absorption spectrum shift to lower energies . for a fast phonon relaxation @xmath107 ( markov limit )
the reorganisation energy dissipates instantaneously and the peaks are shifted to lower energies by approximately @xmath106 and then coincide with the bare exciton energies . figure [ fig:1d - both](b ) displays how the bath - correlation time affects the peak shapes and positions . for longer phonon relaxation times , the reorganisation energy dissipates slowly during the exciton dynamics and the peaks stay closer to the eigenenergies of ( [ eq : h ] ) .
while the spectra shown above are calculated using fitted dipole moments from @xcite , we will now analyse the changes due to rotational averaging on the absorption spectrum . in figure
[ fig:1d - averages ] ( a ) we show the results for the spectrum using the heom for @xmath29 @xmath26 and @xmath108 fs .
we compare different methods of rotational average discussed in the previous section , that is , a single propagation from integrationally averaged dipole moments using @xmath109 eq .
( [ eq : preavdip ] ) , a monte carlo ( mc ) set of 300 orientations and @xmath104 .
the results using three propagations with perpendicular laser directions @xmath110 are not shown since the curve is almost identical to the mc spectrum for 300 realisations .
the mc method matches the experimental conditions closest , since the experiment shows the averaged spectrum of many randomly orientated complexes . to avoid propagating density matrices for many random orientations ,
the @xmath110 approach can be chosen instead as both spectra show excellent agreement . neither the integrated dipole moments , nor the fitted ones yield comparable results to the mc method and only in the markovian limit it is possible to run the propagation with pre - averaged dipole moments . here , we chose a longer bath correlation time @xmath108 fs for the comparison of different rotational averages as the differences vanish for decreasing @xmath105 .
+ in the following we always use several orientations for rotational averages and do not resort to pre - averaged dipole moments . of the fmo complex calculated using the heom with @xmath111 at @xmath1 k , @xmath29 @xmath26 and @xmath108 fs .
we compare different methods of performing the rotational average in the calculations .
the dotted line results from using the fitted values @xmath104 , the dashed line takes the values @xmath109 .
the solid line denotes a monte carlo ( mc ) average over @xmath112 samples .
the spectra are normalised with respect to the area under the graph .
( b ) comparison of the absorption spectra for @xmath30 fs for different values of static gaussian disorder , which is characterised by its standard deviation ( sd ) . in all cases
the rotational average was performed with mc using 300 realisations . ]
if we compare the only rotationally averaged absorption spectra in figure [ fig:1d - both ] ( a ) with the experimentally measured ones @xcite , we find that the rotationally averaged peaks are too narrow . furthermore six peaks are well pronounced while in experiments several peaks overlap and the third exciton eigenstate is not individually resolved .
this result is expected , since there are several additional effects besides rotational averaging , which broaden the peaks and bring the simulation results closer to the experimental data . if we increase either @xmath106 ( not shown ) or the relaxation time @xmath105 , the peaks in the spectrum become broader , see figure [ fig:1d - both ] ( b ) .
both parameters change the spectral density and hence the coupling to the vibrational modes .
the third mechanism is static disorder caused by the slowly fluctuating protein environment .
we model the static disorder by adding a gaussian distributed noise of a given standard deviation ( sd ) to each diagonal term in ( [ eq : h ] ) .
the resulting spectra are shown in figure [ fig:1d - averages ] ( b ) where we obtain broader peaks and show with increasing disorder the diminishing of the third peak in the spectrum .
in addition we observe that the broadening changes the relative peak heights of the second and fourth peak yielding a better agreement with the experimental results . for the absorption spectrum changing the spectral density and inhomogeneous broadening yield similar results and
it is not possible to distinguish which mechanism is relevant for the fmo complex .
it is in the 2d echo - spectra where the difference becomes visible since the disorder results in a broadening along the diagonal frequencies only , while changing the exciton - phonon coupling broadens the peaks in all directions @xcite .
in addition to the information provided by absorption spectroscopy , two - dimensional echo - spectroscopy is used to study time - resolved processes such as energy transfer or vibrational decay as well as to measure intermolecular coupling strengths @xcite .
two - dimensional echo - spectroscopy consists of a sequence of three ultra - short pulses and a resulting signal pulse with fixed time - delay between the second and third pulse @xcite .
the theory of 2d echo - spectroscopy is explained in detail in @xcite .
the 2d echo - spectra are measured in two configurations which correspond to the rephasing ( rp ) and non - rephasing ( nr ) contributions of the third order response function @xmath113\right]\right ] \right).\ ] ] using the impulsive approximation which assumes @xmath73-functions for the temporal envelopes of the pulses , the two components of the spectrum corresponding to @xmath114 read @xmath115 sorting this expression for the rephasing and the non - rephasing parts yields six different pathways , which represent the stimulated emission , the ground - state bleach and the excited state absorption @xcite , shown schematically in figure [ fig : feynmnan ] . by taking the time between second and third pulse as a control parameter as done in ( [ eq : i_rp ] ) and ( [ eq : i_nr ] ) one can study the exciton dynamics which is probed by recording the emitted signal pulse . due to the excited state absorption the dynamics
is more involved compared to the one dimensional case . including a double excited manifold
, we construct an enlarged exciton hamiltonian @xmath116 which is block diagonal and of the form @xmath117 where @xmath65 represents the ground state , @xmath118 is the single - exciton hamiltonian with @xmath119 elements as defined in ( [ eq : h ] ) , and the energy levels of the two excitons are described by @xmath120 .
the two - exciton hamiltonian @xmath116 has @xmath121}^2 $ ] entries which are constructed from the matrix elements of @xmath118 following @xcite . for the fmo complex @xmath116 is a @xmath122 matrix . in site basis , we denote the two exciton states by @xmath123 , where @xmath124 , and the diagonal entries of @xmath120 are given by @xmath125 the off - diagonal elements are given by @xmath126 physically , this means that the @xmath127 and @xmath128 two - exciton states are coupled if they have one exciton in common while a second exciton is shifted from one site to another .
the coupling constant is the same as for the transition of the corresponding exciton shift in the one exciton dynamics .
note also that the linear coupling to the independent baths , each of them associated to one bchl , leads to couplings between the one exciton state @xmath129 and the two exciton states @xmath130 which do both influence the bath at site @xmath16 and are in return influenced by the bath correlations .
following @xcite , we construct the dipole operator associated with the extended hamiltonian ( [ eq:2d - h ] ) @xmath131 the zero to single excitation operators @xmath132 are defined as in ( [ eq:1d - mu ] ) and ( [ eq : mu_1ex ] ) .
the single to double excitation operators are defined according to @xmath133 and @xmath134 . in order to calculate 2d spectra
, we have to evolve the @xmath135 density matrix to obtain the third order response function @xmath136 .
the hierarchical method incorporates the effect of the baths through the auxiliary system of the @xmath137-matrices . when calculating the 2d spectra , we propagate according to the feynman diagrams shown in figure [ fig : feynmnan ] and act on the density matrix and the auxiliary matrices with dipole operator to simulate the interaction with the laser field .
we observe in ( [ eq : s3 ] ) that the 2d spectra depend on the fourth order dipole moments . to take into account the random orientation of the samples we need to compute rotational averages @xmath138 which are performed by sampling @xmath139 laser polarisation - vectors aligned to the vertices of a dodecahedron in one half of the coordinate space .
we take the vertices @xmath140 , @xmath141 , @xmath142 and @xmath143 , where @xmath144 denotes the golden ratio .
we have compared the dodecahedral summation with the full spherical averaging using a @xmath145-shoot mc and obtain differences of less than @xmath146 percent of the mc results . for the simulation of 2d spectra , we calculate the third order response function @xmath147 in ( [ eq : s3 ] ) .
we obtain @xmath148 data points with a spacing of @xmath149 fs in @xmath150 and @xmath151 direction using a numerical time step of @xmath98 fs .
this means that we have to perform over @xmath152 propagation steps for each of the six feynman diagrams .
depending on the delay time @xmath153 our simulations take 48 to 53 minutes on a single nvidia fermi c2050 gpu which provides a @xmath154 fold speed up compared to the cpu algorithm with filtering methods that reduce the hierarchy @xcite .
we always show the sum of the real parts of the rephasing and the non - rephasing spectrum . $ ] of the fmo complex calculated using the heom with @xmath29 @xmath26 , @xmath1 k and @xmath30 fs and a delay time @xmath155 fs .
( b ) oscillations as a function of the delay time for the @xmath156 and @xmath157 diagonal peak and the @xmath158 peak below the diagonal .
the points have been connected to guide the eye . ]
figure [ fig : heom2d - dodeaverage ] ( a ) shows a simulated 2d spectrum for a rotationally averaged sample with no disorder at @xmath1 k. the figure is encoded using a linear colour scale .
we divided the raw data by a constant to obtain an amplitude on a @xmath159@xmath160 range .
the thin white lines have been added to indicate the expected peak positions .
the peaks are denoted by their excitonic energy - level indices @xmath161 , with the first number referring to the @xmath162-axis and the second one to the @xmath163-axis .
the spectrum shows similarities to the experimentally measured one in @xcite insofar as it reproduces the large peak on the diagonal at the second eigenstate and a second near the fourth and fifth .
furthermore we find negative amplitudes above the diagonal and positive peaks below . as expected from the discussion of the absorption spectrum in section [ sec
: broad ] , the peak shapes differ from the experimental ones due to the lack of static disorder .
figure [ fig : heom2d - dodeaverage ] ( b ) shows the peak oscillations for the diagonal peaks @xmath156 and @xmath157 and the off - diagonal peak @xmath158 below the diagonal.we renormalised the peak heights at @xmath155 fs to unity such that the time resolved amplitude of the three peaks are conveniently put into the same panel .
we choose these peaks because figure [ fig : heom2d - dodeaverage ] shows peak @xmath158 as the highest off - diagonal peak in the line where @xmath164 equals the lowest lying eigenstate .
we see that diagonal and off - diagonal peaks oscillate in time .
this indicates that our model imitates the coherent energy transport within the fmo complex that has been observed experimentally @xcite .
theoretically , an off - diagonal peak @xmath161 is expected to undergo oscillations with a frequency corresponding to the difference of its excitonic eigenenergies @xmath165 . for the @xmath158 peak we expect a period of @xmath166 fs which is in good agreement to our numerical result yielding a best - fit period of @xmath167 fs .
when we calculate the peak heights , we integrate over squares of width @xmath168 @xmath26 centred around the white crosses .
we follow the changes of the peak heights with increasing delay time by plotting the absolute value of the spectrum normalised with respect to the value at @xmath155 fs . of the rotationally
averaged spectra of the fmo complex as a function of delay time for different temperatures .
temperature dependence of the peak oscillations for the @xmath158 cross peak .
( b ) oscillations in the peak width of the @xmath156 diagonal peak measured in diagonal and anti - diagonal direction compared to the peak height at @xmath1 k. the points have been connected to guide the eye . ] in order to analyse the robustness of the peak oscillations when the thermal fluctuations increase we plot in figure [ fig : heom2d - nodisorder - peak - oscillations ] ( a ) the oscillations of the @xmath158 cross peak for several temperatures .
we compare with the experimental data in @xcite and observe a qualitative similar behaviour with weaker oscillations and faster decay when temperature is increased .
we do not obtain the picoseconds lasting beatings that have been observed experimentally .
one possible factor is the shape of the spectral density that is used for the heom .
we fit the oscillations in figure [ fig : heom2d - nodisorder - peak - oscillations ] ( a ) by an exponentially decaying sine .
the fit is done separately for the stimulated emission and excited state absorption pathway , which couple to one or two baths respectively .
based on the damping of the oscillations on peak @xmath158 the temperature dependence of the dephasing rate of our simulation is given by @xmath169 @xmath170/k and @xmath171 @xmath170/k .
this is in good agreement to what we expect for the chosen spectral density ( @xmath172 @xmath26/k ) and the value @xmath173
@xmath26/k determined from the experimentally measured data @xcite .
in the experiments an anti - correlation in peak shapes between the diagonal peak width and the height @xcite , or alternatively the ratio of the diagonal peak - width over the anti - diagonal height @xcite has been found .
the width was measured by fitting a gaussian to a diagonal cut through the spectrum and taking its standard deviation .
applying the same procedure to the simulation data yields also anti - correlations shown in figure [ fig : heom2d - nodisorder - peak - oscillations ] ( b ) .
$ ] of the fmo complex with rotational and disorder averages at temperature @xmath1 k. the averages are performed over @xmath174 realisations .
the spectral - density parameters are @xmath29 @xmath26 and @xmath30 fs .
( a ) gaussian disorder with standard deviation ( sd ) @xmath175
@xmath26 at delay time @xmath155 fs .
( b ) sd@xmath176 @xmath26 , @xmath155 fs . ( c ) sd@xmath177
@xmath26 , @xmath178 fs . ] next , we analyse the role of static disorder in the results for the 2d echo - spectra . when including static disorder , the peaks get elongated along the diagonal as observed when comparing the spectra at zero delay time shown in figures [ fig : heom2d - dodeaverage ] ( a ) and [ fig : heom2d - with - average ] ( a ) and ( b ) for increasing values of disorder .
the plots with disorder show better agreement with the experiment @xcite .
figure [ fig : heom2d - with - average ] ( c ) shows the 2d spectra for sd@xmath177 @xmath26 for a long delay time @xmath178 fs for which the system approaches its thermal stationary state . we observe that the excitation moves from the diagonal peaks into the regions with lower frequency @xmath164 .
this occurs as the system relaxes to lower lying eigenstates during the delay time . in experiments by brixner
@xcite the decay of the population into the ground state of the system is also visible .
the absence of the experimentally seen oscillation in the heom calculation in ref .
@xcite has led chen to conclude that the importance of static disorder requires careful reconsideration .
first - principles simulations of static disorder in the fmo complex are computationally extremely time - consuming , since several hundred of realisations are required to reach converged results . in order to study the role of static disorder in the oscillations of the peaks of the spectrum and to ensure numerical convergence , we construct a dimer to model the @xmath158 peak of the fmo spectrum .
for the model dimer we chose the hamiltonian such that the first and fifth fmo complex eigenvalues are reproduced .
the coupling between both sites is set to @xmath179 @xmath26 .
k , @xmath29 @xmath26 and @xmath30 fs . to concentrate on the disorder effects ,
no rotational average is performed and the calculation is done using @xmath180 and @xmath181 ( a ) no static disorder , ( b ) gaussian disorder with standard deviation ( sd ) @xmath175
the average was done over @xmath182 realisations at each delay time .
the points have been connected to guide the eye . ]
the simulations with and without disorder show oscillations on the diagonal and off - diagonal peaks .
the most striking effect of disorder is a decreased oscillatory amplitude .
while the cross peak in the simulation without disorder has lost @xmath183 percent of its amplitude in the first minimum , in the disorder case the loss is only @xmath184 percent . from a fit , we observe that the periods of the disorder curves are slightly shorter @xmath185 fs and @xmath186 fs for the @xmath156 and @xmath187 curves while in the no - disorder plot @xmath188 fs and @xmath186 fs has been obtained , respectively . for the fmo complex
we expect that oscillations are reduced in amplitude by a similar magnitude and therefore still persist within the model .
we have presented calculations of the absorption spectra and the 2d echo - spectra for the fmo complex at different temperatures using the heom for solving the exciton and the vibrational bath dynamics and adding static disorder to account for the fluctuating environment .
important ingredients necessary to find good agreement with experiments for the absorption spectra using the heom propagation method is to perform the rotational average and to include static disorder .
we have also discussed how other simplified propagation methods affect the line shapes and shift the peaks in the spectra .
our rotationally averaged 2d echo - spectra reflect the main features of the experimentally measured ones . at delay time
@xmath155 we find negative contributions above the diagonal and elongated peaks in the diagonal which are transferred to the lowest exciton states after long times on the picosecond scale .
the heom method yields an oscillatory behaviour of peak amplitudes over a wide temperature range . at @xmath1
k we obtain oscillations of cross peak amplitudes up to @xmath112 fs with the expected frequency that corresponds to the difference in excitonic eigenenergies . in addition oscillations on the diagonal peaks
are present .
the long duration of oscillations in the experimental data ( up to @xmath0 ps in ref .
@xcite ) suggests that further refinements of the theoretical models are necessary , for example by adjusting the form of the spectral density which affects the population dynamics @xcite .
remarkably , we find excellent agreement with the measured damping rate of the oscillations due to thermal fluctuations .
this implies a good choice of the parameters of the model and might explain why we obtain different results compared to previous calculations , which were done for a higher value of @xmath189 fs @xcite , leading to an increased dephasing by a factor of two . at ambient temperatures the heom have the advantage to require less terms to converge while approximate methods such as the secular and full redfield approximations do not yield the correct time - evolution of the density matrix and
overestimate the thermalisation rate . in agreement with the experimental measurements we observe the anti - correlated oscillations of the peak heights and peak shapes for the diagonal peaks .
the analysis of a model dimer shows that the cross peak oscillations persist even in the presence of static disorder which mainly results in a reduction of oscillation amplitude by about @xmath190 percent .
this work is supported by the emmy - noether programme of the dfg ( kr 2889/2 ) , the daad project 50240755 , the spanish mincinn ai de2009 - 0088 ( acciones integradas
hispano - alemanas ) , the spanish micinn project fis2010 - 18799 , and the ramn y cajal programme .
we thank akihito ishizaki for helpful discussions .
time on the harvard school of engineering and applied sciences `` resonance '' gpu cluster and support by the seas academic computing team are gratefully acknowledged .
10 m. wendling , t. pullerits , m. a. przyjalgowski , s. i. e. vulto , t. j. aartsma , r. van grondelle , and h. van amerongen .
electron - vibrational coupling in the fenna - matthews - olson complex of prosthecochloris aestuarii determined by temperature - dependent absorption and fluorescence line - narrowing measurements .
, 104:5825 , 2000 . j hu , m luo , f jiang , r - x xu , and yj yan .
pad spectrum decompositions of quantum distribution functions and optimal hierarchical equations of motion construction for quantum open systems .
, 134:244106 , 2011 . | recent experimental observations of time - dependent beatings in the two - dimensional echo - spectra of light - harvesting complexes at ambient temperatures have opened up the question whether coherence and wave - like behaviour plays a significant role in photosynthesis .
we perform a numerical study of the absorption and echo - spectra of the fenna - matthews - olson ( fmo ) complex in _ chlorobium tepidum _ and analyse the requirements in the theoretical model needed to reproduce beatings in the calculated spectra .
the energy transfer in the fmo pigment - protein complex is theoretically described by an exciton hamiltonian coupled to a phonon bath which account for the pigments electronic and vibrational excitations respectively .
we use the hierarchical equations of motions method to treat the strong couplings in a non - perturbative way .
we show that the oscillations in the two - dimensional echo - spectra persist in the presence of thermal noise and static disorder . |
it is now widely accepted that the most of the massive galaxies have their central super - massive black holes ( smbhs ) ( e.g. , kormendy & richstone 1995 ) .
the tight correlation between the mass of smbh , m@xmath6 , and the spheroid mass , m@xmath12 , or the velocity dispersion of the host is observed for both normal and active galaxies in the nearby universe ( magorrian _ et al .
_ 1998 ; ferrarese & merrit 2000 ; mclure & dunlop 2001 ) .
the correlation suggests that the formation of galaxies and their smbh are closely related and probably have the causal or even coeval connection
. the correlation may result from ( i ) mass accretion to the bh due to the star - formation in galaxy spheroids through some physical process such as the stochastic kinematic viscosity by supernovae or by the radiation drag mechanism ( yamada 1994 ; wada & norman 2002 ; kawakatsu & umemura 2002 ) , or ( ii ) termination or suppression of star - formation by the feedback of strong nuclear activity ( hopkins _ et al .
the true origin of the relationship is , however , still far from unveiled especially in observations . in order to see how the formation of galaxies and their smbh are related , and to understand the origins of the m@xmath6-m@xmath12 correlation , it is needed to study the relationship among stellar mass , star - formation , and active galactic nuclei ( agn ) activities of the galaxies at high - redshift when they are majorly developing their stars , structures , and central black holes .
it is also very interesting to see whether such relationship has always been hold or not during the hierarchical growth of massive galaxies ( haehnelt & rees 1993 ; haehnelt & kauffmann 2000 ; granato _ et al .
_ 2004 ; hopkins _ et al .
_ 2006 ) . at @xmath13=0 ,
agn are observed exclusively in the massive galaxies with the stellar mass larger than @xmath14 m@xmath5 ( kauffmann _ et al .
is this also true at high redshift , or do the progenitors of the present - day massive galaxies already have the smbh as massive as they have now ?
indeed , some authors claimed that the m@xmath6/m@xmath12 ratio increases with increasing redshift ( mclure _ et al .
_ 2006 ) , in which case smbh grow earlier than the formation of the entire stars in massive galaxies . it is , however , difficult to directly observe the m@xmath6-m@xmath12 correlation at high redshift .
the techniques to evaluate the bh mass of normal galaxies can not be used for the distant objects . for active galaxies at high redshift ,
massive bhs are likely to have been formed in luminous quasars ( turner 1991 ) and their virial mass can be measured by using the velocity width and the size of the broad - line regions estimated from the quasar luminosity ( kaspi _ et al .
_ 2000 ; mclure & dunlop 2001 ) .
they are in the range of 10@xmath15 - 10@xmath16 m@xmath5 for the luminous quasars at @xmath17 ( shen _ et al .
_ 2008 ) . on the other hand , however , it is difficult to directly observe the properties of the host galaxies , such as morphology , stellar mass , and velocity dispersions , due to the pollution of the light from the very bright nuclei .
adaptive optics imaging of distant quasars succeeded to detect host galaxies only for a few objects ( e.g. , croom _ et al .
_ 2004 ) . akiyama ( 2005 ) investigated the host galaxies of x - ray - selected agn which are less luminous than quasars on the rest - frame uv images and detected extended components for the half of the samples ( 17/31 ) , but their stellar mass is only weakly constrained due to the lack of the deep near - infrared data .
( 2006 ) analyzed the sample of 3c rr radio - loud objects and argue that the m@xmath6-m@xmath12 ratio is larger at high redshift , @xmath18 of the local at @xmath19 .
the line - width of the emission lines in radio - loud objects , however , may be affected by the gas kinematics due to the radio jet activity , which may cause possible uncertainty in bh mass estimation . molecular gas , which is related to star - formation but not directly to the bh activity , is a reliable probe to see the host galaxies of luminous quasars ( yamada 1994 , miolino _ et al .
_ 2007 and reference therein ) .
interestingly , more than a few luminous high - redshift quasars are detected in co lines and the evidence of massive ( @xmath20m@xmath5 ) host galaxies of the quasars at @xmath13=2 - 6 were reported ( e.g. , ohta _ et al .
_ 1996 ; miolino _ et al .
evidence of the growing smbh is also observed for high - redshift star - forming galaxies .
( 2005 ; 2006 ) have revealed that @xmath21% of the sub - mm - selected galaxies ( smgs ) show agn activities and the majority of the active nuclei seem to be heavily obscured by dust .
( 2007 ) also studied the mid - infrared excess galaxies at @xmath22 and found the evidence of obscured compton - thick agn among them by the stacked x - ray data for the sample .
these observations suggest that the formation of smbh is indeed related with the formation of massive galaxies in their infrared ultra luminous ( ulirg ) phase .
time - lag between starburst and agn activity may results in showing the high or low m@xmath6-m@xmath12 ratio in quasar and ulirg phase ( alexander _ et al .
_ 2008 ) . in this paper , we study the general x - ray properties of the massive and the less massive galaxies at @xmath23 - 4 using the new extremely deep near - infrared ( nir ) imaging data obtained in the moircs deep survey ( mods , see section 2 ) to investigate the agn activity among the stellar - mass based sample .
previously , only the x - ray fraction among the nir - selected sources at @xmath19 have been reported ( reddy _ et al .
_ 2005 ; papovich _ et al .
_ 2006 ) and the detailed properties of the x - ray sources associated with the galaxies at such higher redshift has not been studied well . at @xmath8 ,
the x - ray sources detected even in the deepest observation seems dominated by agn population with the x - ray luminosity @xmath24 ( 2 - 10 kev)= 10@xmath25 - 10@xmath26 erg s@xmath3 , which corresponds to the luminosity of the luminous seyfert galaxies in the local universe ( green _ et al .
_ 1992 ; akiyama 2005 ) .
while they may not be at the peak of the agn / quasar activity in each individual galaxy , the lower - limit values of the existing bh mass can still be evaluated if we assume the eddington limit in their luminosity . on the other hand ,
given the hypothetical form of the redshift evolution of the m@xmath6/m@xmath12 ratio , we may discuss the agn activity along the stellar mass for the galaxies at such high redshift . in section 2
, we introduce the nir and the x - ray data and the sample we use , and describe the method to obtain the various properties of the galaxies , such as stellar mass , reddening , star - formation rate .
we show the x - ray properties of these samples in section 3 and discuss the agn activity among the high - redshift galaxies in section 4 .
the cosmological parameters used here are canonical values after wmap , namely , @xmath27=0.3 , @xmath28 , and @xmath29=70 kms@xmath3mpc@xmath3 .
nir magnitude values are in vega system ( oke & gunn 1983 ; fukugita et al 1996 ) .
in order to study the relationship between agn activities and galaxy properties at high redshift , we constructed a sample of @xmath0-band selected galaxies between @xmath13=2 and 4 from the moircs deep survey ( mods , kajisawa _ et al .
_ 2006 ; ichikawa _ et al .
_ 2007 ) . by using multi - object infrared camera and spectrograph ( moircs , ichikawa _
_ 2006 ; suzuki _ et al .
_ 2008 ) equipped with subaru 8.2 m telescope , we obtained deep @xmath30 , @xmath31 , and @xmath32 band images at the sky area in the fields of great observatories origins deep survey field north ( goods - n ) .
we cover the area by the four fields of views ( fov ) of moircs , referred as gt1 , gt2 , gt3 , and gt4 , from north - east to south west .
full description of mods result will be published in the separated papers ( kajisawa _ et al . _ in preparation ) .
gt2 , which includes the hubble deep field north (
hdf - n ) is also the ultra - deep field of mods survey where more than 28h science integrations were obtained in both @xmath30 and @xmath32 bands .
the total sky area ( gt1-gt4 ) used in the analysis of this paper is 103.3 arcmin@xmath33 .
the basic properties of the data obtained and reduced by june , 2008 are summarized in table 1 .
ccccccl field & filter & fwhm ( ch1 ) & depth ( ch1 ) & fwhm ( ch2 ) & depth ( ch2 ) & exp .
time + & & arcsec & mag . &
arcsec & mag . & hours + gt1 & j & 0.59 & 24.9 & 0.59 & 24.8 & 8.0 + & h & 0.58 & 23.8 & 0.59 & 23.7 & 2.5 + & k & 0.58 & 23.7 & 0.53 & 23.8 & 8.3 + gt2 & j & 0.48 & 25.7 & 0.49 & 25.6 & 28.2 + & h & 0.46 & 24.4 & 0.46 & 24.2 & 5.7 + & k & 0.45 & 24.7 & 0.46 & 24.6 & 28.0 + gt3 & j & 0.57 & 24.9 & 0.58 & 24.8 & 6.3 + & h & 0.55 & 23.7 & 0.55 & 23.7 & 3.2 + & k & 0.59 & 23.8 & 0.60 & 23.7 & 10.7 + gt4 & j & 0.58 & 24.8 & 0.59 & 24.8 & 9.1 + & h & 0.58 & 23.8 & 0.59 & 23.8 & 4.3 + & k & 0.59 & 23.6 & 0.60 & 23.7 & 9.8 + in mods , more than 7000 @xmath0-band - selected galaxies are detected down to @xmath34 ( 3@xmath35 ) at gt1-gt4 and 24.6 in the deepest gt2 field .
the moircs images are carefully aligned to the optical images obtained by advanced camera for survey ( acs ) on hubble space telescope ( hst ) , as well as the infrared images taken with irac and mips instruments on spitzer space telescope .
acs and moircs @xmath30 and @xmath31 images are smoothed to be matched with the @xmath32-band images and the colors of the objects are obtained by the apertures with the diameter of 1.2 arcsec . for the irac data , we first measured the flux in apertures in 3.16 , 2.78 , 3.43 , and 3.47-arcsec diameter in 3.6 , 4.5 , 5.8 , and 8@xmath36 m , and then made aperture correction by using the light profile of the @xmath32-band image smoothed to be matched with each irac band .
the multi - band photometric data are then fitted by the galaxy evolutionary models of bruzual & charlot ( 2003 ) so that we obtained the photometric redshift , stellar mass , reddening , and parameters of the star - formation histories such as ages , star - formation - rate , and the decaying parameter @xmath37 by the similar method described in kajisawa & yamada ( 2005 ) .
kajisawa & yamada ( 2005 ) carefully investigated the uncertainty in the evaluation of the stellar mass and found that the stellar mass of the most of the @xmath0-selected galaxies can be determined well for the given sed while there may be degeneracy in other parameters such as age , extinction , and star - formation - decaying time scale .
( in preparation ) also studied the uncertainty of the stellar mass estimation of the current sample to find that the @xmath38 uncertainty is 0.1 dex for m@xmath7=@xmath39m@xmath5 and 0.2 - 0.3 dex for m@xmath7=@xmath40m@xmath5 . in this paper , we used the models with salpeter initial mass function ( imf ) .
reduction by the factor of two in the stellar mass values is needed to compare with those obtained with the chabrier - like imf ( chabrier _ et al .
_ 2003 ) .
the x - ray source catalog in the goods - n field is taken from alexander _ et al . _ ( 2003 ) who conducted 2m - sec observations with the x - ray satellite _ chandra _ at the field .
the on - axis sensitivity limit is @xmath41 erg s@xmath3 at 0.5 - 2 kev ( soft ) band , @xmath42 erg s@xmath3 at the 2 - 8 kev ( hard ) band , and @xmath43 erg s@xmath3 at the 0.5 - 8 kev ( total ) band , respectively .
their exposure maps show that the data is fairly homogeneous over the mods gt1-gt4 fields .
the median positional determination accuracy is @xmath44 arcsec ( alexander _ et al .
_ 2003 ) .
we matched the chandra x - ray source positions in their catalog ( table 3 in alexander _
2003 ) with the @xmath32-band sources by the following procedure .
we first identify the pixel on the @xmath32-band image at the x - ray source position in the catalog and then find the @xmath32-band selected object to which the pixel belongs .
if there is no @xmath32-band object at the pixel , we then extend the search pixels up to three ( 0.35 arcsec ) to find the @xmath32-band counterpart .
in fact , the most of the chandra sources can be identified with the @xmath32-band sources in the first step thanks to the good positional accuracy of the chandra data and the depth of the infrared data . while the depth of nir images at the field gt2 is significantly deeper than other fields , most of the x - ray sources but a few are identified with the @xmath32-band sources above @xmath45 . for the current purpose ,
@xmath0-selected sample within the redshift range , @xmath1 are constructed .
@xmath32-band light samples @xmath46 at rest frame at this redshift range .
we primary use the spectroscopic redshift if available ( wirth _ et al .
_ 2004 ; reddy _ et al .
_ 2006 ; barger _
_ 2008 ) and photometric ones if not . in total , over gt1-gt4
, 2621 galaxies are selected as the sample of which 217 objects have spectroscopic redshift . among them , 1374 galaxies are brighter than @xmath0=23 , which is the conservative completeness limit over gt1-gt4 .
79 objects have the stellar mass larger than 10@xmath47 m@xmath5 and 219 objects have 10@xmath48 m@xmath5@xmath49 m@xmath7 @xmath49 10@xmath47 m@xmath5 .
the accuracy of the photometric redshift , evaluated by using the spectroscopic sample , is @xmath50@xmath13/@[email protected] with 9@xmath53 outlier that has @xmath54@xmath50@xmath13/@xmath51@xmath54@xmath55 and @xmath50@xmath13/@[email protected] with 5@xmath53 ( 1 source ) outlier for the @xmath0-selected galaxies and those associated with the x - ray sources , respectively .
the redshift distribution at @xmath1 is shown in fig.1 .
fig.2 shows the stellar mass versus @xmath56 color of the 2621 objects .
galaxies with @xmath57 , above the completeness limit over gt1-gt4 , are shown by the large dots and the fainter ones are by the small dots . as previously pointed out ( e.g. , kajisawa & yamada 2005 ; 2006 ) ,
the most massive galaxies tend to have the redder colors . at m@xmath7 @xmath58 m@xmath5 , more than 80% of the galaxies have @xmath59 .
the galaxies identified with the chandra sources are also shown in fig.2 and the x - ray identified fractions of the samples are listed in table 2 . in total , 61 total - band ( 0.5 - 8 kev ) x - ray sources , or 52 hard - band ( 2 - 8 kev ) x - ray sources are identified with the @xmath0-band sources .
we found that 21 of 61 , or 19 of 52 objects have the spectroscopic redshift values between @xmath13=2 and 4 . since the majority of the x - ray sources are detected in the hard energy band , 2 - 8 kev , which corresponds to 6 - 24 kev and 10 - 40 kev at @xmath13=2 and 4 , respectively , these x - ray sources are likely to be agns .
indeed , their luminosity range , @xmath60 erg s@xmath3 , is consistent with those of the luminous seyfert galaxies or faint quasars in nearby universe .
while active star - formation also produces luminous x - ray emission , the expected luminosity is more than an order of magnitude lower than the measured values .
the empirical fit by ranalli _
( 2003 ) predicts @xmath24(2 - 10 kev ) = @xmath61 erg s@xmath3 for the star - formation rate ( sfr ) of 1 m@xmath5 yr@xmath3 .
as shown below , the largest sfr estimated for the current sample is 1000 m@xmath62 yr@xmath3 that may produce the x - ray luminosity of @xmath63 erg s@xmath3 , which is about the lower bound of our sample ( see fig.8 below ) .
we conclude that the observed x - ray emission is associated with the agn activity without serious contamination by star - formation .
we first note that the most of the identified x - ray sources are associated with massive galaxies . among the 61 total - band sources , 90% ( 55/61 )
are associated with the galaxies with m@xmath7 @xmath64 m@xmath5 and 48% ( 29/61 ) are those with m@xmath7 @xmath65 @xmath9 m@xmath5 .
table 2 also summarizes these numbers . on the other hand , for the massive galaxies with m@xmath66 m@xmath5 ,
the fraction of the x - ray detection is notably high , 33% ( 26/78 ) and 29% ( 23/78 ) for the total - band and the hard - band sources , respectively .
these values can be interpreted as the lower limit as we ignored the attenuation to the x - ray emission ; we may missed very compton - thick objects with absorption larger than n@xmath67 @xmath68 @xmath69 @xmath70 .
we checked by the several methods that the nir light of the objects is indeed stellar origins and not heavily contaminated by the agn component and the evaluation of the stellar mass based on nir luminosity is not significantly affected by the agn light .
if we adopt the typical x - ray to optical flux - density ( per unit frequency ) ratio for broad - line agn , @xmath71(20 kev)/@xmath71(0.5@xmath36 m ) @xmath72-@xmath73 ( richards _ et al .
_ 2006 ) , the contribution of the agn light in @xmath32-band , namely rest - frame optical band , is estimated to be 19.5 - 22 mag for the sources with @xmath74(2 - 8 kev)=@xmath75 erg s@xmath3 @xmath70 without extinction at the optical wavelength . while the most of the x - ray sources in our sample are fainter than this flux range and their host - galaxy @xmath0-band magnitudes are in the range of @xmath0=19 - 22.5 mag , the contribution of the agn light in @xmath0 band may not be negligible for more than a few cases .
we therefore carefully examined the colors and the optical - mid - infrared spectral energy distributions ( sed ) of the @xmath0-band sources which have the x - ray counterpart .
( 2007 ) examined the properties of the spitzer irac sources in goods - n which have power - law seds and found about a half of them are detected in x - ray to show the agn activity .
we checked how many @xmath0-selected sources may be classified with the power - law sources and found that 13 @xmath0-selected sources are matched with such sed in donley _
( 2007 ) and of them , 9 are detected by _
other four non - xray objects may host hidden very compton - thick agn , which make the agn fraction of the sample even larger .
thus about 15% ( 9/61 ) of the x - ray - identified @xmath0-selected galaxies the nir light may be contaminated by the hot - dust component . in our own photometry ,
7 of the 9 power - law sources ( detected in x - ray ) indeed have the seds monotonically increased in ab magnitude toward the longer wavelength .
however , the rest ( 85% , 52/61 ) of the sample can be well fitted by the stellar - population seds showing the turn off near the rest - frame 1.6@xmath36 m bump , which supports that our stellar - mass evaluation for these objects is robust .
furthermore , even for the 9 power - law sources , we found that our stellar - mass estimation may be reasonable .
we made the similar sed fitting process with regarding the @xmath0 and the irac magnitudes as the upper limits , allowing the contamination by the agn light , but the evaluated stellar mass values do not change very much with a few exceptions .
fig.3 compare the stellar mass obtained by the sed fitting to the all measured photometric values and those with the upper - limit substitution in @xmath0 and the irac bands .
we also show the three examples of the sed fitting results in fig.4a-4c .
fig.4a shows the very typical case which has the middle - range @xmath76 value in the fitting among the sample .
the magenta line is the best - fitted models using the all photometric data points and the blue line is that using the @xmath0 and irac - band data ( open circles ) as the upper limit values . in this case ,
clear turn off around the 5.8@xmath36m - band ( irac ch3 ) data , which corresponds to the 1.6@xmath36m - bump feature in the stellar light , is appeared .
fig.4b shows the case of a relatively red objects , which has the @xmath76 value at @xmath77 ( 10th among the 61 objects ) from the worst .
while the 8@xmath36 m data is deviated from the stellar model , which shows the evidence of hot dust component , the data at the wavelength shorter than 5.8@xmath36 m are well fitted by the stellar model including the balmer jump feature near the @xmath31 band . in this case , the difference in stellar mass from the two different fitting procedures is @xmath78 , which is one of the largest deviation among the sample ( see fig.3 ) .
fig.4c shows the case of a power - law object , which indeed shows the monotonic rising of the flux toward the longer wavelength . even in this case
, we see clear jump of the sed below the @xmath30 band , which is very likely to the balmer jump in the stellar spectrum .
the obtained stellar mass is almost same for the two different fitting procedures while the other parameters of the models ( age , extinction ) are different .
we also found that the @xmath32-band light distributions show the extended profile and not dominated by the point source of the agn components .
5 shows the fwhm of the @xmath32-band images of the 61 x - ray sources . majority ( @xmath79 ) show extended profiles . at the bright end , or in the sharpest image , gt2 ,
the fraction is higher .
the power - law sources are marked with the squares and we see that their @xmath32-band light distributions are also extended in many cases . from these above , we conclude that the observed nir light are not seriously contaminated by the agn light and the obtained stellar mass values are as robust as for the non - x - ray galaxies .
cccc & all & m@xmath7 @xmath65 @xmath9 m@xmath5 & @xmath4 m@xmath5 @xmath49 m@xmath7 @xmath49 @xmath9 m@xmath5 + @xmath0-band sources with @xmath1 & 2621 & 79 & 219 + hard - band x - ray sources & 52 & 23 & 24 + total - band x - ray sources & 61 & 26 & 29 + fig.6 shows the x - ray luminosities of the sources as the function of the stellar mass of the host galaxies .
we obtained the x - ray luminosity at 2 - 10 kev from the hard - band flux adopting the relation @xmath80 to obtain the k - correction factor .
while the soft band ( 0.5 - 2 kev ) is closer to the rest - frame 2 - 10 kev energy range for the objects at @xmath8 , the hard - band flux is less affected by the extinction and we may obtain the x - ray luminosities which are closer to the intrinsic values .
for those which are not detected significantly in the hard - band , we used the total - band energy flux .
the galaxies between @xmath23 and 3 are shown by the blue dots and the galaxies at @xmath81 - 4 by the red dots .
there is no clear trend in fig.6 and thus the stellar mass and the x - ray luminosity does not seem to strongly correlate with each other . in fig.7 , we plotted the specific agn activity , namely the x - ray luminosity divided by the host stellar mass , against their host stellar mass .
note that this is just a different presentation of the data shown in fig.6 . as expected from fig.6 , notable trend that more stellar massive galaxies have smaller specific agn activity is observed .
if the m@xmath6/m@xmath7 ratio is the same among the galaxies and the bhs are similarly active ( in mass accretion ) , the specific agn luminosity is to be constant . for the guide , we plotted the case of the local m@xmath6/m@xmath12 ratio and the 10@xmath53 eddington accretion rate by the dotted line and 100@xmath53 and 1@xmath53 cases by the upper and lower dashed lines , respectively .
fig.7 shows that the specific agn activity is smaller than the 10@xmath53-eddington value for the massive galaxies with m@xmath7 @xmath20 m@xmath5 and larger for the less massive galaxies near m@xmath7 @xmath82 m@xmath5 .
this indicates that bhs in massive galaxies are relatively inactive , or , that more massive bhs are formed in relatively less massive galaxies and they have the similar activities .
we discuss these results further in sec .
it is also possible that attenuation of the x - ray emission in massive galaxies causes such a trend . in order to see whether this is the case
, we checked the _
mips 24@xmath36 m flux data of the sample since the objects with luminous infrared emission may host such obscured agn .
the objects with large circles in fig.7 are those detected in mips 24@xmath36 m band above @xmath83@xmath36jy .
the mips detection fraction among the massive objects , m@xmath7 @xmath84 m@xmath5 , is 39@xmath53 ( 7/18 ) for those with the relatively large specific agn activity ( objects above the dotted line in fig.7 ) and 52% ( 13/25 ) for the rest .
thus the 24@xmath36 m detection rate is similar for the two samples , which does not support that the low specific agn activity is due to the x - ray attenuation in dust - rich environment .
fig.8a and 8b show the distribution of the x - ray luminosity and the star - formation rate ( sfr ) of the galaxies obtained by the sed fitting .
we use the instantaneous sfr in the left panel ( fig.8a ) and the sfr averaged over the past 300 myrs in the right ( fig.8b ) .
the sfr ranges from a few to 1000 m@xmath62 yr@xmath3 and the specific sfr , namely the sfr divided by their stellar mass , ranges from 10@xmath85 to 10@xmath86 yr@xmath3 .
we do not see any clear correlation between @xmath87 ( 2 - 10 kev ) and sfr over 1 - 1000 m@xmath5yr@xmath3 .
if the star - formation in each galaxy and the mass accretion to its bhs are closely correlated and exactly coeval all the time , there should be more tight correlation between @xmath87 and sfr , while a large uncertain here is how much fraction of the sfr is indeed directly related to the formation of the spheroid component .
as also inferred from fig.8 , the specific agn activity and the specific sfr do not show tight correlation .
the dotted lines in the figures show the expected x - ray luminosity due to the star - formation activity predicted from the empirical fit to the local galaxies by ranalli _
the recent models studied by mas - hesse _ et al . _
( 2008 ) give the consistent results .
we see the observed x - ray luminosities are an order - of - magnitude more luminous than the prediction , which implies that they are due to the agn activities and not seriously contaminated even by the very intense star formation . in fig.9
, we also show the distribution of the @xmath0-selected as well as the x - ray - identified objects along the color excess values , e(b - v ) and the average age .
we adopted the calzetti law for the dust extinction in the sed fitting .
the x - ray - identified sources are distributed over the parameter plane and no very strong trend is recognized . ages and e(b - v ) values , however , may be degenerate in the sed fitting , which possibly dilutes any intrinsic trends .
there is a hint that the x - ray fraction is slightly high at the middle range of e(b - v ) between 0.3 and 0.6 for the massive galaxies but at the same time we note that the most reddened objects have few x - ray counterparts .
our results in the last section largely owe to the photometric redshift estimation .
although the 40% of the x - ray sources have the secure spectroscopic redshift values and the comparison between photometric and spectroscopic redshift for the @xmath0-selected galaxies is also reasonable ( kajisawa and yamada 2005 ) , it is worth analyzing the similar sample based on the simple color cut avoiding the uncertainty of the detailed photometric redshift measurements . for the purpose , we defined the sample of _ distant red galaxies _ ( drgs ) , which have @xmath56 @xmath88 ( franx _ et al .
_ 2003 ; van dokkum _
drgs are supposed to be massive galaxies ( @xmath89 ) at @xmath90 ( franx _ et al . _ 2003 ; frster schreiber _
_ 2004 ; reddy _ et al .
_ 2005 ; kajisawa _ et al .
_ 2006 ) or dusty galaxies at intermediate redshift , @xmath91 ( webb _ et al .
_ 2006 ; conselice _ et al .
_ 2007 ; lane _ et al .
. while more than a half of the relatively bright drgs with @xmath92 have @xmath93 , the fainter drgs are likely to be dominated by the galaxies at high redshift ( kajisawa _ et al .
this simple color cut is thus useful to make a sample of massive galaxies at @xmath94 . above @xmath95 , the completeness limit of the color selection over gt1-gt4 fields ,
182 drgs are identified .
fig.10 shows the number counts of the whole drgs as well as drgs with the x - ray counterparts .
their color - magnitude distribution is plotted in fig.11 .
the vertical dotted line in fig.10 shows the detection limit for drg , @xmath0=22.8 .
note that the decrease of the number density of drgs is seen at around @xmath0=22 , which is more rapid than that of the entire @xmath0-band selected objects .
this indicates that there are less red low mass galaxies at high redshift ( kajisawa _ et al .
the x - ray drgs are peaked at @xmath96 mag , which corresponds to the massive galaxies with m@xmath7 @xmath97 @xmath9 m@xmath5 at @xmath98 . at @xmath99 ,
30 - 35% of the drgs ( 19/65 for the hard band , 23/65 for the total band ) are identified with the x - ray sources while it decreases to 21% at @xmath100 .
these analysis for the sample of drgs with the simple color cuts thus support the trend discussed in the previous section , namely , a large x - ray fraction among the bright ( @xmath101 ) or massive objects , and decrease of the x - ray fraction at the fainter magnitude . among the @xmath32-band selected galaxies at @xmath23 - 4 ,
the x - ray emissions are exclusively associated with the massive objects , with m@xmath7 @xmath102 m@xmath5 . as the _ chandra _ 2m - sec data is deep enough to constrain the presence of moderately - luminous agn even at such high redshift
, we may discuss the possible bh mass limits as a function of their host galaxy mass .
the limiting x - ray flux @xmath103(2 - 8 kev)=@xmath104
erg s@xmath3 @xmath70 corresponds to @xmath24(2 - 10 kev ) = @xmath105 erg s@xmath3 to @xmath106 erg s@xmath3 at @xmath23 to 4 , which corresponds to the luminosity of the agn such as luminous seyfert galaxies in nearby universe .
if we assume the bolometric fraction of the x - ray emission at 2 - 10 kev , @xmath107 and the ratio of the bolometric to the eddington luminosity of the agn , @xmath108 , the observed x - ray luminosity can be connected to the black hole mass by the relationship , @xmath109 @xmath110 .
the absence of luminous agn in the less massive galaxies indicates that the associated bh mass or the activity of the agn is lower than the limit .
unfortunately from the current data we can not discriminate which is the case ; we need further spectroscopic observations to evaluate virial bh mass to break this degeneracy .
instead , we here discuss the two cases with rather extreme assumptions to see which the more natural interpretation is .
we first consider the case that smbh mass does not depend on the host stellar mass , namely similar massive ( or less massive ) bhs exist in massive or less massive host galaxies . in this case , the x - ray non - detection in less massive galaxies below m@xmath7 @xmath111 m@xmath5 is due to the lower activity ( less mass accretion rate ) of their bh .
this is the opposite trend seen for the _ x - ray detected _ sample above @xmath111 m@xmath5 as shown in fig.7 and thus not favorable ; it will be strange that the activity of agn falls off suddenly at @xmath14 m@xmath5 .
we then consider the second case that the m@xmath6/m@xmath12 ratio is constant and does not evolve from the local value , m@xmath6/m@xmath12@xmath112 and not depends on the host stellar mass .
we also assume m@xmath7 @xmath68 m@xmath12 . in this case , the decrease of specific agn activity among the x - ray _ detected _ sample above @xmath14 m@xmath5 shown in fig.7 can be interpreted as the decrease of relative mass accretion rate ; we see that the eddington ratio of smbh in the host with m@xmath113 @xmath114 m@xmath5 ( or the corresponding m@xmath115 @xmath116 m@xmath5 ) is @xmath117 and it becomes lower , @xmath108=0.1 - 0.01 , for the more massive objects . on the other hand ,
few x - ray agn is detected for galaxies with the stellar mass less than 10@xmath16 m@xmath5 ( fig.2 ) , which should host smbh with m@xmath6 @xmath118 m@xmath5 with the assumption above .
this bh mass expected from the no - evolution hypothesis , which may be the upper limit values if we consider the lower spheroidal mass fraction , is comparable with the bh mass detection limit for the current _ chandra _ x - ray data , @xmath119 m@xmath5 and @xmath120 m@xmath5 for @xmath121 at @xmath23 and @xmath122 , respectively , and @xmath123 and @xmath124 m@xmath5 for @xmath125 . thus the latter interpretation seems more natural .
our results are at least consistent with that , at @xmath8 , the galaxies with stellar mass with @xmath14 m@xmath5 or less do not host the black hole which is significantly ( i.e. , more than an order of magnitude ) heavier than that expected from the local relationship , i.e. , m@xmath6@xmath126 m@xmath5 unless the accretion rate is very low , @xmath127 . mclure _ et al . _
( 2006 ) argue for the sample of 3c rr radio - loud objects that m@xmath6/m@xmath12 ratio is an increasing function with redshift and the ratio is @xmath128 larger at @xmath19 .
these 3c rr galaxies are very massive , m@xmath7 @xmath129 m@xmath5 and their bh mass is estimated to be m@xmath6 @xmath130 m@xmath5 .
if the trend is also true for the galaxies in our sample , the galaxies with m@xmath7 @xmath14 m@xmath5 may have the bh with m@xmath6 @xmath131 m@xmath5 , which is not favored in our results .
the trend that the agn activity seen only among the massive galaxies is very similar with that seen for the local galaxies .
( 2003 ) investigated the host properties of the @xmath132 agns among the @xmath133 galaxies at @xmath134 and found that agn with all luminosity reside almost exclusively in massive galaxies , with m@xmath7@xmath135m@xmath5 .
while we have to consider the x - ray detection limits carefully , the current results are consistent with that agn are observed only among the massive galaxies even at @xmath8 .
on the other hand , a steep number - density evolution of massive galaxies from @xmath98 to @xmath136 is reported ( caputi _ et al .
it is very likely only 10 - 30% of the present - day massive ( @xmath137 m@xmath62 ) galaxies have been formed at @xmath98 and if their progenitors already exist at such high redshift , they may be less massive galaxies which will be merged with other objects or forming more stars to grow .
it is an interesting question whether such progenitors have already hosted the massive bhs which are observed in their present - day descendants or not . in our results , galaxies with smaller mass ( @xmath138 a few @xmath139 m@xmath5 )
do not appear to host the bh as massive as those located in the present - day massive ( @xmath137 m@xmath62 ) galaxies . in this sense , it is unlikely that massive bh had already been formed in the less - massive progenitors of their present - day host galaxies .
we may argue that massive bhs have been developed as the host galaxies have grown their stellar mass , and not significantly prior to that .
it is also surprising that a very large fraction ( @xmath140% ) of the massive galaxies host the _ active _ agn . if the m@xmath6/m@xmath7 ratio for the massive galaxies does not differ from the local value , @xmath112 , many of these agn are active with the eddington ratio @xmath141 as we can see in fig.7 .
while the observed x - ray luminosity and sed - fitting - based sfr are not strongly correlated and the bh formation ( mass accretion ) and the star formation seems not coeval exactly ( fig.8 ) , it can be said that the active duration in the agn duty cycle of the high - redshift massive galaxies is large .
at the same time , we already noted that the negative dependence of the specific agn activity on the stellar mass is observed .
we also investigated how the specific agn activity depends on the averaged stellar age obtained in the sed fitting .
the results are shown in fig.12 .
while they are scattered out and no tight correlation exists , we see a broad trend that the agn activity is high in the young host galaxies with the stellar age less than @xmath142 yr and relatively low in the galaxies which are older than @xmath143 yr ; 8 of 11 galaxies with @xmath144 @xmath145 yr have @xmath146 and 9 of 14 galaxies with @xmath144 @xmath147 yr have @xmath148 . from these above
, we may infer a possible scenario that the formation of the bhs in massive galaxies continues over the large fraction ( i.e. , @xmath149 ) of the period of their star formation although not instantaneously coeval . in the @xmath0-band - selected very massive ( @xmath150 m@xmath5 ) galaxies at @xmath8 ,
the bh growth is still ongoing but as the star - formation activity peaked earlier in more massive galaxies , as inferred from their relatively older age and lower specific sfr , their agn activities may also be peaked earlier and only relatively lower - level activity is seen at the observed epoch .
we investigated the x - ray properties of the @xmath0-band selected galaxies at @xmath1 with the stellar mass of 10@xmath151 - 10@xmath152 m@xmath5 in 103 arcmin@xmath33 in goods - n . @xmath153
x - ray sources , most of which must be agn , are identified with the galaxies and we found that they are exclusively ( @xmath6590@xmath53 ) associated with the massive objects with the stellar mass @xmath102 m@xmath5 . on the other hand , a very large fraction , 30 - 35% of the galaxies with stellar mass larger than 10@xmath47 m@xmath5 shows the agn activity with the x - ray luminosity larger than @xmath154 erg s@xmath3 .
we also found that the specific agn activity , namely the x - ray luminosity divided by the stellar mass is smaller in the more massive galaxies .
the absence of luminous agn in the less massive galaxies implies that the associated bh mass is smaller or the activity of the agn is lower .
our results are consistent with the idea that the m@xmath6/m@xmath7 ratio in @xmath8 massive and less massive galaxies is similar to the present - day value
. since the x - ray luminosity and the sfr are not correlated , the growth of bh and the star formation in the host seems not exactly coeval . at the same time
, the large x - ray fraction of high - redshift massive galaxies also indicates that a significant fraction of the agn activity continues over the comparable time scale of the star - formation in the host .
we thank the staff of the subaru telescope for their assistance with the development and the observation of moircs .
this research is supported in part by the grant - in - aid 20450224 for scientific research of the ministry of education , science , culture , and sports in japan .
data analysis was in part carried out on common use data analysis computer system at the astronomy data center , adc , of the national astronomical observatory of japan . | we investigate the x - ray properties of the @xmath0-band - selected galaxies at redshift @xmath1 by using our deep near - infrared images obtained in the moircs deep survey project and the published _
chandra _ x - ray source catalog .
61 x - ray sources with the 2 - 10 kev luminosity @xmath2 erg s@xmath3 are identified with the @xmath0-selected galaxies and we found that they are exclusively ( 90% ) associated with the massive objects with stellar mass larger than @xmath4 m@xmath5 .
our results are consistent with the idea that the m@xmath6/m@xmath7 ratio of the galaxies at @xmath8 is similar to the present - day value . on the other hand , the agn detection rate among the very massive galaxies with the stellar mass larger than @xmath9 m@xmath5 is high , @xmath10% ( 26/78 ) .
they are active objects in the sense that the black - hole mass accretion rate is @xmath11 - 50% of the eddington limit if they indeed have similar m@xmath6/m@xmath7 ratio with those observed in the local universe .
the active duration in the agn duty cycle of the high - redshift massive galaxies seems large . |
the question whether traditional leading twist qcd is the relevant description of hera small-@xmath0 data has still no clear - cut answer .
a particularly successful saturation model proposed by golec - biernat and wsthoff ( gbw ) @xcite suggests that the common picture of a proton looking like a dilute set of partons might be unjustified .
we have recently shown _ directly _ that the proton is a dense system already at present hera energies @xcite .
the method we used is summarized in the next section .
we then dedicate a section to a by - product of our analysis : the measurement of total dipole - proton cross section .
wave diffraction allows to obtain a picture of a microscopic object .
a fourier transform relates its density profile to the square root of the intensity of light measured on the interference pattern .
it turns out that one can analyse the hera data for diffractive vector meson production in this way .
indeed , at high energy , such a process is equivalent to elastic diffusion of dipoles .
the picture is the following @xcite : in an appropriate reference frame , the photon of virtuality @xmath1 splits in a @xmath2 dipole which scatters elastically off the target proton before recombining into a meson .
the size of the interacting dipole is distributed around a mean value @xmath3 and the total flux @xmath4 is the scalar product of photon and meson wave functions @xmath5 .
the density profile is extracted by taking the fourier transform of the square root of the differential cross section @xmath6 with respect to the momentum transfer @xmath7 ( @xmath8 ) .
the variable conjugated to @xmath7 is the impact parameter of the dipole relatively to the center of the proton .
one then sees that this profile is related to the @xmath9-matrix element for elastic dipole - proton scattering ( a key observation is that @xmath9 is essentially real at high energy ) .
it reads @xmath10 as a phenomenological model is needed for the meson wave function @xmath11 which appears in @xmath4 , checks of model - independence of the method are in order .
we checked that the dipole size is indeed distributed around a mean size @xmath12 independently of the model ( @xmath13 is the mass of the meson ) .
we also verified that the dipole flux @xmath4 is strongly constrained by the well - known photon wave function @xmath14 , and thus is quite model - independent .
we applied formula ( [ eq : s ] ) to the zeus data for diffractive electroproduction of longitudinal @xmath15 mesons @xcite .
as saturation is a high energy effect , we considered the lowest available values of @xmath0 ( @xmath16 ) only .
we varied the virtuality @xmath1 of the photon so that the effective scale @xmath17 is always larger than @xmath18 .
the result of our analysis for 3 different values of the photon virtuality is represented on fig.1 .
the estimated error shown stems from the lack of data for large momentum transfer @xmath19 .
it does not include experimental errors on the measured quantities .
why is @xmath9 a good estimator of the importance of saturation effects ? as argued before , the value of @xmath9 tells us about how dense the proton looks .
@xmath20 means blackness : it is the unitarity limit . for a more quantitative interpretation
, one observes that @xmath21 is the interaction probability of a dipole that hits the proton at impact parameter @xmath22 .
if this probability is significant , it means that more rescatterings are likely to occur .
we see that the interaction probability is more than 50% and could already reach 75% at the center of the proton .
a more common parameter characterizing the saturation regime is the saturation scale @xmath23 . roughly speaking
, it is defined as the maximal virtuality for which the photon sees the proton as a dense medium .
we found @xmath23 of order 1 to 1.5 @xmath24 near the center of the proton , and 0.2 @xmath24 on the periphery .
we refer the interested reader to ref.@xcite for all details of the analysis and more results .
having the @xmath9-matrix , one can easily obtain the total dipole - proton cross section .
it is an interesting quantity because of its universality .
one needs the forward differential cross section from the data and the dipole flux @xmath4 and average dipole size @xmath3 from the model .
the cross section reads @xmath25 in pratice , we give an upper and a lower bound for @xmath3 .
an approximate lower bound is obtained when the dipole size distribution is weighted by a flat ( saturated ) cross section ( i.e. @xmath26 ) .
the upper bound comes for a colour transparent cross section @xmath27 .
a realistic average value can be computed by taking an interpolating weight .
we took gbw cross section , and we checked that the expectation value @xmath3 is not very much dependent on the exact point where the transition occurs , as long as it is around @xmath28 .
the result is plotted on fig.[fig : stot ] .
we have shown that perturbative saturation effects are already important at hera .
our argument was that the proton appears quite dense to a relatively small size probe .
we were able to give a quantitative estimate of its darkness by measuring the @xmath9-matrix element for elastic dipole - proton scattering .
it turns out that the probability that a dipole of size @xmath29 fm undergoes an inelastic interaction is more than @xmath30 at small impact parameters , and for the high energy data from hera .
this translates into a saturation scale @xmath23 lying between @xmath31 and @xmath32 which is consistent with the assumptions contained in gbw model .
more exclusive data for quasi - elastic processes at large momentum transfer on one hand , and impact parameter - dependent analysis on the other hand , would be very welcome in the future since saturation effects are expected to be maximal for central collisions .
+ [ _ the interesting question whether this statement is in contradiction with the successful description of vector meson photoproduction at large @xmath33 by 2-gluon exchange enhanced by bfkl corrections was raised by martin mcdermott during the discussion . from our result ,
for high energies ( @xmath34 ) , low @xmath1 and large @xmath33 diffractive processes , one would expect significant saturation effects . in any case
, it would be very instructive to construct a model of gbw type for large @xmath33 processes .
_ ] n.n .
nikolaev and b.g .
zakharov , _
_ * c49 * ( 1991 ) 607 ; _ z. phys _ * c53 * ( 1992 ) 331 ; _ z. phys .
_ * c64 * ( 1994 ) 651 ; _ jetp _ * 78 * ( 1994 ) 598 ; j. nemchik , n.n .
nikolaev , b.g .
zakharov , _
_ * b341 * ( 1994 ) 228 ; a. h. mueller , nucl .
* b415 * ( 1994 ) 373 . | through an analysis of diffractive vector meson production , we show that the proton appears quite dense to a small size probe at present hera energies .
this means that saturation effects are already important . |
ferroelectric materials feature in many modern day electronic devices including computer memory and capacitors , and are a simple setting for studying quantum criticality @xcite . in this paper
we focus on the family of displacive ferroelectrics where the optical lattice modes condense , forming a structural distortion . near to quantum criticality excitations
can become highly degenerate and new phases can emerge .
motivated by recent experiments that signal the emergence of novel quantum critical behavior in ferroelectrics @xcite , we explore the possibility that transverse components of polar fluctuating phonons conspire to drive a first order displacive metaelectric transition and investigate the implications for the inverse susceptibility
. the soft - mode optical phonons in ferroelectrics can be well - described by a bosonic field theory .
if the dynamics were not damped by free electrons and the interactions remain short - ranged then the general quantum critical behavior would adhere to the well - established rules reviewed in ref .
@xcite . however , in ferroelectrics the motion of the atoms in optical modes leads to the emergence of electric dipoles .
a good description of these long - range dipole forces is essential to properly describe the ferroelectric transition .
the effect of long - range dipolar forces was first studied by @xcite and @xcite .
@xcite found that anisotropies associated with the dipolar interaction led to a universality class in the classical ferroelectric .
the quantum ferroelectric phase transition in the mean - field approximation , and its universality class , was studied by @xcite .
however , recent experimental evidence points to new physics that emerges close to quantum criticality ; for example the coexistence of a quantum paraelectric phase with a quantum ferroelectric phase in @xmath1o - exchanged srtio@xmath2 provides strong evidence for a first order phase transition @xcite .
additional motivation to study ferroelectrics arises from the inverse dielectric constant behavior of srtio@xmath2 which falls at low temperature before increasing as @xmath3 at intermediate temperatures and rises as @xmath4 at high temperature .
one suggestion is that new phenomena are driven by the coupling of acoustic to optical phonons @xcite .
however , inspired by the ramifications of quantum fluctuations in ferromagnets @xcite , we show that the transverse coupling of fluctuating phonons can drive a first order metaelectric transition . having realized that fluctuations can cause the emergence of a first order transition it is natural to search for further phase reconstruction .
motivated by the development of a textured fflo phase @xcite , and evidence for a textured ferromagnetic state near to the ferromagnetic first order transition @xcite , here we search for the emergence of an analogous textured ferroelectric phase . finally , to connect to prevailing experimental methods
, we derive an appropriate expression for the inverse susceptibility that is consistent with recent experimental results @xcite over a wide range of the phase diagram , and demonstrate that the transverse coupling of fluctuating phonons could cause it to have a characteristic minimum at low temperature .
we adopt a bosonic field theory to describe the soft optical phonon modes that should recover the main physical behavior of the system .
the order parameter of the theory is the local polarization @xmath5 , which is formally defined for one unit cell at @xmath6 containing @xmath7 atoms of charge @xmath8 each individually displaced through @xmath9 by the optic mode .
as the optical phonon softens , the action develops an instability and the order parameter must describe both thermal and quantum fluctuations .
following @xcite we describe the action in three - dimensional space and imaginary time via the ginzburg - landau phenomenology @xmath10\phi_{\alpha}({\mathbf{q}})\phi_{\beta}({\mathbf{-q}}){\nonumber\\}\!\!\!\!\!\!\!\!&&+\!\!\!\!\!\!\sum_{\alpha,\beta,\{{\mathbf{q}}_{i}\}}\!\!\!\!\!\!\left(u+v\delta_{\alpha,\beta}\right)\phi_{\alpha}({\mathbf{q}}_{1})\phi_{\alpha}({\mathbf{q}}_{2})\phi_{\beta}({\mathbf{q}}_{3})\phi_{\beta}({\mathbf{q}}_{4})\biggr\}{\text{d}}\tau{\,,}\end{aligned}\ ] ] where @xmath11 is the lattice constant , @xmath12 is the speed of the phonons , @xmath13 , the dimensionless momenta @xmath14 , and the second summation is carried out under the conservation of momentum @xmath15 . since the field @xmath16 describes an electric dipole , the action includes a long - range dipole interaction , and also a coupling to the underlying lattice through the parameters @xmath17 , @xmath18 , @xmath19 , and @xmath20 .
the terms @xmath21 and @xmath22 that describe the local anharmonic interactions give a net positive contribution which ensures that the polarization remains bounded . in general these parameters
are tensorial , but for simplicity we have assumed that they adopt cubic symmetry .
estimates for the parameters shown in table [ tab : srtio3params ] were obtained from _ ab initio _
calculations @xcite in the two key ferroelectrics srtio@xmath2 and ktao@xmath2 @xcite .
the typical energy scale of ferroelectric fluctuations along ( 100 ) is @xmath23 ; using this definition we can then employ a dimensionless bosonic matsubara frequency @xmath24 , and a dimensionless temperature @xmath25 . throughout the paper
we adopt the units @xmath26 .
.model parameters for the ferroelectrics srtio@xmath2 and ktao@xmath2 @xcite . [ cols="^,^,^,^,^,^,^,^",options="header " , ] to establish the connection to previous work we first consider the mean - field phase diagram that is sketched in fig .
[ fig : meanfieldphasediagram ] . making the ansatz that the ground state is uniform we obtain the action @xmath27 , where @xmath28 .
when @xmath29 the polarization @xmath30 , @xmath31 has an ising configuration , whereas when @xmath32 the polarization @xmath33 exhibits diagonal order .
the term proportional to @xmath22 controls the polarization direction in the ferroelectric phase , whereas the @xmath21 term is rotationally invariant .
we note that whilst sweeping @xmath22 through @xmath34 with @xmath35 the first order rotation of polarization direction is accompanied with a continuous change in the magnitude of the polarization .
this is driven by a similar mechanism to the blume - emery - griffiths model involving two bosonic fields @xcite . within the mean - field approximation
the condition for stability of the polarization is that the net coefficient of the quartic term is positive which translates to @xmath36 when @xmath29 and @xmath37 if @xmath32 .
if these conditions are not fulfilled then higher order terms must be included and rather than undergo a second order transition at @xmath38 , the system might have a first order ferroelectric transition at mean - field level .
we can neglect the higher order terms such as @xmath39 provided that the model remains stable , which requires that @xmath40 . here
we wish to investigate whether near criticality the fluctuating modes can conspire to drive an otherwise second order transition to become first order . in order to access this behavior
we now go beyond mean - field and consider the consequences of quantum fluctuations on the system .
to account for fluctuation corrections to the system @xcite employed the renormalization group , which is tailored to study the well - established second order ferroelectric transition .
however , motivated by recent experiments @xcite we wish to explore the possibility of a first order metaelectric transition . therefore , rather than considering just the corrections due to slow fluctuations that are encompassed by renormalization group
, we need to consider fluctuations @xmath41 over all length scales in the polarization @xmath42 around the saddle - point solution @xmath16 . when @xmath43 we can neglect fluctuations in @xmath41 beyond second order which reduces the action to @xmath44{\nonumber\\}&+&\tilde{\beta}\sum_{\tilde{\omega},{\mathbf{q}}}{\bm\psi}^{\text{t}}(\tilde{\omega},{\mathbf{q}}){\mathbb{g}}^{-1}{\bm\psi}(-\tilde{\omega},-{\mathbf{q}}){\,,}\end{aligned}\ ] ] where @xmath45 , the diagonal inverse green function takes the form @xmath46 , and the off - diagonal terms are @xmath47 .
we now integrate over quantum fluctuations to yield the free energy @xmath48 where @xmath49 is the dimensionless inverse temperature . if @xmath50 and @xmath51 or if @xmath52 and @xmath53 then @xmath54 . in this regime
we can expand the inverse green function in its off - diagonal terms @xmath55 using @xmath56 which enables us to describe the renormalization of fluctuations by off - diagonal coupling .
this yields @xmath57\!\!-\!\!\ln\!\left[\!\frac{\tilde{\beta}\xi^{\alpha}_{{\mathbf{0}}}}{2}\!\right]\!\right)\!-\!\frac{1}{4\tilde{\beta}}{\operatorname{tr}}\left(ugug\right)\!,\end{aligned}\ ] ] where @xmath58^{1/2}$ ] . to remove the fluctuations of the static uniform component of @xmath59 , which are included in @xmath60
, we must introduce the second logarithm .
this has the effect of regularizing the divergence which would otherwise develop from the first logarithm .
this expression , except for the final fluctuation correction term , agrees with that of ref .
@xcite , and is analogous to the coupling of transverse ferromagnetic fluctuations that led the emergence of first order behavior @xcite .
the condition for stability is the same as for the mean - field case .
the momentum integrals are in general evaluated numerically .
however , to further investigate the diagonal ordered phase we make the approximation that the cuboid brillouin zone boundary ( @xmath61 ) that bounds the momentum space integral can be replaced with a spherical boundary that encloses the same total phase space , so has radius @xmath62{6\pi^{2}}$ ] . in the low temperature limit with the polarization aligned in the @xmath63 direction , the resulting integrals
can then be evaluated analytically to yield @xmath64\nonumber\\ & \!\!+&\!\!\!\frac{u^{2}\phi^{2}}{6\pi^{2}}\left[\frac{2\pi}{\sqrt{\xi+\pi^{2}}}-2\ln\left(\frac{\pi}{\sqrt{\xi}}+\sqrt{1+\frac{\pi^{2}}{\xi}}\right)\right]{\,,}\end{aligned}\ ] ] with @xmath65 , and were found to be in good agreement with the corresponding numerical result . the phase behavior of the system is shown in fig .
[ fig : phasediagram ] .
the forbidden region indicates where the action polarizability and free energy would diverge without considering higher order corrections to the original action .
when considered within the framework of mean - field phenomenology , here the system could undergo a first order paraelectric - ferroelectric transition .
however , the corrections due to quantum fluctuations renormalize the action , causing a metaelectric boundary to peel away from the first order transition associated with the forbidden region .
this metaelectric transition is consistent with recent experimental evidence for a first order phase transition @xcite in @xmath1o - exchanged srtio@xmath2 . in both of the planes
considered , the line of first order metaelectric transitions covers an extensive region of the phase diagram , terminating in a tricritical point at @xmath66 .
the first order transition at small @xmath21 is destroyed at non - zero temperature , with the tricritical point moving up the line of transitions to @xmath67 .
this critical behavior does not depend on the long - range dipole interactions since the lowest order term in @xmath19 and @xmath20 averages to zero on integrating over momenta @xcite .
the @xmath68 symmetry could be destroyed by applying a uniaxial electric field misaligned to the lattice .
a further ramification of the quantum fluctuation corrections is that the rotation of the polarization from ising to diagonal order no longer occurs where @xmath22 turns negative .
though , as for the mean - field case , the magnitude of the polarization is conserved ; fluctuations have renormalized the quartic terms and shifted the phase boundary in fig .
[ fig : phasediagram](b ) .
this behavior can also be recovered by a renormalization group analysis @xcite .
one experimental probe of the metaelectric transition is the changing behavior of the heat capacity @xmath69 . before the metaelectric transition ( small negative @xmath17 ) the relevant optic mode is `` soft '' and so the heat capacity follows the familiar debye form @xmath70 , whereas after the metaelectric transition ( large negative @xmath17 ) , the relevant optic modes are `` stiff '' and so the heat capacity has an exponential dependence on temperature . at high temperature , in both cases
the heat capacity has the expected classical behavior @xmath71 .
having confirmed the existence of a possible metaelectric behavior , we now turn to consider the stability of the phase in the vicinity of the transition .
recent studies of itinerant ferromagnetism have suggested that such first order behavior can be preempted by the development of textured magnetic order analogous to that seen in the fflo phase of superconductors @xcite .
this leaves open the question as to whether a textured phase can develop in the vicinity of the metaelectric transition .
our strategy to explore this possibility is to assume that the inhomogeneous phase is formed continuously , which allows us to develop a landau expansion in the polarization @xmath72 and texture wave vector @xmath73 .
the onset of an inhomogeneous phase is signaled by the coefficient of the @xmath74 term turning negative . in our analysis
we search primarily in the vicinity of the metaelectric transition at @xmath75 and consider a trial state with uniform polarization @xmath60 that is for simplicity superimposed by an inhomogeneous component @xmath76 .
we then expand the free energy to quartic order in @xmath73 and discover that the presence of a textured phase makes a contribution to the total energy of @xmath77 $ ] .
short of the first order transition where @xmath78 , the coefficient of @xmath79 is positive so the phase is not modulated . after the first order transition @xmath80 turns positive driving the coefficient of @xmath79 negative , revealing a finite @xmath81 instability in the region highlighted in fig .
[ fig : phasediagram](b ) .
the modulation carries polarization @xmath82 . though the analysis is restricted to the consideration of a potential continuous transition into the textured phase , and a simple form for the texture , it is sufficient to validate its existence .
refinements to include a putative first order transition or further textured phases would only enlarge the region of the phase diagram over which inhomogeneities could be observed .
leaving aside potential textured phases we now turn to consider the behavior of the susceptibility across the phase diagram .
the inverse susceptibility provides an experimental window @xcite onto the quantum critical properties of ferroelectrics .
deep in the paraelectric regime where @xmath83 , the contribution to the inverse susceptibility is @xmath84 , which is consistent with barrett s formula @xcite for a gapped system . in the quantum critical regime
we see three characteristic types of behavior for the inverse susceptibility @xmath85 where the first term is from the diagonal contributions to eqn . , and the latter two terms are the off - diagonal contribution , and @xmath86 . at low temperature effects of long - range dipole interactions
prevail as the off - diagonal fluctuating contribution renormalizes the on - diagonal terms , with the linear temperature dependence of the term proportional to @xmath19 giving a positive slope to the inverse susceptibility whereas the @xmath87 term could provide a negative slope . at higher temperatures
the @xmath88 contribution from the mean - field term dominates , which is also characteristic of quantum critical behavior and is in good agreement with recent experimental results @xcite .
we note that the @xmath88 behavior is recovered by other models , including a diagrammatic resummation @xcite , the quantum spherical model @xcite , renormalization group studies @xcite , a self - consistent phonon model @xcite , and an analogy to the temporal casimir effect @xcite .
the behavior has also been observed experimentally @xcite . in both srtio@xmath2 and ktao@xmath2 the initial linear negative slope and
the quadratic @xmath89 term conspire to cause a characteristic minimum in the inverse susceptibility . using estimates for the parameters in table [ tab : srtio3params ] ,
the minimum occurs at @xmath90 in both srtio@xmath2 and ktao@xmath2 which is in good agreement with the experimental values of @xmath91 and @xmath92 respectively @xcite
. finally , at high temperatures a classical term @xmath93 from the longitudinal fluctuating term dominates from @xmath94 which is again in good agreement with the experimental observations @xcite .
in this paper we have found that the polar fluctuating phonons can drive a displacive ferroelectric through a first order metaelectric transition .
long range dipolar interactions did not affect this critical phase behavior @xcite . however , long - range dipole interactions introduced into the action through the term @xmath95 were pivotal in creating the correction to the inverse susceptibility @xmath96 that could explain the characteristic inverse susceptibility minimum @xcite , as well as provide important corrections to the self - consistent phonon treatment @xcite
. however , another mechanism , coupling of the soft optic modes to acoustic phonons could be significant .
it has already been understood @xcite that a coupling with the acoustic phonons @xmath97 of the form @xmath98 leads to a correction in @xmath99 of @xmath100 that could explain the characteristic minimum in the inverse susceptibility , and also has the capability of driving a first order transition @xcite .
this work and the results presented here motivate further experimental investigations into the inverse susceptibility and putative metaelectric transition that could shed light on the origin of the phase structure .
though the coupling to acoustic phonons complicated the solid state system , ultracold atoms in an optical lattice with long - range dipole interactions @xcite present a clean system that could provide a powerful tools to help unravel the properties of the generic hamiltonian .
one important simplification was to model the ferroelectric with undamped dynamics .
damping would primarily arise due to free electrons , which can be introduced controllably through doping .
analogous to `` avoided criticality '' at a magnetic critical point which leads to non - fermi liquid behavior and superconductivity , ferroelectrics might also be expected to adopt novel behavior ; for example doped srtio@xmath2 @xcite , whereas undoped snte @xcite and gete @xcite become superconducting at low temperatures .
this area presents a promising avenue of research .
further open questions are to determine whether with just a change of parameters @xcite the same formalism be applied to order - disorder ferroelectrics , and to consider the consequences of the coupling of fluctuating polarization and magnetization that could arise in eutio@xmath2 @xcite . | we present a microscopic model of the quantum paraelectric - ferroelectric phase transition with a focus on the influence of coupled fluctuating phonon modes .
these may drive the continuous phase transition first order through a metaelectric transition and furthermore stimulate the emergence of a textured phase that preempts the transition .
we discuss two further consequences of fluctuations , firstly for the heat capacity , and secondly we show that the inverse paraelectric susceptibility displays @xmath0 quantum critical behavior , and can also adopt a characteristic minimum with temperature .
finally , we discuss the observable consequences of our results . |
the sequencing of the human reference genome , announced ten years ago , provided a roadmap that is the basic foundation for modern biomedical research @xcite .
this monumental achievement was enabled by developments in dna ( homopolymer ) sequencing technology that allowed data production which exceed the original descripion of sanger sequencing @xcite .
linear polymers are the simplest physical systems that can be studied in the framework of random walks models .
they are long chain - like molecules formed by repetition of a basic unit or segment , where more importantly the polymer is _ flexible _ , i.e. , it can assume different geometric configurations .
recently , the study of homopolymers has been established by various techniques in condensed matter physics .
@xcite , studied the behavior of single file translocation of a homopolymer through an active channel under the presence of a driving force by using langevin dynamics simulation .
previous works on homopolymers which studied the denaturation of circular dna are extensions of the poland - scheraga model @xcite .
experimentally , the viscoelastic properties of a binary mixture of a mesogenic side - chain block copolymer in a low molecular weight nematic liquid crystal are studied for mass concentrations ranging from the diluted regime up to a liquid crystalline gel state @xcite .
although monte carlo simulations play an important role for the study of phase transitions and critical phenomena , some well - known difficulties arise when one uses standard algorithms ( one - flip algorithms ) @xcite for the study of random walks models .
these difficulties have been overcomed by the development of alternative monte carlo methods , such as parallel - tempering @xcite , cluster algorithms @xcite , multicanonical algorithms @xcite , and more recently the wang - landau method @xcite .
this method has been applied with great success to many systems , in particular to polymers in lattice @xcite . in the present paper , using the wang - landau method , we investigate the computer simulations of a homopolymer model with exact solution in canonical and microcanonical formalism . in section [ models ]
we give an introduction to the three studied models and we briefly present the mathematical background . in section [ simulations ]
we outline shortly how the simulations were carried out . in section [ results ]
we show and discuss the results for all the models .
we consider a homopolymer chain of @xmath0 monomers units of length @xmath1 which may assume any configuration on a two - dimensional latice @xcite .
one end of the polymer is fixed and its taken as the origin of coordinates , shown by an open circle in figure [ model ] , a ) .
the other end of this linear chain is subject to an externally applied tension @xmath2 , acting along the positive @xmath3-axis .
a possible realization of this model is sketched in figure [ model ] , a ) .
each polymer unit is permitted to lie either parallel or antiparallel to the @xmath3-axis and we assign the works @xmath4 and @xmath5 to these two orientations .
we denote by @xmath6 the distance between the ends of the polymer chain , on the @xmath3-direction .
each monomer unit has the additional possibility of lying perpendicular to the @xmath3-axis , in the @xmath7 or @xmath8 directions .
we associate a positive energy @xmath9 to such a perpendicular monomer and the distance between the ends of the chain on the @xmath10-direction is denoted by @xmath11 .
the hamiltonian of this model can be written as @xmath12 where @xmath13 and @xmath14 are the number of monomers along the @xmath15 and @xmath16 directions respectively , and similarly for @xmath17 and @xmath18 .
since the tension in the @xmath10 direction is zero , we can assume @xmath19 .
then @xmath20 @xmath21 @xmath22 from which we find @xmath23 @xmath24 and @xmath25 the number of configurations of the polymer consistent with a given end - to - end distance in the @xmath3-direction , @xmath26 ( the dimensionless length of the polymer ) , and @xmath27 ( the number of monomers lying perpendicular to the @xmath3-axis ) , is @xmath28 from eq .
( [ h_1 ] ) , if we set @xmath29 , we can write for a given energy level @xmath30 and from eq .
( [ n_1 ] ) @xmath31 adding the equations ( [ e_1 ] ) and ( [ n1_1 ] ) we obtain @xmath32 inserting eqs .
( [ nx_1 ] ) , ( [ nx1_1 ] ) and ( [ ny_1 ] ) into ( [ omega_1 ] ) , using eqs .
( [ e_1 ] ) , ( [ n1_1 ] ) , and ( [ u_nx_1 ] ) , and setting @xmath33 we obtain the number of configurations with energy @xmath34 as @xmath35 using the definition of the entropy and defining @xmath36 , we obtain @xcite : @xmath37 in this model we consider that each polymer unit is parallel to the @xmath3-axis ; no antiparallel move is allowed .
additionally the polymer units have the possibility of lying on the @xmath7 and @xmath8 directions . in figure [ model ] b ) is sketched a possible configuration of this model , where we depicted by an impenetrable wall the forbidden region along the @xmath3-axis .
the hamiltonian for the model @xmath38 can be written as @xmath39 where @xmath40 is the number of monomers along the @xmath15 direction and similarly for @xmath41 and @xmath42 .
assuming again @xmath43 we obtain the following equations : @xmath44 and @xmath45 for this model the number of configurations of the polymer consistent with a given @xmath26 and @xmath27 is @xmath46 from eq .
( [ h_2 ] ) , if we set @xmath29 , we can write the energy as @xmath47 and for @xmath48 we obtain @xmath49 after similar calculations as for model @xmath50 , subsection [ model1 ] , we obtain the density state and end - to - end distance per monomer , respectively @xmath51 and @xmath52 in this model we consider that each polymer unit is parallel to the @xmath3-axis and we allow only the possibility of lying on the @xmath7 direction .
thus , we restrict to the situation of positive values for both axes . in figure [ model ] c ) we show a possible configuration of this model .
the hamiltonian for the model can be written as @xmath53 where @xmath40 is the number of monomers along the @xmath15 direction and similarly for @xmath41 . then using the method previously described
, we obtain @xmath40 and @xmath41 for this model @xmath54 and @xmath55 for this model the number of configurations with energy @xmath34 is given by @xmath56 using the equation for @xmath57 we obtain the length of the polymer as @xmath58
in our simulations we followed the prescriptions of ref .
we define a monte carlo step ( mcs ) as giving sequentially to any unit the possibility of changing its direction with identical probability to any allowed direction or remaining in the same one . at the beginning of the simulation we set @xmath59 for all energy levels , where @xmath60 .
the random walk in the energy space runs through all energy levels from @xmath61 to @xmath62 with a probability @xmath63,1\right\rbrace , \ ] ] where @xmath34 and @xmath64 are the energies of the current and the new possible configurations .
after @xmath0 trial moves we update @xmath65 and @xmath66 , where @xmath67 , @xmath68 and @xmath69 ( where @xmath70 is the so - called modification factor and @xmath71 is a histogram accumulated for each @xmath72 ) . the flatness of the histogram is checked after a number of monte carlo ( mc ) steps and usually the histogram is considered flat if @xmath73 , for all energies , where @xmath74 is an average over the energies .
if the flatness condition is fulfilled we update the modification factor to a finer one and reset the histogram @xmath75 .
the simulations are continued up to @xmath76 and the microcanonical averages were accumulated from the very beginning ( @xmath77 ) , results obtained by ferrera and caparica @xcite . having in hand the density of states , one can calculate the canonical average of any thermodynamic variable as @xmath78 where @xmath79 is the microcanonical average accumulated during the simulations and @xmath80 , @xmath81 is the boltzmann constant and @xmath82 is the temperature .
in figure [ model ] we depicted the three models which were presented in section [ models ] .
the difference between the models is given by the allowed moves .
in the first model , denoted by a ) in the figure , we allow the polymer unit to move along the positive or the negative directions of the @xmath3-axis , the same for the @xmath10-axis .
the end - to - end distance of the polymer on @xmath3-direction is denoted by @xmath6 and with @xmath11 we denote this distance on @xmath10-direction . in the second model , b ) in figure [ model ] , the antiparallel motion along the @xmath3-axis is forbidden . in the third considered model there
are allowed moves only on the positive side of both @xmath3 and @xmath10-axis . using the simulated and the exact density of states in eq .
we calculate the specific heat given by @xmath83 and the mean end - to - end distance @xmath84 where @xmath34 is the energy of the configurations and @xmath85 and @xmath86 are the corresponding @xmath3-coordinates of the ends of the polymer . in figure
[ ge ] we plot in semi - logarithmical scale the density of states @xmath57 for all the three models . here
we plot the results obtained from the simulations ( symbols in the figure ) and also the exact theoretical results ( continuous lines in the figure ) for a polymer of @xmath87 monomers . in figure [ lx ]
we plot the end - to - end distance along the @xmath3-axis , @xmath6 , as a function of the temperature . here
we rescale @xmath6 by the total number of monomers , @xmath87 in this case .
we observed a very good agreement between the simulation results ( symbols in the figure ) and the theoretical results ( continuous lines in the figure ) , given by equations ( [ lxn_1 ] ) , ( [ lxn_2 ] ) , and ( [ lxn_3 ] ) , corresponding to model @xmath50 , model @xmath38 , and model @xmath88 respectively . in figure [ cv ]
we plot the specific heat per monomers for polymers with @xmath87 .
the specific heat have a tail proportional to @xmath89 in the high temperature limit . finally , in table i we present the location of peak of the specific heat for each model obtained by the wang - landau simulations and compare with the results calculated with the exact density of states .
the simulations were carried out for 100 independent runs , adopting the @xmath90 flatness criterion .
one can see that in the three cases the exact results fall into the error bars .
[ cols="^,^,^ " , ] * table i : * _ temperatures of the peak of the specific heat from simulations , compared with the exact values . _
we have carried out wang - landau simulations of a simple polymer model which has exact solutions in both the microcanonical and the canonical ensembles .
here we considered three two - dimensional models : in the first model we allowed moves in all possible directions , in the second model the moves along the negative @xmath3-axis are forbidden , while in the last model we allowed only moves along the positive direction for both @xmath3 and @xmath10- axis .
we have shown that updating the density of states only after each @xmath0 trial moves and halting the simulations when @xmath76 @xcite , defined during the simulations , we obtain quite accurate results , compared with the available analytical exact results .
we have obtained a very good agreement between the simulations and the exact results also for the studied physical quantities : the end - to - end distance and the specific heat . as expected , due to the difference of the allowed directions of motion , in the limit of high temperatures
the end - to - end distance has the highest value for the third model and the lowest for the first model .
cunha - netto a g , dickman r and caparica a a , ( 2009 ) _ comput . phys .
commun . _ * 180 * , 583 ; silva c j , cunha - netto a g , caparica a a and dickman r , ( 2006 ) _ braz .
j. phys . _
* 36 * , 3a 619 ; seaton d t , wst t and landau d p , ( 2010 ) _ phys .
e _ * 81 * , 011802 .
-axis and @xmath10-axis , ( b ) model 2 : is forbidden the backward motion on the @xmath3-axis , and ( c ) model 3 : the random walk moves only along the positive directions of @xmath3-axis and @xmath10-axis.,width=287,height=336 ] for ( a ) model 1 ; ( b ) model 2 and ( c ) model 3 .
the line is the exact result and the dots are the simulational results using the procedure for wang - landau simulation @xcite .
the error bars are less than symbols.,width=283,height=283 ] for ( a ) model 1 ; ( b ) model 2 and ( c ) model 3 .
the line is the exact result and the dots are the simulational results using the procedure for wang - landau simulation @xcite .
the error bars are less than symbols.,width=283,height=283 ] | in this work we apply wang - landau simulations to a simple model which has exact solutions both in the microcanonical and canonical formalisms .
the simulations were carried out by using an updated version of the wang - landau sampling .
we consider a homopolymer chain consisting of @xmath0 monomers units which may assume any configuration on the two - dimensional lattice . by imposing constraints to the moves of the polymers we obtain three different models .
our results show that updating the density of states only after every @xmath0 monomers moves leads to a better precision .
we obtain the specific heat and the end - to - end distance per monomer and test the precision of our simulations comparing the location of the maximum of the specific heat with the exact results for the three types of walks .
* keywords * : homopolymer , monte carlo , wang - landau |
the linear - chain structure of three @xmath1 clusters in c isotopes has been a long standing problem , for example , as discussed in refs .
historically , the @xmath6 state in @xmath7c was suggested to be a candidate for the linear - chain structure of three @xmath1 clusters @xcite . however , according to @xmath3 cluster model calculations in the 1970s , it was pointed out that the @xmath6 state is not a linear - chain state , but a weak coupling state , where three @xmath1 clusters are weakly interacting with each other @xcite . in recent years , the @xmath6 state attracts an interest as the @xmath1 condensed state in analogy to bose einstein condensation in nuclear matter @xcite . instead of the @xmath6 state ,
the @xmath8 state was suggested to have a geometric configuration of three @xmath1 clusters @xcite . in the recent works on @xmath7c with antisymmetrized molecular dynamics ( amd ) and fermionic molecular dynamics ( fmd ) calculations @xcite ,
the @xmath8 state was suggested to have an open triangle structure of three @xmath1 clusters and was assigned to the experimental @xmath9 state at 10.3 mev .
however no linear - chain structure of three @xmath1 clusters nor linear - chain band was found in the @xmath7c system , and therefore , searching for a linear - chain state in c isotopes is still an attractive subject .
linear - chain structures in c isotopes have been discussed in earlier works with 3@xmath1+@xmath10 cluster models @xcite and the skyrme mean - field model @xcite . in the study by itagaki _
et al_. @xcite , they assumed a symmetric 3@xmath1 cluster configuration with valence neutrons in molecular orbitals .
it was argued that the linear - chain structure of @xmath0c may be unstable because there is no local minimum in the energy curve against the bending mode . however , to confirm the stability of an excited state , it is essential to take into account the orthogonality to lower states .
in contrast to the work of ref .
@xcite , in our previous study on @xmath0c with an amd method @xcite , we found the linear - chain structure and its rotational band consisting of the @xmath11 , @xmath12 , and @xmath13 states in @xmath0c . in ref .
@xcite , we adopted a method of @xmath14-@xmath15 constraint amd in combination with the generator coordinate method ( gcm ) called the @xmath14-@xmath15 constraint amd + gcm @xcite .
although the 3@xmath1 core is not _ a priori _ assumed in the framework , various 3@xmath1+@xmath16 structures are obtained in the energy variation . as a result of energy variation and gcm calculations ,
the linear - chain states appear and show @xmath2be+@xmath1 correlation . here
, the @xmath2be+@xmath1 correlation means an asymmetric @xmath17+@xmath1 configuration with two valence neutrons distributing around the correlating @xmath17 .
we note that three @xmath1 clusters have one geometric configuration , and therefore @xmath2be+@xmath1 ( @xmath17+@xmath1 ) correlation does not have a weak coupling feature , but a strong coupling feature . in other words ,
this @xmath2be is not an isolated one , but a polarized configuration of the middle @xmath1 cluster and valence neutrons in the linear - chain structure . for excited states of @xmath0c ,
there exist many experimental works suggesting the @xmath2be+@xmath1 cluster structure in the observed excited states above the @xmath2be+@xmath1 threshold @xcite .
we expect that these states might be possible candidates for the linear - chain states .
then we consider that the @xmath2be+@xmath1 correlation is one of the key features of the linear - chain structure .
our aim is to confirm the @xmath2be+@xmath1 correlation in the linear - chain states of @xmath0c . as discussed in ref .
@xcite , our previous work indicates that the stability of the linear - chain structure in @xmath0c is owed to the valence neutron effects and the orthogonality to lower states .
since the bending mode is suppressed by the orthogonality in a @xmath0c system , a @xmath3 cluster core with a straight line configuration is considered to be rather stable against the bending mode .
therefore , in the present study , we consider only the linear configuration and omit the bending mode to find the origin of the @xmath2be+@xmath1 correlation . for @xmath0c ,
this ansatz is probably valid .
in such a linear system , the @xmath2be+@xmath1 correlation can be described by the asymmetry of three @xmath1 clusters and valence neutron motion . here
the first question is whether three @xmath1 clusters favor an asymmetric @xmath17+@xmath1 configuration which were not incorporated in the work by itagaki _
the second question is whether the two neutrons localize and form the @xmath2be cluster or if they are moving around all three @xmath1 clusters as described in the molecular orbital picture @xcite .
unfortunately , in our method used in the previous work @xcite , detailed behavior of valence neutrons in the molecular orbitals is not sufficiently described because a single particle wave function is approximated by a gaussian wave packet . in the present paper , to investigate the linear - chain structures of @xmath0c we extend the molecular orbital model by explicitly incorporating asymmetric configurations of the linear - chain structure .
we solve the valence neutron motion in the linear @xmath3 system by using a linear combination of @xmath18 orbits around @xmath1 clusters as done in the molecular orbital model . by analyzing the calculated results
, we will confirm the @xmath2be+@xmath1 correlation in the linear - chain structure of @xmath0c and clarify its origin . to see the effect of three @xmath1 clusters and that of valence neutron motion
, we compare the results of @xmath0c with those of @xmath3 and @xmath3+@xmath4 . here
, we comment that these 3@xmath1 and @xmath3+@xmath4 do not necessarily correspond to excited states of realistic @xmath7c and @xmath19c systems .
although the @xmath8 state of @xmath7c is suggested to have an open triangle structure , it is different from the straight - line chain structure assumed in the present analysis .
even if a linear - chain structure does not exist in realistic @xmath7c and @xmath19c , analysis of @xmath3 and @xmath3+@xmath4 systems with the linear - chain configuration is useful to understand the origin of @xmath2be+@xmath1 correlation suggested in the realistic linear - chain states in @xmath0c .
we also calculate @xmath5c and discuss systematics of correlations in the linear - chain structure of c isotopes .
this article is organized as follows . in sec .
[ framework ] , we explain the framework of the generalized molecular orbital model .
the calculated results are shown in sec .
[ results ] .
we discuss the motion of valence neutrons and correspondence to the previous work in sec . [ discussion ] .
finally , in sec .
[ summary ] , we give a summary .
in this section , we explain the framework of the generalized molecular orbital model . in this model , we express the valence neutron motion in the linear - chain structures by a linear combination of configurations written by the @xmath18 orbits centered at @xmath1 clusters . in the present model ,
the wave function of a carbon isotope with a mass number @xmath20 , @xmath21c , is described by a superposition of basis wave functions which consist of three @xmath1 clusters with a linear - chain configuration and @xmath22 valence neutrons with various configurations : @xmath23 where @xmath24 is the antisymmetrizing operator and @xmath25 represents a set of the configuration for valence neutrons , @xmath26 .
the subscripts l , m , and r mean left , middle , and right positions for three @xmath1 clusters , respectively .
we label the left , middle , and right @xmath1 clusters @xmath27 , @xmath28 , and @xmath29 .
the coefficients @xmath30 are determined by diagonalization of hamiltonian and norm matrices as explained later .
the wave function of @xmath31 is given by the @xmath32 configuration whose center is shifted at a certain position @xmath33 , @xmath34 \label{single_particle_spatial_generalizedmo},\end{aligned}\ ] ] where @xmath35 .
a linear - chain structure is expressed by the positions @xmath33 on the @xmath36-axis and it is characterized by inter - cluster distances , @xmath37 and @xmath38 : @xmath39 as seen in fig .
[ linear_chain_figure ] . without loss of generality
, we can limit the distances @xmath40 .
the distances between @xmath1 clusters l and m , and m and r are set to @xmath37 and @xmath38 , respectively.,width=298 ] for spatial wave functions of valence neutrons , we define the @xmath18 orbits centered at one of three @xmath1 clusters , @xmath41 where @xmath42 is a small value .
the @xmath43 represents the @xmath18 orbits around @xmath31 , and it has the @xmath36-component of orbital angular momentum @xmath44 . in general , the @xmath45-th valence neutron wave function @xmath46 in eq .
[ wavefunction_generalizedmo ] is given by a product of @xmath43 and a spin wave function , @xmath47 @xmath48 and @xmath49 are the eigenstates of the @xmath36-component of total angular momentum @xmath50 with @xmath51 and @xmath52 , respectively , which are energetically favored because of the spin - orbit force , while @xmath53 and @xmath54 are the unfavored @xmath55 states . in the present model ,
we omit the @xmath56 orbits because the orbits constructed from the @xmath56 orbits have higher nodes than those from @xmath57 and @xmath57 orbits ; therefore the @xmath56 orbits are unfavored . in the present calculation
, we assume the first and second valence neutrons occupy the favored orbits and the third and fourth valence neutrons occupy the unfavored ones . as a result
, the valence neutron wave functions can be written with the labels @xmath58 and @xmath59 as , @xmath60_{t_1 } , \\
\varphi^{n}_{2 } ( t_{2 } ) & = ( p_{-})_{t_2 } \chi_{n\downarrow}\equiv [ -3/2]_{t_2 } , \\
\varphi^{n}_{3 } ( t_{3 } ) & = ( p_{+})_{t_3 } \chi_{n\uparrow}\equiv [ + 1/2]_{t_3 } , \\ \varphi^{n}_{4 } ( t_{4 } ) & = ( p_{-})_{t_4 } \chi_{n\downarrow}\equiv [ -1/2]_{t_4},\end{aligned}\ ] ] and a set of the configuration for valence neutrons , @xmath25 , can be explicitly written as @xmath61 with @xmath62 .
it means that the total wave function is a superposed wave function given by the configuration mixing for valence neutrons with respect to @xmath62 indicating positions of the @xmath18 orbit centers , @xmath63 the coefficient @xmath64 is determined by the diagonalization to minimize the energy of the state , @xmath65 , for a given @xmath3 configuration parametrized by @xmath66 and @xmath67 . in this model
, we introduce an asymmetric correlation of a spatial configuration of @xmath1 clusters by the position of the middle @xmath1 cluster . when the middle @xmath1 cluster shifts to one side considerably , @xmath68 fm , a linear three @xmath1 system has the correlating @xmath17 and the isolated @xmath1 configuration .
this improvement is the most different point from symmetric molecular orbital models .
moreover , we incorporate a multineutron correlation by the configuration mixing of valence neutron orbits .
we project the wave functions onto parity and angular momentum eigenstates by using the parity projection operator @xmath69 and the angular - momentum projection operator @xmath70 .
the parity projection operator @xmath69 is defined as @xmath71 where @xmath72 is the parity operator .
the angular - momentum projection operator @xmath70 is defined as @xmath73 where @xmath74 are the euler angles , @xmath75 is the wigner s @xmath76 function , and @xmath77 is the rotation operator .
we will show energy curves of intrinsic , parity projected , and parity and angular - momentum projected wave functions . for the effective two - body interactions , we use the volkov no
. 2 interaction @xcite as the central force and the spin - orbit term of the g3rs interaction @xcite as the @xmath78 force .
we take the same interaction parameters as those in refs .
@xcite , i.e. , the majorana exchange parameter @xmath79 ( @xmath80 ) , the bartlett exchange parameter @xmath81 , and the heisenberg exchange parameter @xmath82 in the central force , and @xmath83 mev and @xmath84 mev in the @xmath78 force .
these parameters are the same as those adopted in the studies for @xmath85be @xcite , and @xmath2be @xcite , except for a small modification in the strength of the spin - orbit force to fit the @xmath86 energy of @xmath7c @xcite . for the width parameter of single - particle gaussian wave packets in eq .
, we used the value @xmath87 @xmath88 , which is also the same as those in the studies for c isotopes @xcite . to discuss the correlation between valence neutrons in a linear - chain structure , which may be incorporated by the diagonalized wave function with the configuration mixing in eq .
[ wavefunction_generalizedmo_2 ] , we also introduce one slater determinant wave function having no explicit two - body correlation in principle , and compare the results . in the framework of one slater determinant wave function , each single particle wave functions for a valence neutron is given by a linear combination of @xmath89_t$ ] orbits with respect to @xmath90 instead of the configuration mixing in eq .
[ wavefunction_generalizedmo_2 ] : @xmath91 where @xmath92_{t}.\ ] ] @xmath93 , @xmath94 , @xmath95 , and @xmath96 are chosen for the first , second , third and fourth valence neutrons as well as eq .
[ wavefunction_generalizedmo_2 ] and @xmath97 are the coefficients for the orbits @xmath89_{t}$ ] .
@xmath98 means the character of the orbit : the @xmath99 orbit , negative parity , and @xmath36 component of angular momentum @xmath100 . in the limit of @xmath101 and @xmath102 , these @xmath98 orbits coincide with the @xmath99 orbits of a usual molecular orbital model . in the present study ,
the coefficients @xmath97 are determined by a variational calculation . in a variational calculation
, we apply the following conditions for the coefficients , @xmath103 the model space spanned by this one slater determinant wave function is a subspace of the generalized molecular orbital model .
we also introduce three kinds of test wave functions with fixed coefficients which are expected from naive pictures .
we name three orbits ` mo ' , ` po ' , and ` ao ' . in ` mo ' , the coefficients are chosen as @xmath104 which indicates the molecular orbital limit where valence neutrons spread over the whole of three @xmath1 clusters . ` ao '
corresponds to the atomic orbital limit with valence neutrons localizing around the middle @xmath1 cluster : @xmath105 ` po ' means the partial orbit in which valence neutrons distribute around the correlating @xmath17 as @xmath106 this distribution describes a naively expected be correlation .
however , please note that realistic be correlation in the linear - chain state of @xmath0c is different from this po configuration as shown later .
for example , the @xmath107 orbits for ` mo ' , ` po ' , and ` ao ' are written as follows : @xmath108_{\text{l } } + [ + 3/2]_{\text{m } } + [ + 3/2]_{\text{r } } , \\ \text{`po ' } & : & \pi^{-}(+3/2 ) = [ + 3/2]_{\text{l } } + [ + 3/2]_{\text{m } } , \\ \text{`ao ' } & : & \pi^{-}(+3/2 ) = [ + 3/2]_{\text{m}}. \end{aligned}\ ] ] neutrons in ` mo ' orbits gain the kinetic energy because valence neutrons distribute in a wide region along the @xmath36-axis , while those in ` ao ' orbits gain maximum potential energy , as shown later .
we applied the generalized molecular orbital model to c isotopes : @xmath0c , @xmath7c , @xmath19c , and @xmath5c . in this section
, we show the calculated results . [
cols="^ " , ] we discuss the motion of valence neutrons and their effect in the linear @xmath3 structure by comparing the results of the generalized molecular orbital model with those calculated by using simpler wave functions , ` mo ' , ` po ' , ` ao ' , and one slater determinant defined in sec .
[ subsec : one_slater ] . in fig .
[ energy_neutron_motion_5.2fm ] , we show the angular - momentum projected energy curves of @xmath0c and @xmath19c as a function of the difference @xmath109 of the @xmath1-cluster intervals .
the top panel shows the @xmath0c(@xmath9 ) energy curves for the ` mo ' , ` po ' , ` ao ' , and one slater determinant wave functions compared with the generalized molecular orbital ones , and the bottom panel shows the @xmath19c(@xmath110 ) energy curves .
it should be noted that one slater determinant wave function is equivalent to the generalized molecular orbital wave functions in @xmath19c , because two models are coincident with each other in the case of a single valence neutron .
this is not necessarily the case in @xmath0c because the generalized molecular orbital wave functions contain two neutron correlations due to the configuration mixing , while one slater determinant does not have two neutron correlations . in spite of the difference of the model space ,
the obtained one slater determinant wave functions for @xmath0c give the same @xmath111 energy curve as that of the generalized molecular orbital model in all regions of @xmath109 .
actually , the obtained @xmath0c wave functions of these two models are found to be equivalent to each other with about 99% overlap .
this means that the two - body correlation between valence neutrons is not essential in the linear chain structure of @xmath0c but the intrinsic wave functions can be described by a simple product of single - particle wave functions as written by one slater determinant in eq .
[ eq : one_slater ] .
next , we describe the energy curves obtained by using three kinds of test wave functions , ` mo ' , ` po ' , and ` ao ' . although the energy curves are higher than those of the generalized molecular orbital model and one slater determinant , all of these wave functions show soft energy curves against the asymmetry @xmath109 similarly to the case of the generalized molecular orbital model .
the energy minima of ` po ' and ` ao ' reproduce an asymmetric configuration .
in particular , the degree of the asymmetry of ` ao ' , @xmath112 fm , is in good agreement with that of the generalized molecular orbital model , though the depth of the ` ao ' energy minimum is larger .
we discuss valence neutron wave functions in the one slater determinant wave function of @xmath0c , which almost coincides with that obtained by the generalized molecular orbital model . as mentioned before , the coefficients @xmath113 ( @xmath90 ) in eq .
[ eq : one_slater ] are optimized for each @xmath109 . at the energy minimum , @xmath114 fm ,
the ratio of the coefficients is @xmath115 the largest coefficient @xmath116 indicates that valence neutrons gather around the middle @xmath1 cluster(@xmath28 ) .
since the @xmath28 is shifted to one side , the concentration of the amplitude at the @xmath28 means that valence neutron distribution leans toward the narrow side of two of three @xmath1 clusters .
that is , the asymmetric configuration of linear three @xmath1 clusters and valence neutrons following the middle @xmath1 cluster make @xmath2be+@xmath1 correlation , which is consistent with the result in ref . @xcite . for more quantitative discussion of the asymmetry , the reduction of the positive parity component in the intrinsic wave function can be a good measure . at the @xmath114 fm corresponding to the energy minimum of linear - chain states in @xmath0c ,
the magnitude of the positive parity component in the linear 3@xmath1 system without valence neutrons is 89% , while that with two valence neutrons for the whole @xmath0c system is found to be 82% .
the reduction of the positive parity component , i.e. , the enhancement of the asymmetry by valence neutrons owes to the concentration of the neutron distribution at the middle @xmath1 cluster .
incidentally , 82% for @xmath0c at the energy minimum , @xmath114 fm , is very close to the magnitudes of the positive parity component , 81% , in the linear - chain structure of @xmath7c at the energy minimum , @xmath117 fm . in the next section ,
we show that this localization of valence neutrons can be understood by a mean field potential from three @xmath1 clusters with the linear - chain structure . in @xmath19c ,
the situation is quite similar to @xmath0c .
although the energy curves of ` mo ' , ` po ' , and ` ao ' are higher than that of the generalized molecular orbital model , they give energy minima at asymmetric configurations . as well as @xmath0c , the degree of the asymmetry of ` ao ' at the energy minimum agrees well with that of the generalized molecular orbital model . here , we also analyze the distribution of the valence neutron in @xmath19c . at the energy minimum ,
the ratio of the coefficients @xmath113 ( @xmath118 ) is @xmath119 as well as @xmath0c , the valence neutron gathers around the middle @xmath1 cluster and the @xmath85be+@xmath1 correlation is found in @xmath19c . surprisingly , the ratio of coefficients and the behaviors of energy curves in @xmath0c are almost the same as those in @xmath19c .
this is consistent with the fact that the linear @xmath0c can be described by one slater determinant wave function and valence neutrons have no explicit dineutron correlation . in this analysis
, we find that valence neutrons in @xmath0c are moving rather independently in the mean field constructed from the linear three @xmath1 clusters .
in other words , a valence neutron in @xmath0c is not affected by another valence neutron .
c using ` mo ' , ` po ' , ` ao ' , one slater determinant , and the generalized molecular orbital model with the linear - chain structure whose length is 5.2 fm.,width=298 ] finally , we show the energy curves of @xmath5c to study the systematics of correlations in linear chain structures of c isotopes . for @xmath5c , we do not calculate the @xmath9 energy curve because of the computational cost .
however , we have already seen in the results of @xmath120c that the @xmath9 energy curves are essentially the same as those of the positive - parity energy curves .
therefore , in this paper , we discuss the positive - parity energy curves of @xmath5c . in fig .
[ 16c_energy_+_5.2fm ] , we show the positive - parity energy curves of one slater determinant and the generalized molecular orbital model wave functions .
the behaviors of the energy curves are different from those of other c isotopes .
the most striking feature is that the positive - parity energy curve has no energy minimum at an asymmetric configuration .
although it is soft around @xmath121 fm , an asymmetric correlation seems to disappear in @xmath5c .
moreover , in @xmath5c , the energy curve of the generalized molecular orbital model is lower by 1 mev than that of one slater determinant wave function in the @xmath122 fm region . this means the configuration mixing in the generalized molecular orbital model gains some correlation energy due to multi - neutron correlation in the linear - chain structure of @xmath5c . comparing the wave function at the symmetric configuration @xmath121 fm between the one slater determinant model and the generalized molecular orbital model , the overlap with each other
is found to be 95% , which is still large , but smaller than the 99% overlap for the case of @xmath0c .
the ratios of the coefficients @xmath97 ( @xmath123 , @xmath90 ) in the one slater determinant wave function are @xmath124 as same as @xmath0c and @xmath19c , the valence neutrons gather around the middle of the linear @xmath1 clusters .
however , in @xmath5c , the @xmath3 core does not show the @xmath7be+@xmath1 correlation but it is symmetric at the energy minimum .
furthermore , the valence neutrons also have the symmetric configuration because they distribute around the middle @xmath1 cluster , which exists just at the center of linear - chain structure .
this result indicates a linear - chain structure in @xmath5c has a symmetric configuration .
the question to be answered in the future is whether or not the linear - chain structure is stable in excited states of @xmath5c .
firstly , the formation of three @xmath1 clusters should be checked by such a framework free from model assumptions of clusters as the amd method @xcite . secondly , because the orthogonality to low - lying states is essential for the stabilization of linear - chain states as found in @xmath0c @xcite , it is necessary to investigate the structure of @xmath5c covering low - lying states to high - lying states systematically .
in this section , we give further discussion of the behavior of valence neutrons in the linear - chain structure . as already mentioned , valence neutrons tend to distribute around the middle @xmath1 cluster .
it seems in contrast to a simple molecular orbital picture , where valence neutrons are considered to distribute widely surrounding all @xmath1 clusters .
the tendency of the concentration around the middle @xmath1 is understood by a mean - field potential formed by linear three @xmath1 clusters as follows . to see the effect of the mean - field potential for a valence neutron from linear three @xmath1 clusters we calculate the single - folding potential using the volkov no . 2 force , @xmath125 of three @xmath1 clusters with the linear - chain structure . here
@xmath126 is the density distribution of linear three @xmath1 clusters . in fig .
[ 12c_folding_potential ] , we show the contour plot of the folding potential on the @xmath127 plane .
valence neutrons feel this potential in linear three @xmath1 clusters . as clearly seen , the folding potential is deepest in the narrow side of two of three @xmath1 clusters and becomes shallow around the isolated @xmath1 cluster ; therefore , valence neutrons tend to gather around the narrow side to gain potential energy . around the narrow side ,
valence neutrons feel attractions from the two @xmath1 clusters and gain larger potential energy , while around the isolated @xmath1 cluster , valence neutrons feel an attraction only from one @xmath1 cluster and gain less potential energy . here
, we note that valence neutrons do not distribute the deepest region of the potential because of pauli blocking with @xmath1 clusters .
in fact , neutron wave functions in @xmath18 orbits , @xmath128 @xmath129 , vanish on the @xmath130 axis . in the present calculations ,
the root mean square distance @xmath131 from the @xmath130 axis for a @xmath18 orbit with the width parameter @xmath87 @xmath88 is @xmath132 roughly speaking , valence neutrons in @xmath18 orbits feel the potential around the @xmath133 surface of the linear three @xmath1 structure .
the folding potential along the @xmath134 fm line is shown in fig .
[ 12c_folding_potential_x=2.0fm ] .
the behavior of this energy curve is consistent with that seen in fig .
[ 12c_folding_potential ] .
that is , the folding potential is deepest in the narrow side of two of three @xmath1 clusters and becomes shallow around the isolated @xmath1 cluster .
the mean field constructed from linear three @xmath1 clusters induces the concentration of valence neutron wave functions around the middle @xmath1 cluster . clusters with the linear - chain structure for @xmath134 fm .
the crosses show the positions of three @xmath1 clusters , width=298 ] as shown above , the mean - field potential induces the concentration of valence neutron wave functions around the middle @xmath1 cluster , while the localization of the wave functions loses kinetic energy . as a result of the balance of potential energy gain and kinetic energy loss , valence neutron wave functions are optimized . here
, we describe features of three kinds of test wave functions , ` mo ' , ` po ' , and ` ao ' , and discuss valence neutron motion from a viewpoint of kinetic and potential energies . in ` mo ' , valence neutrons spread over the whole of three @xmath1 clusters . `
po ' describes a naively expected be correlation with valence neutrons distributing around two of the three @xmath1 clusters . `
ao ' means valence neutrons around the middle @xmath1 cluster . `
mo ' gains larger kinetic energy because of the spread of valence neutron wave functions , while it gains less potential energy because of no concentration around the middle @xmath1 cluster . on the other hand , ` ao ' gains larger potential energy and looses kinetic energy because of the localization .
the failure to reproduce the results of the generalized molecular orbital model by the ` po ' model indicates that the be correlation in linear three @xmath1 clusters is different from the naively expected be correlation which is incorporated in the ` po ' wave function . in the obtained results of the generalized molecular orbital model , both the kinetic energy gain and potential energy gain play important roles . to demonstrate this feature
, we show the @xmath9 energy curves of @xmath0c calculated by mixing two of the three ` mo ' , ` po ' , and ` ao ' configurations in comparison with the results of the generalized molecular orbital model in fig .
[ 14c_mixing_0+_5.2fm ] .
the point is that the ` mo+ao ' curve is very close to that of the generalized molecular orbital model .
this indicates that the linear - chain state in @xmath0c can be interpreted as an intermediate state of ` mo ' and ` ao ' to gain both of the kinetic energy and potential energy . on the other hand , other sets of mixing , ` mo+po ' and ` po+ao ' fail to reproduce the results of the generalized molecular orbital model and ` mo+ao ' mixing .
energy surfaces of @xmath0c with the linear - chain structure against the difference between the distances of @xmath1 clusters .
the energy surface for the mixing configurations of ` mo ' , ` po ' , and ` ao ' are illustrated as well as that of the generalized molecular orbital model.,width=298 ] next , we discuss the relation between the present results and our previous results with amd @xcite . in the present results , the minimum point of the @xmath9 energy surface ( @xmath114 )
has @xmath2be+@xmath1 correlation and 80.4 mev binding energy . in the amd results ,
the main component of the @xmath11 linear - chain state also has @xmath2be+@xmath1 correlation and 80.6 mev binding energy .
the binding energies are almost the same and the valence neutron behaviors are qualitatively similar .
moreover , the overlap between these wave functions is significant as 81% .
therefore , we can conclude that the present results correspond to the amd results .
this good correspondence between the present and amd results indicates that the previous amd calculation properly describes the valence neutron configurations in the linear - chain structure of the @xmath11 in @xmath0c . in ref .
@xcite , we expected that the linear - chain states ( rotational band consisting of the @xmath11 , @xmath12 , and @xmath13 states ) might be the candidates for the states observed recently in @xmath2be+@xmath1 breakup reactions .
the @xmath2be+@xmath1 correlation confirmed in the present study supports again this expectation .
we comment on the quantum fluctuation of the @xmath1-@xmath1 distances .
the full solution does not correspond to well defined distances but to a superposition of the wave functions which have various distances because the energy curves are very soft with respect to asymmetry and quantum fluctuation may be large . however , because of the existence of minimum at the asymmetric point , components of asymmetric configuration are expected to largely mix in the superposed wave functions .
it means the importance of be+@xmath1 correlation
. finally , we comment on the negative parity states of the linear - chain structures in @xmath0c .
because of the asymmetry of the intrinsic state from the @xmath2be+@xmath1 correlation , one can expect negative parity states as a parity partner of the positive - parity linear - chain states @xcite .
however , in the present calculations , energies of negative parity states are much higher , more than 10 mev , than the positive - parity linear - chain states ; therefore , the negative - parity linear - chain states can not appear near the positive - parity linear - chain states .
it is consistent with the previous study on @xmath0c with an amd method @xcite which shows a negative - parity linear - chain state does not appear .
we investigated the linear - chain structures of @xmath0c using a generalized molecular orbital model . in this model
, we incorporated asymmetric configurations of linear three @xmath1 clusters to study the @xmath2be+@xmath1 correlation in @xmath0c .
we also investigated be+@xmath1 correlations in linear @xmath3 and @xmath3+@xmath4 systems to see the effect of three @xmath1 clusters and that of valence neutron motion .
the linear - chain structure of @xmath5c is also studied .
it was found that the linear - chain structure of @xmath0c shows the @xmath2be+@xmath1 correlation , which is characterized by the asymmetric @xmath17+@xmath1 configuration with two valence neutrons distributing around the correlating @xmath17 .
compared with the results for the linear @xmath3 and @xmath3+@xmath4 systems , we find that a linear 3@xmath1 system prefers the asymmetric 2@xmath1+@xmath1 configuration , whose origin is the many body correlation incorporated by the parity projection .
this configuration causes an asymmetric mean field for two valence neutrons , which induces the concentration of valence neutron wave functions around the correlating 2@xmath1 .
these are the origins of the @xmath2be+@xmath1 correlation in the linear - chain structure of @xmath0c . in analysis of motion of valence neutrons
, we find that both the kinetic energy gain and potential energy gain play important roles in the linear three @xmath1 clusters with valence neutrons .
a symmetric configuration of linear three @xmath1 clusters is indicated in @xmath5c .
we also find valence neutrons in a linear - chain structure of @xmath5c have multi - neutron correlation , while those in @xmath0c have no explicit two - neutron correlation . as a future problem ,
it is important to clarify whether or not a linear - chain structure with a symmetric configuration appears in @xmath5c . for this aim
, the amd + gcm method is a promising framework for systematic study of @xmath5c from low - lying states to high - lying states .
the computational calculations of this work were performed by supercomputers in yitp and kek .
this work was supported by the yipqs program in yitp .
it was also supported by the grant - in - aid for the global coe program
the next generation of physics , spun from universality and emergence " from the ministry of education , culture , sports , science and technology ( mext ) of japan , and grant - in - aid for scientific research ( nos .
22@xmath13584 , 18540263 , and 22540275 ) from jsps .
99 h. morinaga , phys . rev .
* 101 * , 254 ( 1956 ) .
h. morinaga , phys .
* 21 * , 78 ( 1966 ) .
w. von oertzen , z. phys .
a * 357 * , 355 ( 1997 ) .
n. itagaki , s. okabe , k. ikeda , and i. tanihata , phys .
c * 64 * , 014301 ( 2001 ) .
m. milin and w. von oertzen , eur .
j. a * 14 * , 295 ( 2002 ) .
e. uegaki , s. okabe , y. abe , and h. tanaka , prog .
57 * , 1262 ( 1977 ) .
y. fukushima and m. kamimura , _ in proceedings of the international conference on nuclear structure , tokyo , 1977 _ , edited by t. marumori [ j. phys .
* 44 * , 225 ( 1978 ) ] .
y. suzuki , h. horiuchi , and k. ikeda , prog .
phys . * 47 * , 1517 ( 1972 ) .
m. kamimura , nucl . phys . *
a351 * , 456 ( 1981 ) .
a. tohsaki , h. horiuchi , p. schuck , and g. rpke , phys .
lett . * 87 * , 192501 ( 2001 ) .
y. funaki , a. tohsaki , h. horiuchi , p. schuck , and g. rpke , phys .
c * 67 * , 051306(r ) ( 2003 ) . y. funaki , a. tohsaki , h. horiuchi , p. schuck , and g. rpke , eur .
j. a * 24 * , 321 ( 2005 ) .
k. kato , h. kazama , and h. tanaka , prog .
. phys . * 77 * , 185 ( 1987 ) .
t. neff and h. feldmeier , nucl .
* a738 * , 357 ( 2004 ) .
y. kanada - enyo , prog .
. phys . * 117 * , 655 ( 2007 ) .
t. suhara and y. kanada - enyo , prog .
* 123 * , 303 ( 2010 ) .
w. von oertzen , h. g. bohlen , m. milin , tz .
kokalova , s. thummerer , a. tumino , r. kalpakpakchieva , t. n. massey , y. eisermann , g. graw , t. faestermann , r. hertenberger , and h .- f .
wirth , eur .
phys . j. a * 21 * , 193 ( 2004 ) .
j. a. maruhn , n. loebl , n. itagaki , m. kimura , nucl . phys . * a833 * 1 ( 2010 ) .
t. suhara and y. kanada - enyo , phys .
c * 82 * , 044301 ( 2010 ) .
m. milin , s. cherubini , t.davinson , a. di pietro , p. figuera , d. miljanic , a. musumarra , a. ninane , a. n. ostrowski , m. g. pellegriti , a. c. shotter , n. soi , c. spitaleri , m. zadro , nucl . phys . *
a730 * , 285 ( 2004 ) .
n. soi , m. freer , l. donadille , n. m. clarke , p. j. leask , w. n. catford , k. l. jones , d. mahboub , b. r. fulton , b. j. greenhalgh , and d. l. watson , nucl .
a738 * , 347 ( 2004 ) .
d. l. price , m. freer , n. i. ashwood , n. m. clarke , n. curtis , l. giot , v. lima , p. mc ewan , b. novatski , n. a. orr , s. sakuta , j. a. scarpaci , d. stepanov , and v. ziman , phys .
c * 75 * , 014305 ( 2007 ) .
p. j. haigh , n. i. ashwood , t. bloxham , n. curtis , m. freer , p. mcewan , d. price , v. ziman , h. g. bohlen , tz .
kokalova , ch .
schulz , r. torabi , w. von oertzen , c. wheldon , w. catford , c. harlin , r. kalpakchieva , and t. n. massey , phys .
c * 78 * , 014319 ( 2008 ) .
a. volkov , nucl . phys .
* 74 * , 33 ( 1965 ) .
n. yamaguchi , t. kasahara , s. nagata , and y. akaishi , prog .
. phys . * 62 * , 1018 ( 1979 ) .
s. okabe and y. abe , prog .
theor . phys .
* 61 * , 1049 ( 1979 ) . n. itagaki and s. okabe , phys .
c * 61 * , 044306 ( 2000 ) .
n. itagaki , t. otsuka , k. ikeda , and s. okabe , phys .
lett . * 92 * , 142501 ( 2004 ) .
y. kanada - enyo and h. horiuchi , prog .
* 93 * , 115 ( 1995 ) .
y. kanada - enyo , h. horiuchi and a. ono , phys .
c * 52 * , 628 ( 1995 ) ; y. kanada - enyo and h. horiuchi , phys .
c * 52 * , 647 ( 1995 ) .
y. kanada - enyo and h. horiuchi , prog .
suppl . * 142 * , 205 ( 2001 ) .
y. kanada - enyo , m. kimura , and h. horiuchi , c. r. physique * 4 * , 497 ( 2003 ) . | we investigate the linear - chain structures in highly excited states of @xmath0c using a generalized molecular orbital model , by which we incorporate an asymmetric configuration of three @xmath1 clusters in the linear - chain states . by applying this model to the @xmath0c system
, we study the @xmath2be+@xmath1 correlation in the linear - chain state of @xmath0c . to clarify the origin of the @xmath2be+@xmath1 correlation in the @xmath0c linear - chain state , we analyze linear @xmath3 and @xmath3+@xmath4 systems in a similar way .
we find that a linear 3@xmath1 system prefers the asymmetric 2@xmath1+@xmath1 configuration , whose origin is the many body correlation incorporated by the parity projection .
this configuration causes an asymmetric mean field for two valence neutrons , which induces the concentration of valence neutron wave functions around the correlating 2@xmath1 .
a linear - chain structure of @xmath5c is also discussed . |
the first measurement of @xmath7 was reported two years ago by the sample collaboration@xcite , and led to @xmath8 a linear combination of strange electric and magnetic form factors has been measured by the happex collaboration@xcite : @xmath9(0.48{\rm gev}^2 ) = 0.023 \pm 0.034 \pm 0.022 \pm 0.026~.\ ] ] further efforts are underway by various groups .. ] it is important to recall that @xmath10 and @xmath11 contain more than just strangeness effects . even in a world of only two flavours ( up and down ) @xmath4 would be nonzero due to isospin violation .
thus , the true effects of strange quarks can only be extracted from an experimental determination of @xmath4 if isospin violating effects can be calculated .
dmitrainovi and pollock@xcite , and also miller@xcite , have studied the isospin violating contribution to @xmath11 within the nonrelativistic constituent quark model .
ma has used a light - cone meson - baryon fluctuation model.@xcite more recently , a model - independent study of isospin violating effects ( using heavy baryon chiral perturbation theory ) has been published@xcite and it is this work which will be emphasized below , after a brief review of attempts to calculate the authentic strange quark effects .
many attempts have been made to calculate the contribution of strange quarks to the `` strange '' electric and magnetic form factors , @xmath4 . in principle
a lattice qcd calculation could give the definitive answer , and an exploratory calculation has been performed in the quenched theory.@xcite the errors due to finite lattice spacing , finite lattice volume and quenching are not yet known , but the existing results , @xmath12 and @xmath13 , still provide important inputs to the discussion .
one might consider using chiral perturbation theory to calculate the strangeness contributions to @xmath4 , but both form factors have a free parameter at their first nonzero order in the chiral expansion , so the magnitude of neither form factor can be predicted from chiral symmetry alone .
however , two experimental inputs are sufficient to fix these parameters , and chiral symmetry does determine the @xmath14-dependence of the form factors at leading chiral order .
this tact has been taken by the authors of ref .
@xcite , who use the sample and happex measurements as input . beyond lattice qcd and chiral perturbation theory , there are many models and dispersion relation methods which have been employed in the effort to determine the strange quark contributions to @xmath4 .
( the authors of refs . @xcite have collected some predictions from the literature . )
the various methods lead to differing results . for @xmath15 ,
most predictions lie in the range @xmath16 and it has often been noted that this tendency toward a negative number does not seem to be supported by the experimental data , refs .
predictions for the magnitude and sign of @xmath17 also span a large range . a precise experimental measurement would help to distinguish between the various models of strangeness physics , but only after the isospin violating contribution has been calculated and subtracted .
in a world with no strange quark , @xmath10 and @xmath11 do not vanish . instead , @xmath18 where @xmath19 and @xmath20 are isospin violating quantities .
if both the strange and isospin violating components of @xmath4 are small , then contributions which are both isospin violating _ and _ strange are doubly suppressed .
the following discussion considers @xmath21 in a strange - free world .
constituent quark model calculations have led to a vanishing result for @xmath22 and a very mild @xmath14 dependence : @xmath23@xcite .
there is no symmetry which would force @xmath22 to vanish exactly , but perhaps the constituent quark model is trying to anticipate a `` small '' result .
a light - cone meson - baryon fluctuation model permits a large range , @xmath24.@xcite heavy baryon chiral perturbation theory ( hbchpt ) is a natural tool for the study of @xmath21 . it is a model - independent approach which employs a systematic expansion in small momenta ( @xmath25 ) , small pion masses ( @xmath26 ) , small qed coupling ( @xmath27 ) , large chiral scale ( @xmath28 ) and large nucleon masses ( @xmath29 ) .
it is appropriate to use @xmath30 with @xmath31 , and then the hbchpt lagrangian can be ordered as a single expansion , @xmath32 for the explicit form of the lagrangian , see ref . @xcite and references therein . for the present discussion , it is simply noted that @xmath33 contains parameters @xmath34 , @xmath35 and @xmath27 ; @xmath36 contains 11 parameters ( 7 strong and 4 electromagnetic ) ; @xmath37 contains 43 parameters ; @xmath38 contains hundreds of parameters and @xmath39 has even more .
happily , it will be shown that @xmath19 is parameter - free at its first nonzero order , and @xmath20 is parameter - free at its first and second nonzero orders except for a single additive constant .
the coupling of a vector current ( e.g. @xmath0 ) to a nucleon begins at first order in hbchpt , @xmath33 , but is isospin conserving . to be precise ,
recall the usual notation , @xmath40u(\vec{p})~,\ ] ] where @xmath41 denotes a particular flavour of quark .
the sachs form factors for that flavour are @xmath42 an explicit calculation using @xmath43 leads to isospin violating vector form factors which vanish exactly . at first glance
this might seem surprising , but it can be readily understood as follows .
an isospin violating factor , such as @xmath44 , is suppressed by two hbchpt orders . moreover , the @xmath45 term in eq .
( [ deffs ] ) has an extra explicit @xmath46 suppression factor , so isospin violating @xmath45 terms can not appear before @xmath38 .
meanwhile , @xmath47 is constrained by noether s theorem ( qcd s flavour symmetries : upness and downness ) to be unity plus momentum - dependent corrections , and dimensional analysis therefore requires a large scale , @xmath29 or @xmath28 , in the denominator of all corrections .
this demonstrates that both @xmath19 and @xmath20 vanish in hbchpt until the fourth order lagrangian : @xmath38 . a leading order ( lo ) calculation of @xmath19 or @xmath20 involves tree - level terms from @xmath38 plus one - loop diagrams built from @xmath48 .
referring to ref .
@xcite for details of the calculation and renormalization , the results are @xmath49 , \\
\left.g_m^{u , d}(q^2)\right|_{\rm lo } & = & { \rm constant } - \frac{16g_a^2m_n}{(4\pi{f})^2}(m_n - m_p ) \int_0 ^ 1{\rm d}x\,{\rm ln}\left(1-x(1-x)\frac{q^2}{m_{\pi^+}^2 } \right).\end{aligned}\ ] ] notice that the electric form factor contains no unknown parameters , and the magnetic form factor has only a single parameter ( an additive constant ) .
the lo results for @xmath19 and @xmath50 are plotted in fig .
1 . the contribution of isospin violation to @xmath17 is @xmath51 .
consider next - to - leading order ( nlo ) . here
, one expects tree - level terms from @xmath39 plus one- and two - loop diagrams built from lower orders in the lagrangian .
since small hbchpt expansion parameters without uncontracted lorentz indices come in pairs ( e.g. @xmath14 , @xmath52 , @xmath53 ) , the @xmath39 counterterms can contribute to @xmath47 but not to @xmath45 .
thus @xmath20 is independent of these parameters at nlo , although @xmath19 is not .
it is also found that no two - loop diagrams contribute to @xmath20 at nlo , although in principle they could have .
furthermore , unknown coefficients from @xmath37 are also permitted to appear within loops , but none of them actually contribute .
this means that the nlo corrections to @xmath20 are basic one - loop diagrams .
the explicit result is given in ref .
it needs to be stressed that the nlo contribution is parameter - free ; the only new quantities ( with respect to lo ) are the well - known nucleon magnetic moments . the lo+nlo result for @xmath50 is shown in fig .
2 . notice that the nlo corrections serve to soften the @xmath14-dependence .
the nlo correction to @xmath22 is @xmath54 the value of @xmath22 itself is not determined by chiral symmetry alone , and it receives contributions from physics other than the `` pion cloud '' of hbchpt ( consider , for example , isospin violation due to vector mesons ) .
the pion cloud contribution to @xmath22 is estimated in ref .
@xcite via a physically - motivated cutoff in hbchpt , and is comparable in size to the nlo contribution of eq .
( [ nlo ] ) .
the full result for the pion cloud contribution to @xmath20 is shown in fig .
2 with error bands to reflect truncation of the hbchpt expansion : the narrow band assumes @xmath55 and the wide band assumes @xmath56 .
the strange vector form factors of the proton are basic to an understanding of proton structure .
the contribution due to strange quarks has proven to be a theoretical challenge .
isospin violation also contributes to the so - called `` strangeness '' form factors , and this contribution must be calculated and subtracted from experimental data before the strange quark contribution can be identified .
the present work indicates that chiral symmetry is of great value for discussions of the isospin violating effects . despite the large number of parameters in the lagrangian , @xmath19 is parameter - free at leading order , and @xmath20 has only one ( @xmath14-independent ) parameter at leading order , and no parameters at next - to - leading order .
the isospin violating effects computed here are large compared to some models of the strange quark effects , but small compared to other models .
the experimental results for the full `` strangeness '' form factors in eqs .
( [ sameq ] ) and ( [ hapeq ] ) are not precise enough to indicate their size relative to the isospin violating contributions found in this work
. it will be interesting to see what future experiments reveal .
99 b. a. mueller _ et al .
lett . * 78 * , 3824 ( 1997 ) .
k. a. aniol _
82 * , 1096 ( 1999 ) .
d. t. spayde _
et al . _ , nucl - ex/9909019 ( 1999 ) .
v. dmitrainovi and s. j. pollock , phys .
c * 52 * , 1061 ( 1995 ) .
g. a. miller , phys .
c * 57 * , 1492 ( 1998 ) .
ma , phys .
b * 408 * , 387 ( 1997 ) .
r. lewis and n. mobed , phys .
d * 59 * , 073002 ( 1999 ) . s. j. dong , k. f. liu and a. g. williams , phys .
d * 58 * , 074504 ( 1998 ) .
t. r. hemmert , u .- g .
meissner and s. steininger , phys .
b * 437 * , 184 ( 1998 ) ; t. r. hemmert , b. kubis , u .- g .
meissner , phys .
c * 60 * , 045501 ( 1999 ) .
m. musolf _
et al _ , phys .
rep . * 239 * , 1 ( 1994 ) ; e. j. beise _
et al . _ , nucl - ex/9602001 ( 1996 ) ; l. l. barz _
et al . _ ,
a * 640 * , 259 ( 1998 ) . | the strange form factors of the proton are basic to an understanding of proton structure , and are presently the focus of many experiments .
before the strangeness effects can be extracted from data , it is necessary to calculate and remove effects due to isospin violation , which exist independently of the strange quark but which contribute nevertheless to the experimentally measured `` strange '' form factors .
a discussion of the isospin violating contributions to vector form factors is given here in the context of heavy baryon chiral perturbation theory . |
recently , dienes , dudas and gherghetta ( ddg ) @xcite have suggested the intriguing possibility that grand unification may occur at intermediate or even low energy scales in models with extra spacetime dimensions compactified on orbifolds of radii @xmath0 . above the compactification scale @xmath1 ,
the vacuum polarization tensors for the gauge fields of the minimal supersymmetric standard model ( mssm ) receive finite corrections from a tower of kaluza - klein ( kk ) excitations which contribute at one loop . as a consequence ,
the gauge couplings develop a power - law rather than logarithmic dependence on the ultraviolet cut off of the theory , @xmath2 @xcite .
while this is not running in the conventional sense , the gauge couplings nonetheless evolve rapidly as a function of @xmath2 , so that it is possible to achieve an accelerated unification . in the minimal scenario proposed by ddg , all the non - chiral mssm fields , the two higgs doublets and the gauge multiplets ,
live in a @xmath3 dimensional spacetime , and have an associated tower of kk excitations .
the chiral mssm fields are assumed to lie at fixed points of the orbifolds , and thus have no kk towers .
this is the simplest way of avoiding the difficulties associated with giving mass to chiral kk states .
ddg demonstrated that an approximate gauge unification could be achieved in this scenario , at a grand unification scale @xmath4 that is much smaller than its usual value in supersymmetric theories , @xmath5 gev .
however , as pointed out in ref .
@xcite , strict comparison to the low energy data reveals that for tev - scale compactifications , the ddg model predicts a value for @xmath6 that is higher than the prediction in conventional unified theories , which is already @xmath75 standard deviations higher than the experimental value in conventional supersymmetric unified theories , compared to the world average , @xmath8 @xcite . ] . assuming that the unification point coincides with the string scale , then a specific model of string scale threshold corrections is required before one can claim that unification in the ddg model
is actually achieved .
it is the purpose of this paper to point out that there are a number of simple variations on the ddg proposal that achieve gauge unification much more precisely than the minimal scenario described above .
to begin , however , let us consider the variations that are as successful as the minimal case . as pointed out by ddg ,
one possibility is to allow @xmath9 generations of matter fields to experience extra dimensions , and to add to the theory their chiral conjugate mirror fields , so that suitable kk mass terms may be formed .
assuming that the orbifold is @xmath10 , then one may take the mirror fields to be @xmath11 odd , so that unwanted zero modes are not present in the low - energy theory .
given that the kk excitations of the matter fields form complete su(5 ) multiplets , it is perhaps not surprising that if unification is achieved for @xmath12 , it will also be preserved for @xmath13 , at least for some range of @xmath14 . what is less obvious is that the kk excitations of the matter fields may be chosen to form _ incomplete _ su(5 )
multiplets , and an approximate unification may still be preserved if only _
some _ of the mssm gauge and higgs fields experience extra dimensions . in section 2
, we will present three models that demonstrate ( i ) that it is not necessary for all gauge groups to live in the higher dimensional bulk in order to achieve unification , and ( ii ) a precise unification may sometimes be obtained without introducing any exotic matter , beyond the mirror fields described above . in section 3
we study these cases quantitatively , taking into account weak - scale threshold corrections , and two - loop running up to the compactification scale .
for tev scale compactifications @xcite , we will see that all of the nonminimal scenarios we present in this paper unify more precisely than the minimal ddg scenario , and do not require large threshold corrections at the unification scale .
in section 4 we summarize our conclusions , and make some brief comments on the phenomenological implications of these models when the compactification scale is low .
we assume @xmath3 spacetime dimensions , with @xmath15 dimensions each compactified on a @xmath11 orbifold of radius @xmath16 .
the fields that experience extra dimensions are periodic in the @xmath15 new spacetime coordinates @xmath17 , and are either even or odd under @xmath18 .
for example , in the case where @xmath19 , these ` bulk ' fields have expansions of the form @xmath20 @xmath21 where @xmath22 indicates the kk mode .
the only other fields in the theory are those which live at the orbifold fixed points @xmath23 or @xmath24 , and have no kk excitations .
the effect of a tower of kk states on the running of the mssm gauge couplings was computed by ddg , and is given in a useful approximate form by @xcite @xmath25 \,\,\ , .
\label{eq : drun}\end{aligned}\ ] ] here the @xmath26 are the beta function contributions of a single kk level , and @xmath27 is given by @xmath28 for the scenarios considered by ddg , these beta functions are @xmath29 where @xmath9 is the number of generations of matter fields that experience extra dimensions .
ddg observe that a sufficient condition for gauge unification to be preserved is that the ratios @xmath30 be independent of @xmath31 and @xmath32 .
thus , they point out that in the scenario above @xmath33 we will now show that there are a variety of other models , each with a different set of mssm fields living at the orbifold fixed points , that lead to @xmath34 . first ,
notice that the @xmath26 of the minimal scenario can be decomposed into the contributions from the kk excitations of each mssm field , as shown in table 1 .
an overline denotes a mirror field required so that the given kk tower is vector - like .
.contributions to @xmath26 .
[ cols= " < , < , < " , ] c|ccccscenario & @xmath19 & @xmath35 & @xmath19 & @xmath35 + & @xmath36 tev & @xmath36 tev & @xmath37 gev & @xmath37 gev + + minimal & 0.1473 & 0.1509 & 0.1320 & 0.1344 + 1 & 0.1263 & 0.1283 & 0.1206 & 0.1215 + 2 & 0.1047 & 0.1049 & 0.1076 & 0.1077 + 3 & 0.1044 & 0.1044 & 0.1055 & 0.1056 + + minimal & 0.1947 & 0.2006 & 0.1678 & 0.1717 + 1 & 0.1566 & 0.1606 & 0.1495 & 0.1509 + 2 & 0.1255 & 0.1258 & 0.1297 & 0.1297 + 3 & 0.1243 & 0.1246 & 0.1258 & 0.1258 + @xmath38 minimal scenario : this is the @xmath12 scenario of ddg , which we include as a point of reference .
the beta functions for this scenario are given in eq .
( [ eq : bmin ] ) .
note that for @xmath19 and @xmath39 tev the low energy value of @xmath6 is @xmath730 standard deviations above the experimental central value , @xmath40 @xcite , and improves to @xmath716 standard deviations in the 1996 review of particle physics , we would obtain results too high by about @xmath41 standard deviations .
the experimental determination of @xmath42 has since improved . ]
if one assumes that colored mssm superpartners at the weak threshold are all at @xmath43 while noncolored sparticles are at @xmath71 tev .
these results agree qualitatively with those in ref .
@xcite , where a different approximation for weak scale threshold effects was used .
@xmath38 scenario 1 : in this scenario , the gauge fields and leptons live in the bulk , while the higgs and quarks live at orbifold fixed points
. the beta functions for this scenario were given in eq .
( [ eq : bsc1 ] ) .
notice that our previous observation that this scenario satisfies the relation @xmath44 more accurately than the minimal case does translate into a better predictions for @xmath6 .
for @xmath19 and @xmath39 tev , and assuming the same choice for weak scale threshold corrections applied to the minimal scenario above , we find agreement with the experimental value of @xmath6 at the @xmath45 standard deviation level .
@xmath38 scenario 2 : in this scenario , the su(2 ) gauge multiplet and higgs fields are confined to the fixed point , while precisely one generation of right - handed up and down quarks , and three generations of right - handed leptons live in the bulk .
the kk beta function contributions are given by @xmath46 this scenario achieves unification much more precisely than the minimal one . in the case where @xmath39 tev and @xmath19 , the predicted value of @xmath6 is only 1.8 standard deviations _ below _ the experimental central value , ignoring all threshold corrections . allowing the superparticle spectrum to vary as in table 4 , we find this prediction varies between @xmath47 thus
, unification can be achieved in this model without any string scale threshold corrections .
@xmath38 scenario 3 : in this example , the su(3 ) gauge multiplet is confined to the orbifold fixed point , and hence we are only allowed to place leptons and higgs fields in the bulk . if we let @xmath48 and @xmath49 represent the number of right - handed and left - handed kk lepton excitations ( including their chiral conjugate partners ) , then the constraint @xmath50 implies that @xmath51 , which has no solution for only three generations of left - handed lepton fields .
however , there is a simple way to circumvent this problem .
notice that if we add su(5 ) * 5*+@xmath52 pairs in which only the lepton compontents live in the bulk , then the condition on @xmath48 and @xmath49 given above will hold , since differences in zero mode beta function pairs will remain unchanged . a solution may then be obtained by choosing @xmath53 and @xmath54 , which implies the existence of two such * 5*+@xmath52 pairs , when one generation of right - handed and three generations of left - handed mssm lepton fields are assigned to the bulk .
the beta functions for this scenario are then given by @xmath55 notice in this case that the su(3 ) gauge coupling only evolves logarithmically , since @xmath56 .
unification may be achieved at a low scale by virtue of the power - law evolution of @xmath57 and @xmath58 , as can be seen from fig . 1 .
this scenario is about as successful as scenario 2 , predicting @xmath6 only @xmath59 standard deviations below the experimental central value , ignoring all threshold corrections , and assuming that the exotic matter fields have masses of @xmath43 . allowing the sparticle mass spectrum to vary between @xmath43 and 1 tev , as in table 4
, we find that the scenario 3 prediction for @xmath6 varies between @xmath60 for @xmath19 and @xmath39 tev .
again , unification is achieved without the need for threshold corrections at the high scale .
although this scenario does indeed involve some new matter fields , the choice is relatively minimal , and may be completely natural from the point of view of string theory .
what is interesting about the scenarios we ve presented is that gauge unification can be achieved in so many different ways .
each of our scenarios unifies more precisely than the minimal ddg model , and none requires large ( or in two cases any ) threshold corrections at the unification scale .
these models illustrate two other interesting points as well : ( 1 ) one can achieve unification when some of the standard model gauge groups are confined to a brane .
( 2 ) there are some models that unify more precisely than ddg that _ do not _ require any additional matter fields with exotic quantum numbers , beyond the vector - like kk towers of certain mssm fields that are chosen to live in the bulk . before concluding , we comment briefly on some of the other phenomenological implications of these scenarios when the compactification scale is low .
for a more complete discussion of the phenomenology of standard model kk excitations in models with tev scale compactification @xcite , we refer the reader to refs .
@xcite . in scenarios 1 and
2 the gluon has a tower of kk excitations , which leads to a significant bound on the compactification scale .
the kk gluon excitations are massive color octet vector mesons , with couplings both to zero mode gluons and to all the quarks .
thus , the kk gluons are in every way identical to flavor universal colorons , and are subject to the same bounds . recall that in coloron models , one obtains a massive color octet from the spontaneous breaking of su(3)@xmath61su(3 ) down to the diagonal color su(3 ) . in the case where the two su(3 ) gauge couplings are equal , the coloron couples to quarks exactly like a gluon , or a kk gluon .
the couplings of colorons or kk gluons to zero mode gluons are completely determined by su(3 ) gauge invariance , and hence are also the same .
thus the relevant bound on the lowest kk gluon excitation is given by @xmath62 gev at the 95% confidence level @xcite , which follows from consideration of the dijet spectrum at the tevatron .
this constraint places a lower bound on the scale for all the kk excitations in scenarios 1 and 2 .
we can obtain a similar bound on the compactification scale in scenario 3 from the production and hadronic decay of @xmath63 boson kk excitations , which have standard model couplings to the quarks : @xmath64 gev @xcite .
other direct collider bounds on the compactification scale require a more detailed analysis , given the nonstandard @xmath65 and @xmath66 couplings in our models .
this issue will be considered elsewhere @xcite .
scenario 1 is particularly interesting when one takes into account that interaction vertices involving fields that all live in the higher dimensional bulk respect a conservation of kk number .
( one can think of this as arising from the conservation of kk momentum following from translational invariance in the extra dimensions . ) hence , in scenario 1 , the kk excitations of the electroweak gauge fields can not couple to the lepton zero modes , and we obtain both @xmath66 and @xmath65 bosons with otherwise standard couplings , that are naturally leptophobic !
these states would likely be within the reach of the lhc for tev scale compactifications .
the other two scenarios present a more complicated phenomenology , since some generations of a given mssm matter field live in the bulk while others live on the brane .
it follows that the kk excitations of a standard model gauge field would have generation - dependent couplings to the zero mode matter fields , and may contribute to a variety of quark and lepton flavor - changing processes .
finally it is worth pointing out that in scenario 3 , the fact that the kk @xmath63 boson excitations ca nt couple to zero mode left - handed lepton fields , also leads to a leptophobic @xmath65 .
while the purpose of the present work was to focus on gauge unification in these nonminimal scenarios , a more quantitative discussion of the tev scale phenomenology of the scenarios described here will be presented in a separate publication @xcite .
we are grateful to keith dienes for his comments on the manuscript , and thank marc sher and carl carlson for useful discussions .
we thank the national science foundation for support under grant phy-9800741 . | we consider gauge unification in nonminimal models with extra spacetime dimensions above the tev scale .
we study the possibility that only a a subset of the supersymmetric standard model gauge and higgs fields live in the higher dimensional ` bulk ' . in two of the models we present ,
a choice for bulk mssm matter fields can be found that preserves approximate gauge unification .
this is true without the addition of any exotic matter multiplets , beyond the chiral conjugate mirror fields required to make the kaluza - klein excitations of the matter fields vector - like . in a third model , only a small sector of additional matter fields
is introduced . in each of these examples
we show that gauge unification can be obtained without the necessity of large string scale threshold corrections .
we comment briefly on the phenomenology of these models in the case where the compactification scale is as low as possible .
epsf wm-99 - 103 + .5 in * gauge unification in nonminimal models with extra dimensions * 0.3 in christopher d. carone 0.1 in _ nuclear and particle theory group + department of physics + college of william and mary + williamsburg , va 23187 - 8795 _ .5 in |
the study of atmospheric neutrinos of sub - gev and multi - gev energies has been of paramount importance in the recent past , providing in particular the first clear evidence in favor of neutrino oscillations , and hence of non - vanishing neutrino masses .
these neutrinos are produced mainly in the decay of pions and kaons ( e.g. @xmath2 , with @xmath3 ) produced by the cosmic rays ( crs ) hitting the upper atmosphere and generating air showers with a significant hadronic component . due to the interplay between the mesons decay and their interactions in the air ,
as the energy increases ( with the associated relativistic dilation of their decay times ) the mesons lose an increasing amount of energy before they decay , and also the muons produced can reach the ground before decaying . as a consequence of this ,
the predicted atmospheric neutrino spectrum results steeper than the cr spectrum typically by an extra power of the energy , with d@xmath4 , while d@xmath5 @xcite . comparing the charged mesons decay length , @xmath6 , where in particular @xmath7 km(@xmath8 gev ) and @xmath9 km(@xmath10 ) , it is clear that above 100 gev the pions traverse several attenuation lengths ( corresponding to approximately 120 g/@xmath11 ) in the atmosphere before decaying , and due to this the neutrino fluxes above a few hundred gevs actually arise mainly from @xmath12 decays . kaons in turn are also significantly attenuated before decaying for energies above the tev , with the effect that the prompt neutrinos from charmed particle decays become increasingly important . several groups have studied the charm particle production by crs @xcite , which actually requires to take into account next to leading order ( nlo ) processes , which increase the charm production by more than a factor of two , and the cross sections are sensitive to the values of the partonic distribution functions at very small values of the scaling variable @xmath13 , what introduces further uncertainties in the results , due to the need to extrapolate beyond the measured range . even if the charmed mesons ( @xmath14 ) are produced at a rate @xmath15 with respect to the non - charmed ones , their very short decay time , with @xmath16 km(@xmath17 pev ) , implies that the prompt neutrino spectral slope just follows the cr slope , and hence the prompt neutrinos eventually dominate the atmospheric neutrino fluxes above a pev in the vertical direction , and somewhat above that energy at large zenith angles .
it is important at this point to take into account the fact that the primary cr spectrum shows a steepening at the so - called _ knee _ , corresponding to @xmath18 ev , with d@xmath19 for @xmath20 .
the associated steepening in the induced neutrino fluxes has been obtained in the literature @xcite , but these works usually assume for simplicity that the dominant cr component consists of protons , and only some crude attempts have been done trying to generalise these predictions to heavier compositions @xcite .
another contribution to the diffuse neutrino fluxes produced by crs are those resulting from the interactions of the crs present all over the galaxy with the ambient gas in the inter stellar medium ( ism ) @xcite .
the very low densities in the ism ( @xmath21 ) imply that essentially all mesons produced decay in this case without suffering any attenuation , and hence the neutrino fluxes are just the conventional ones from @xmath22 , @xmath12 and @xmath23 decays , and moreover their spectral slopes follow the behaviour of the primary cr spectrum .
these neutrino fluxes are proportional to the column density of the ism along the particular direction considered , and hence the flux is maximal in the direction to the galactic center , it is generally enhanced near the galactic plane while it is minimum in the direction orthogonal to it .
along directions near the plane of the milky way , these diffuse fluxes of galactic neutrinos may overcome the atmospheric neutrino background at energies larger than @xmath24 ev ( the actual energy depends on the model adopted for charm production and on the assumed ism column density , see below ) , but in the direction orthogonal to the plane they remain below the expected flux from prompt atmospheric neutrinos from charmed particle decays up to the highest energies @xcite . here
again the usual simplifying working hypothesis is that the cr flux consists mainly of protons , and only crude attempts have been done trying to generalise the predictions to heavier compositions @xcite .
although some works have found a strong sensitivity to the assumed composition both for atmospheric and galactic cr induced neutrinos , this issue is generally not treated appropriately and hence a more thorough analysis is required in order to estimate the neutrino fluxes with some confidence for @xmath1 ev .
the very high energy neutrino fluxes produced by crs both in the atmosphere and in the galaxy are one of the important targets of present and future neutrino telescopes , such as amanda , baikal , antares , icecube , etc .. they are interesting per se , but they also represent the main background in the search of the fluxes arising from other potential neutrino sources , such as active galactic nuclei ( agns ) @xcite and gamma ray bursts ( grbs ) @xcite , which might give rise to detectable fluxes just in the energy range @xmath25@xmath26 ev @xcite .
it is hence of primary importance to establish with some confidence the actual value of the cr induced neutrino fluxes , and the aim of the present work is to point out that the predictions are quite sensitive to the detailed composition and to the behaviour of the spectrum of the individual nuclear components of the crs with energies above those corresponding to the knee in the spectrum .
a major problem that one has to face here is that although the existence of the knee has been known for more than forty years , there is still no consensus on the underlying physics responsible for this feature , and different proposed explanations for it can lead to quite different predictions for the behaviour of the cr composition and the individual spectra . on the observational side , due to the indirect nature of the measurements , which are based on the analysis of extensive air showers ,
also the situation is far from settled . a large class of possible scenarios to explain the knee
are based on a rigidity dependent effect , such as a change on the cr acceleration efficiency at the sources @xcite or on a change in the escape mechanism of crs from the galaxy @xcite , both of which , being magnetic effects , depend on the ratio @xmath27 , where @xmath28 is the cr charge .
hence , in these scenarios the different nuclear components essentially steepen their spectra at an energy @xmath29 , and in this way the light components become more suppressed at smaller energies than the heavy ones .
this leads to the prediction that the cr composition should become increasingly heavier above @xmath30 , in agreement with the latest observations of kascade @xcite and the eas - top / macro experiments @xcite .
in addition , the change in the spectral slope of each individual component usually turns out to be steeper than the change in slope of the total observed spectrum , which is @xmath31 .
in particular , in the so - called diffusion / drift model @xcite in which the knee is due to a change in the escape mechanism of crs from the galaxy from one dominated by normal diffusion ( with diffusion coefficient @xmath32 ) to one dominated by drift effects ( with @xmath33 ) , the spectral slope of each individual nuclear component changes by an amount @xmath34 . as we will show below , these rigidity dependent scenarios have a profound impact on the predictions of high energy diffuse neutrino fluxes , and hence a clear understanding of the physics responsible for the knee and also better measurements of air showers at these energies are crucial to predict the expected neutrino fluxes and to extract conclusions from them once they will be measured .
figure 1 shows the observed cosmic ray differential spectrum from several experiments @xcite . as mentioned above , the total cr spectrum is well described by power laws d@xmath35 , with the spectral index changing from @xmath36 to @xmath37 at the so - called knee , occurring at @xmath18 ev .
the cr spectrum turns harder again at the ankle , corresponding to @xmath38 ev , a feature usually explained as the crossover between the dominance of the galactic cr component below the ankle and of the extragalactic one above it .
a simple way of describing the galactic cr flux within a rigidity dependent scenario is to assume that each galactic component of charge @xmath28 is given by @xmath39 where @xmath40 ( @xmath41 ) is the cr flux below ( above ) the knee , and is given by @xmath42 .cosmic ray fractional abundances ( for @xmath43 tev ) and low energy spectral indices from hydrogen to nickel .
[ cols= " > , > , > , > , > , > , > , > , > , > , > , > " , ] in this expression , @xmath44 is the total cr flux at the reference energy @xmath45 , hereafter adopted as @xmath46 tev , @xmath47 is the fractional cr abundance at the same energy , @xmath48 is the low energy spectral index , @xmath49 is the spectral index change across the knee ( which is the same for all cr components ) and @xmath50 is a parameter that fixes the position of the knee , which is conveniently chosen to fit the observations .
the expression in eq .
( [ crfit1 ] ) provides then a smooth interpolation ) as @xmath51 , where the parameter @xmath52 will control the ` width ' of the knee , with @xmath53 leading to an abrupt change in slope .
we find however that @xmath54 provides a very satisfactory fit to the cr data for the @xmath55 case considered .
] between the flux @xmath40 at low energies @xmath56 and @xmath41 at high energies @xmath57
. = 5.truein=3.5truein the cr abundances at @xmath43 tev and the low - energy spectral indices are shown in table 1 , and were taken from the data compiled in @xcite . for the extragalactic component , we will assume for definiteness that it consists of protons ( this would anyhow only affect the predictions for @xmath58 ev ) and that it is given by @xmath59 i.e. a flux similar to that considered in @xcite .
for the purpose of illustration , we display in figure 1 the cases of both @xmath55 , corresponding to the diffusion / drift model ( when a kolmogorov spectrum of fluctuations is assumed for the random magnetic field component in which cr particles propagate @xcite ) , and @xmath60 , corresponding to an extreme model advocated in @xcite .
figure 1 shows the fits that result from taking @xmath55 with the parameter value @xmath61ev , and @xmath60 with @xmath62ev .
while the former exhibits an excellent agreement with the experimental observations , it can be seen that the latter shows an abrupt suppression above @xmath63 ev which results from the steep suppression of the different cr components above their knees @xcite , and due to this we will hereafter just focus on the @xmath55 case .
the cr particles reaching the top of the atmosphere produce secondary fluxes of hadrons and leptons that are usually described by means of coupled cascade equations @xcite , which can be solved analytically under appropriate symplifying assumptions . in this work
we will compute the fluxes of muon neutrinos and antineutrinos , which are the ones more readily detectable at underground detectors . at the energies considered the neutrino flavor oscillations will have no effect on the atmospheric neutrinos , although they may be relevant for the neutrinos produced by cr interactions with the ism ( eventually redistributing in that case the produced @xmath64 and @xmath65 fluxes evenly among the three flavors , something that can be incorporated into the final results straightforwardly ) .
the main contributions to the neutrino flux in the range of interest of this work ( namely , for @xmath66 gev ) arise , on the one hand , from the decay of the mesons @xmath67 and @xmath68 , which produce the so called conventional atmospheric neutrinos that dominate at low energies , and on the other hand , from the decay of the charmed mesons @xmath69 and @xmath70 , which give rise to the so called prompt charm neutrino flux that dominates at very high energies .
the muon contribution to the atmospheric neutrino flux is completely negligible in this context , since above @xmath71 gev the muons reach the ground and are stopped before decaying .
the transport equation associated to a given cascade component @xmath72 can be written as @xmath73 where as usual @xmath74 is the slant depth , @xmath75 is the interaction length in air and @xmath76 is the decay length .
both characteristic lengths are measured in g/@xmath11 units ( i.e. they include the factor @xmath77 that corresponds to the local density of the atmosphere ) . also notice that the decay length is linear in the energy due to the lorentz time dilation factor , while the interaction length exhibits instead a much milder energy dependence . the last term in equation ( [ casc ] )
corresponds to the production / regeneration term given by @xmath78 which actually couples the transport equations of different cascade components .
notice that @xmath79 here refers to the total @xmath80air cross section , while @xmath81 corresponds to the process @xmath80air
@xmath82 @xmath83 anything .
since the energies involved in the nuclear collisions between crs and atmospheric nuclei are much higher than nuclear binding energies , only nucleon - nucleon interactions are relevant in this context .
the different cascade components that we will take into account are hence @xmath84 ( that corresponds to the nucleon component in which protons and neutrons are grouped together ) and @xmath85 and @xmath70 . as in @xcite , the @xmath86 can be considered as contributing to the pion flux .
moreover , we will assume for simplicity that the only non - negligible terms will be those corresponding to regeneration ( i.e. @xmath87 and @xmath88 ) and to meson production by nucleons ( i.e. @xmath89 ) @xcite .
let us consider in particular fluxes with the form @xmath90 ( i.e. we consider the separation of variables in the fluxes , in which the energy dependent factor has a power - law behavior ) .
then , it turns out that @xmath91 where @xmath92 is the production / regeneration moment defined by @xmath93 with @xmath94 . assuming that the interaction lengths are energy independent and that the differential production distribution is feynman scaling , the production / regeneration moments themselves become energy independent . indeed , for the relevant interaction lengths and production / regeneration moments involving nucleons and non - charmed mesons
, we consider the energy independent values given in @xcite . when necessary , we also take the @xmath95dependence from fits to the data compiled and plotted in @xcite .
moreover , following @xcite , the charm interaction lengths and regeneration moments are all set equal to the kaon ones . for the production of charmed particles , however , one has to take into account the energy dependent moment given by equation ( [ zmoment ] ) , since the results are very sensitive to the particular cross section adopted for the charm production process .
indeed , the various estimates of the prompt atmospheric fluxes calculated so far are found to differ by almost 2 orders of magnitude @xcite . due to their relatively large mass , charm quarks
are usually considered to be produced in hard processes which can be well described by perturbative qcd ( pqcd ) . to leading order ( lo ) in the coupling constant , the processes that contribute to the charm production cross section are the gluon - gluon fusion process @xmath96 and the quark - antiquark annihilation process @xmath97 . however , at the next to leading order ( nlo ) the gluon scattering process @xmath98 shows up , giving a very significant contribution to the total cross section which increases it by a factor of @xmath99 @xcite . concerning higher order corrections , their contribution is certainly small since there are no qualitatively new channels opening up .
several parton distribution functions have been proposed and provide theoretical pqcd predictions that fit the available accelerator data reasonably well . however , in order to calculate the atmospheric charm flux the parton distribution functions need to be extrapolated to very small parton momentum fractions , typically corresponding to @xmath100 gev/@xmath101 @xcite , which at the highest energies is well outside of the measured region ( @xmath102 , and the uncertainty in the extrapolation affects significantly the final results @xcite . in order to illustrate this uncertainty range , we display the results obtained with two different structure distribution functions , namely the cteq3 parton distribution function @xcite and the golec - biernat , wsthoff ( gbw ) model @xcite , which includes gluon saturation effects .
the corresponding charm production cross sections were obtained either directly from the fit given in @xcite ( for the results based on the gbw saturation model ) , or by fitting the results given in figure 4 of @xcite for the cteq3 structure functions set with @xmath103 , with the charm mass @xmath104 gev , and interpolating the results for different energies . for the nucleon - air cross section
, we used the parametrization given by @xcite @xmath105{\rm{mb}}\ .
\label{sigman}\ ] ] once all relevant interaction lengths and production / regeneration moments are determined , the nucleon and meson atmospheric fluxes can be calculated from the coupled cascade equations given by ( [ casc ] ) and ( [ coupling2 ] ) .
recalling that these expressions were obtained under the assumption of power - law nucleon and meson spectra , we will first discuss the results for an initial nucleon flux given according to @xmath106 , discussing the general case further below . under the assumptions mentioned above
, the nucleon flux develops independently from the meson fluxes and is given by @xmath107 where the nucleon attenuation length is defined as @xmath108 concerning the meson cascade equations , they are usually solved by considering separately the low energy solution @xmath109 ( which neglects the interaction and regeneration terms , since @xmath110 at low energies ) and the high energy solution @xmath111 ( which neglects instead the decay term , since @xmath112 at very high energies ) @xcite .
the high energy solution is given by @xmath113 where the meson attenuation length @xmath114 is defined analogously to equation ( [ lambdan ] ) .
for the low energy case , the solution is obtained by replacing @xmath115 in the last equation . recalling that @xmath116 , the resulting expression couples the dependence on the energy and the slant depth , and hence it does not lead to just a simple power law energy spectrum . however
, for not very small values of @xmath74 the solution reduces to @xmath117 in equations ( [ nucleons])([mesonsl ] ) , @xmath118 and @xmath119 are to be evaluated for the nucleon spectral index @xmath120 .
notice also that the high energy meson flux has the same spectral index as the nucleon flux , while the low energy meson solution is flatter by an extra power of the energy , due to the energy dependence implicit in @xmath121 .
the neutrino flux produced via the meson weak decay can be calculated following the same scheme described above for the atmospheric nucleon and meson fluxes .
indeed , the neutrino flux is given by a transport equation analogous to equation ( [ casc ] ) , but actually much simpler since it contains only the source terms arising from meson decay , i.e. @xmath122 where @xmath123 the decay distribution can be put in terms of @xmath124 , the inclusive neutrino spectrum in the decay of meson @xmath125 , by means of the relation @xmath126 where @xmath127 is the branching ratio for the decay of meson @xmath125 into a state with the given neutrino @xmath128 . in the ultrarelativistic limit ,
the inclusive neutrino spectrum scales as @xmath129 @xcite .
moreover , we have already derived the asymptotic solutions for the meson fluxes , which are of the form @xmath130 , with @xmath131 ( @xmath132 ) for the high ( low ) energy fluxes ( see equations ( [ mesonsh ] ) and ( [ mesonsl ] ) ) .
thus , the source terms given by equation ( [ nuflux2 ] ) can be rewritten as @xmath133 where the @xmath134dependent meson decay moments are defined by @xmath135 the decay moments used in this work were calculated by fitting the @xmath134 dependence of the relevant moments tabulated in @xcite for @xmath136 , which were determined with lund monte carlo simulation programs .
then , the asymptotic low and high energy neutrino fluxes can now be determined from the equations ( [ nuflux1 ] ) and ( [ nuflux3 ] ) , which involve the meson decay moments and the corresponding high and low energy meson fluxes derived in equations ( [ mesonsh ] ) and ( [ mesonsl ] ) .
the low energy solution reads @xmath137 as expected , the neutrino flux develops rapidly , on the scale of a nucleon interaction length , and then remains stable . hence , at ground level ( @xmath138 ) the low energy flux will be @xmath139 similarly , the high energy flux at ground is @xmath140 where @xmath141 is the zenith angle of the cr incidence direction considered and @xmath142 with @xmath143 and @xmath144 the meson s mass and mean life , respectively , and @xmath145 km a typical scale height for density variations in the atmosphere . since the production / regeneration moments @xmath118 and @xmath119 are the same as those appearing in the meson fluxes , they are to be evaluated for the nucleon spectral index @xmath120 .
equation ( [ atmosh ] ) is actually valid for zenith angles sufficiently small so that the curvature of the earth can be neglected , and assumes a density profile of an isothermal atmosphere ( i.e. @xmath146 ) , which is appropriate in the stratosphere ( @xmath147 km ) where most particle interactions occur @xcite .
for very large angles , one can still use equation ( [ atmosh ] ) with the simple prescription of replacing @xmath148 km ) , where for a given line of sight that corresponds to an angle @xmath141 at the observer s position , the angle @xmath149 is the zenith angle that would be observed at a location where this same line of sight is at a height @xmath150 with respect to the ground @xcite .
for instance , for crs incident in the horizontal direction ( @xmath151 ) , one has that @xmath152 . in order to obtain a general expression for the atmospheric neutrino fluxes deep in the atmosphere
, the asymptotic solutions given by equations ( [ atmosl2 ] ) and ( [ atmosh ] ) can be joined together by means of the interpolation function @xmath153 which is actually analogous to the interpolation function given in equation ( [ crfit1 ] ) to join the fluxes associated to each galactic cr component below and above their knees .
we described so far a calculational scheme that provides the nucleon , meson and neutrino fluxes produced by an initial power law nucleon flux @xmath106 hitting the top of the atmosphere . we now seek the corresponding results for the full cr spectrum arriving at the earth , taking into account both the cr composition and the break in the cr spectra at the knee .
let @xmath154 be the cr flux associated to the cr component of nuclei of charge @xmath28 and mass @xmath155 ( with @xmath156 being the average mass number corresponding to element @xmath28 ) .
this nuclear component provides a nucleon flux @xmath157 given by @xmath158 .
if the flux of the cr component of charge @xmath28 is @xmath159 , the corresponding nucleon flux is then given by @xmath160 > from these considerations , the procedure to calculate the neutrino flux produced by the galactic cr flux parametrized by equations ( [ crfit1 ] ) and ( [ crfit2 ] ) is now straightforward .
indeed , the cr component of charge @xmath28 is represented by two power law fluxes @xmath161 and @xmath162 , with spectral indices @xmath48 and @xmath163 , respectively .
for each power law flux , we can compute the corresponding nucleon flux given by equation ( [ crnucleon ] ) and use it to calculate the associated neutrino flux .
then , the final neutrino flux produced by the given cr component can be obtained by interpolating the two solutions , using again the interpolation function analogous to equation ( [ crfit1 ] ) . finally , the total neutrino flux results from summing over the contribution of all galactic cr components together with the additional contribution from the extragalactic component given by equation ( [ xgflux ] ) .
figure 2 shows the atmospheric neutrino fluxes that correspond to the @xmath55 rigidity dependent scenario , with the parametrization formulae and parameter values given in section 2 .
the figure shows the contribution coming from different components and for different cases , namely the vertical and horizontal conventional , prompt charm / gbw and prompt charm / cteq3 fluxes . for the prompt charm flux calculated with the gbw model
, the figure also shows the total vertical and horizontal neutrino fluxes .
finally , in the latter case the conventional , charm and total neutrino fluxes produced by the extragalactic cr component is also indicated .
= 4.7truein=3.2truein as is clear from equation ( [ crnucleon ] ) , the induced neutrino fluxes are sensitive , on the one hand , to the cr composition assumed and , on the other hand , to the model assumed to reproduce the steepening of the cr spectrum at the knee .
previous works that have taken the knee into account have assumed for simplicity that the dominant cr component consists of protons alone , having a spectral change @xmath164 at @xmath30 . in order to observe the effect of the cr composition on the neutrino fluxes ,
figure 3 compares the total horizontal atmospheric fluxes ( with the prompt charm component calculated with the gbw model ) as obtained for different assumptions on the cr spectrum , namely the @xmath55 rigidity dependent scenario ( that corresponds to the results given already in figure 2 ) , the same total cr spectrum but assuming it consists only of protons , and the cr spectrum used in @xcite ( which is assumed to be constituted by protons and has a single spectral index change of @xmath164 at @xmath165ev ) .
the latter cr spectrum has a normalization which is lower by a factor of @xmath166 compared to the former ones , and indeed it seems to lay somewhat below the present observational data . comparing the results for the same total cr spectrum
, one can observe that the neutrino flux corresponding to the composition of different nuclear species is clearly below that produced by a cr spectrum formed only by protons , and also that this effect is more significant at higher energies . indeed , equation ( [ crnucleon ] ) shows that the nuclear cr component of charge @xmath28 and mass @xmath156 gives a contribution to the nucleon flux which is suppressed by a factor of @xmath167 below the respective knee , while it becomes suppressed by a factor @xmath168 well above it .
this effect was not appropriately treated in @xcite , where the suppression was considered to be given by a factor of @xmath169 , and moreover the impact of the change in the composition at the knee was never discussed before .
the strong suppression of the @xmath128 flux produced by the heavier components implies that the light ( h and he ) components are still responsible for a large fraction of the neutrinos above their knee , and hence the change in slope of the individual components is also reflected in the change in slope of the neutrino fluxes .
the neutrino fluxes produced during the propagation of the cr particles through the ism can be determined following the procedure described in the preceding section for the production of atmospheric neutrinos . for simplicity , we consider that the crs are distributed homogeneously within the galactic disk .
the highly relativistic cr particles interact with the ambient gas in the ism , which is a very low - density , non - relativistic plasma constituted chiefly by atomic and molecular hydrogen that can be described by an homogeneous nucleon density @xmath21 @xcite .
the treatment follows essentially the same considerations described above , but in this case one has to consider that the interaction of the secondaries produced in the nucleon - nucleon interactions is completely negligible due to the extremely low ism densities , which allow for their decay well before interactions can take place , irrespective of the energy considered .
hence , the main contributions to the diffuse neutrino flux produced in the galaxy arise from the decay of pions and muons , while the additional contributions coming from production and decay of heavier mesons can be safely disregarded . as in the preceding section ,
let us first consider an initial nucleon flux given by @xmath106 .
notice that in this context the atmospheric slant depth should be replaced by the ism column density traversed along the line of sight ( i.e. @xmath170 , where @xmath171 is the distance measured from the border of the galactic disk along the line of sight ) .
due to the extreme faintness of the ism , we should now take into account that corresponds to the interaction length for nucleons propagating through the ism .
notice that @xmath172 , expressed in g/@xmath11 , is actually very similar to the one corresponding to nucleon propagation in air .
in contrast , the meson decay length @xmath121 , which is linear in the mass density of the medium , will be much smaller in the ism than in the atmosphere . ]
indeed , it turns out that @xmath174 where the nucleon - nucleon total cross section is @xmath175\ { \rm mb}\ , \label{sigmann}\ ] ] and where @xmath176 stands for the center of mass energy squared @xcite .
hence , considering equation ( [ atmosl1 ] ) with @xmath173 , the neutrino flux produced from pion decay in the ism is @xmath177 in order to determine the muon decay contribution , it suffices to estimate it directly from the results obtained already for pion decay . from the decay kinematics , the mean fraction of energy ( relative to the parent particle ) in the @xmath178 decay
is @xmath179 for the @xmath65 and 0.79 for the muon @xcite .
analogously , in the @xmath180 decay the effective fraction of energy for the resulting @xmath65 is 0.35 @xcite , i.e. a fraction @xmath181 relative to the original pion .
hence , one expects a muon decay contribution approximately a factor @xmath182 larger than the neutrino flux from pion decay ( for instance , a factor of @xmath183 for a nucleon spectral index @xmath184 , in sensible agreement with more detailed calculations @xcite ) .
= 4.1truein=2.6truein = 4.1truein=2.6truein as commented above , the galactic neutrino flavor oscillations may redistribute the produced @xmath65 and @xmath64 fluxes evenly among the three flavors . in that case
, the calculation of the electron neutrino flux produced by cr interactions with the ism is also required in order to determine the galactic muon neutrino flux reaching the earth .
this flux is mainly produced in the @xmath180 decay , in which the effective fraction of energy for the @xmath64 is 0.3 @xcite .
thus , the @xmath64 flux is approximately a factor @xmath185 times the @xmath65 flux produced in the muon decay ( i.e. a factor of @xmath186 for a nucleon spectral index @xmath184 , again in good agreement with previous , more detailed calculations @xcite ) .
hence , if the neutrino flavors are redistributed by oscillations , the ( @xmath187 ) galactic flux will be a factor @xmath188 of that depicted in figure 4 , which neglects the neutrino oscillations . in order to calculate the galactic neutrino flux induced by the full cr spectrum
, the procedure follows the same steps already described above for the atmospheric fluxes .
the total galactic neutrino fluxes are shown in figure 4(a ) for @xmath189 and 20 kpc , and correspond to the @xmath55 rigidity dependent scenario ( with the same parametrization used for the previous figures and detailed in section 2 ) .
for comparison , the figure also shows the total atmospheric neutrino flux in the horizontal direction with the prompt charm component calculated using the gbw model , as well as the waxman - bahcall flux predictions for the astrophysical diffuse neutrino fluxes produced in their model of grbs @xcite .
analogously , figure 4(b ) shows the results corresponding to the same total cr spectrum but assuming it to consist only of protons . in agreement with previous results @xcite
, we observe that the galactic flux in the direction orthogonal to the plane ( corresponding for instance to @xmath189 kpc ) remains below the atmospheric flux up to the highest energies . along directions near the galactic plane ( e.g. for @xmath190 kpc )
, the flux of galactic neutrinos produced in the ism instead overcomes the atmospheric flux at energies larger than @xmath24 ev . the energy at which the galactic flux actually dominates depends to a large extent on the particular model adopted for charm production , as discussed above , and the results naturally also vary according to the assumed ism column density and the zenith angle of arrival direction in the atmosphere .
these figures show very clearly the significant effect of suppression in the neutrino fluxes that results from considering a rigidity dependent scenario .
indeed , above @xmath191 ev the waxman - bahcall flux is found to overcome the background of galactic neutrinos with trajectories nearly contained in the galactic plane .
however , if the same total cr spectrum is taken to be formed by protons alone , the resulting background flux of galactic neutrinos turns out to be larger than the waxman - bahcall flux up to @xmath192 ev , and then both fluxes stay with comparable magnitude up to the highest energies .
hence , rigidity dependent scenarios result in a significantly reduced background for the search of astrophysical neutrino sources of current interest , such as for instance grbs or agns .
we computed in this work the high energy diffuse fluxes of muon neutrinos ( and antineutrinos ) produced by crs hitting the upper atmosphere or interacting with the ism in the galaxy , showing that the results are quite sensitive to both the total cr spectrum and the cr composition assumed . concerning the former , we compared previous results ( which adopted an overall normalization for the cr spectrum actually somewhat below present observational data , and a single and abrupt slope change representing the knee ) with those obtained from considering a smooth cr spectrum that soundly fits the observations and reproduces its main relevant features ( namely , the knee , the second knee and the ankle ) . with respect to the latter , we showed that taking into account a cr composition that turns heavier above the knee ( i.e. in agreement with the scenarios that explain the knee as due to a rigidity dependent effect ) the induced neutrino fluxes become significantly suppressed , thus making their detection harder but also reducing the background for the search of other astrophysical neutrino sources . in particular , we compared our results with the waxman - bahcall predictions for the neutrino fluxes produced in their model of grbs , and found indeed that the signal - to - background ratio for this model becomes significantly enhanced when considering a cr composition with the contribution of different nuclear species , as opposed to the case in which the full cr spectrum is assumed to be composed of protons alone . an analogous effect would lower the neutrino flux produced by the extragalactic component ( which gives the dominant contribution to the total neutrino flux above @xmath193 ev ) if it were not constituted just by protons , as we assumed here .
it is then clear that if the diffuse neutrino fluxes predicted in the grbs models @xcite or the even larger ones associated to some agn models @xcite do actually exist , the observation of atmospheric or galactic cr induced neutrino fluxes above @xmath25 ev will remain probably hopeless .
let us also mention that it may be of interest to consider the atmospheric muon fluxes produced by crs .
for instance , it has recently been suggested @xcite that the observation of the down - going atmospheric muons with neutrino telescopes would provide an indirect measure of the prompt atmospheric neutrino flux , and hence this could be used to confront the nlo qcd predictions .
indeed , due to the charmed particle semileptonic decay kinematics , it turns out that the prompt muon flux coincides with the prompt neutrino flux to within @xmath194 , irrespective of the energy and independently of the model used to treat the atmospheric charm production @xcite . on the other hand ,
the atmospheric conventional muon flux is about a factor of @xmath195 larger than the conventional muon neutrino flux within the energy range of interest of this work , and exhibits roughly the same energy dependence @xcite .
completely analogous effects as those described here for the neutrino fluxes should then be expected for the atmospheric muons produced by crs reaching the earth .
finally , we stress the key importance of determining confidently the cr composition around and above the knee , since it appears in this context as decisive in order to estimate reliably the diffuse high energy neutrino background produced by crs .
moreover , solving the cr composition puzzle will also provide a valuable means of testing the different proposals concerning the origin and nature of the knee , and will thus shed new light on this long standing problem .
work supported by conicet and fundacin antorchas , argentina .
100 volkova l v , 1980 yad . fiz . * 31 * 1510 + volkova l v , 1980 sov .
. phys . * 31 * 784 ( translation ) gaisser t k , 1990 cosmic rays and particle physics ( cambridge university press , cambridge ) lipari p , 1993 astropart . phys . * 1 * 195 thunman m , ingelman g and gondolo p , 1996 astropart . phys . * 5 * 309 pasquali l , reno m h and sarcevic i , 1999 phys .
d * 59 * 034020 martin a d , ryskin m g and stato a m , 2003 _ preprint _ hep - ph/0302140 ingelman g and thunman m , 1996 _ preprint _ hep - ph/9604286 stecker f w , 1979 apj * 228 * 919 domokos g , elliott b and kovesi - domokos s , 1993 j. phys .
g * 19 * 899 berezinskii v s _
et al _ , 1993 astropart .
* 1 * 281 stecker f w and salamon m h , 1996 space sci . rev . * 75 * 341 waxman e and bahcall j , 1999 phys .
d * 59 * 023002 ahrens j _ et al _ , 2003 _ preprint _ astro - ph/0305196 fichtel c e and linsley j , 1986 apj * 300 * 474 jokipii j r and morfill g e , 1986 apj * 312 * 170 kobayakawa k , honda y s , and samura t , 2002 phys .
d * 66 * 083004 syrovatsky s i , 1971 comm .
space phys .
* 3 * 155 wdowczyk j and wolfendale a w , 1984 j. phys .
g * 10 * 1453 ptuskin v s _
et al _ , 1993 astron .
* 268 * 726 candia j , roulet e and epele l n , 2002 j. high energy phys . jhep12(2002)033
kampert k - h _ et al _ ( kascade collaboration ) , 1999 proc .
. cosmic ray conf .
( salt lake city , ut , 1999 ) og.1.2.11 + kampert
k - h _ et al _ ( kascade collaboration ) , 2001 _ preprint _ astro - ph/0102266 .
aglietta m _
et al _ ( eas - top and macro collaborations ) , 2003 _ preprint _ astro - ph/0305325 candia j , mollerach s and roulet e , 2002 j. high energy phys .
jhep12(2002)032 candia j , mollerach s and roulet e , 2003 j. cosmol .
jcap05(2003)003 nagano m and watson a a , 2000 rev .
* 72 * 689 wiebel - sooth b and biermann p l , 1998 astron . astrophys . * 330 * 389 hrandel j r , 2003 astropart . phys . * 19 * 193 gelmini g , gondolo p and varieschi g , 2003 phys .
d * 67 * 017301 nason p , dawson s and ellis r k , 1988 nucl .
b * 303 * 607 beenakker w _
et al _ , 1989 phys .
d * 40 * 54 matsuura t , van der marck s c and van neerven w l , 1989 nucl .
b * 319 * 570 lai h _ et al _ ( cteq collaboration ) , 1995 phys . rev .
d * 51 * 4763 golec - biernat k j and wsthoff m , 1999 phys .
d * 59 * 014017 + bartels j , golec - biernat k j and kowalski h , 2002 phys .
d * 66 * 014001 mielke h h _
et al _ , 1994 j. phys .
g * 20 * 637 berezinskii v s _
et al _ , 1990 astrophysics of cosmic rays ( amsterdam : north holland ) hagiwara k _ et al _ ( the particle data group ) , 2002 phys .
d * 66 * 010001 | scenarios in which the knee of the cosmic ray spectrum depends on the particle rigidities usually predict that the cosmic ray composition becomes heavier above the knee and have associated a change in the spectral slope of each individual nuclear component which is steeper than the change ( @xmath0 ) observed in the total spectrum .
we show that this implies that the very high energy ( @xmath1 ev ) diffuse neutrino fluxes produced by cosmic rays hitting the atmosphere or colliding with the interstellar medium in the galaxy will be significantly suppressed , making their detection harder but also reducing the background for the search of other ( more challenging ) astrophysical neutrino sources .
.5 cm .5 cm 21truecm 15truecm |
the phenomenon of detonation was discovered in the late ninetheenth century as a supersonic combustion wave that propagates around a thousand times faster than an ordinary flame .
the precise nature of detonation was elucidated in the works of mikhelson , chapman and jouguet at the turn of the twentieth century and of zeldovich , von neumann and dring in the 1940 s ( znd theory , @xcite ) .
it was established that a detonation is a shock wave that compresses and heats the reactive gas to temperatures sufficiently high to ignite chemical reactions within a short distance from the shock .
gas expansion caused by the heat release in these reactions creates pressure waves that , on the one hand , support the propagation of the shock and , on the other , accelerate the flow relative to the shock .
a self - sustained detonation is possible if such flow acceleration is sufficient to make the flow reach sonic conditions relative to the shock .
the presence of a sonic state at some distance from the shock isolates the flow between the shock and the sonic point from the influence of the conditions downstream of the sonic locus , thus making the wave self sustained .
most actual detonations are self sustained and therefore represent an important class of detonations . in znd theory , a detonation is assumed to be one - dimensional and in a steady - state in a galilean frame moving with the wave .
the theory is successful in explaining the fundamental nature of a detonation as a coupled shock - reaction zone system .
however , in experiments with gaseous detonations , it was observed as early as 1926 that detonations tend to be unsteady and multi - dimensional with rather complex structures and dynamics @xcite .
to explain these dynamics , theoretical efforts focused on analysis of the stability of znd solutions began in the 1960 s in the works of zaidel , pukhnachev , schelkin and , most comprehensively , erpenbeck ( see @xcite for an early literature review ) .
the theory of linear stability of gaseous detonations was found to be rather involved especially with regard to specific numerical computations for real gaseous mixtures .
much later , lee and stewart @xcite revisited the stability problem and solved it for an idealized system by employing a relatively straightforward normal - mode approach .
solving the linear stability problem for real complex gaseous mixtures or for systems with losses remains an open problem .
linear stability theory is successful in predicting , for idealized systems , why and how znd detonations are unstable to linear perturbations .
it is able to predict the neutral stability boundary and the most unstable modes and thus the characteristic length scales of various multi - dimensional structures seen in experiments .
however , the predictive power of linear stability theory is limited by its very nature as a linear theory .
gaseous detonations are known to be highly nonlinear and unsteady .
the lead shock of such a detonation wave propagates with strong oscillations in its velocity and with a highly non - uniform flow behind the shock that involves additional shocks propagating transversely to the lead shock with triple points forming at their intersection .
the triple points in complex detonation waves propagating in tubes with walls covered with soot leave fish - scale traces called detonation cells @xcite , and these detonations are termed cellular .
theoretical prediction of the origin and structure of cellular detonations remains an outstanding and challenging open problem .
however , some progress has been achieved with use of the tools of asymptotic analysis that allow for insight into the nature of the problem under various limiting conditions .
for example , large activation energy @xcite , small heat - release @xcite , slow time evolution and small shock curvature @xcite , weak nonlinearity @xcite , strong overdrive @xcite and other limiting assumptions lead to relatively simple asymptotic models .
these asymptotic theories in general describe idealized systems and are thus also limited in their predictive power .
nevertheless , they provide important information about the mechanisms involved in the existence of particular qualitative traits in the real phenomenon .
two fundamental structural and dynamical properties of gaseous detonations are first that they propagate in a galloping regime in narrow tubes , as characterized by large amplitude pulsations in an essentially one - dimensional wave , and second that the structure of detonation fronts in large channels / tubes or open environments is cellular . from a theoretical point of view
, the basic problem is to describe and explain , at least at a qualitative level , the origin and dynamics of these galloping and cellular detonations .
importantly , such detonations have been reproduced and extensively studied in numerical simulations of the reactive * * navier - stokes / euler equations with simpified descriptions of the chemical reactions and equations of state @xcite .
thus , from a modeling standpoint , the use of these equations is frequently appropriate .
some features of pulsating and cellular detonations have been reproduced theoretically in @xcite .
however , the following questions remain : 1 ) can an asymptotic theory predict the observed period - doubling sequence of bifurcations in one - dimensional detonations and the self - sustained cellular structures in two - dimensional detonations ?
2 ) how quantitatively close are the asymptotic predictions to the numerical results from the reactive navier - stokes / euler equations ?
our aim here is to develop a theory that captures the linear and nonlinear dynamics of self sustained detonations , especially with regard to the bifurcation sequence seen in numerical simulations of one - dimensional detonations and the cellular structures found in two - dimensional detonations .
our theory is asymptotic and relies on a number of approximations , namely small heat release , large activation energy , slow time evolution , weak curvature and the newtonian limit ( @xmath0 , where @xmath1 is the ratio of specific heats ) .
we build on the theory developed in @xcite by employing , as an additional approximation , the newtonian limit ( also used in @xcite in one spatial dimension for euler equations ) for two - dimensional detonations with retained dissipative effects .
the resulting system is a coupled set of three nonlinear partial differential equations .
when the dissipative terms are neglected , it is a hyperbolic system for which we compute the traveling wave solutions analogous to znd waves , their multi - dimensional linear stability properties and full nonlinear dynamics .
we provide a quantitative comparison with the results from the reactive euler equations for all three asymptotic predictions : the steady - state solutions , the linear stability spectrum and the cellular structure .
the analysis of the reduced system with retained dissipative effects is postponed for future work . as we show in section [ sec : weakly - nonlinear - approximation ] , the two - dimensional reactive navier - stokes equations reduce to a forced version of the unsteady , small disturbance , transonic equations ( utsd ) given by @xmath2 where @xmath3 and @xmath4 represent leading - order corrections to the @xmath5 velocity , @xmath6 velocity and reaction progress variable .
the right - hand side of ( [ eq : asymptotic - model - lambda-1 ] ) is the leading - order contribution of the reaction rate assumed for simplicity to follow a single step arrhenius kinetics .
the parameters @xmath7 and @xmath8 are the rescaled viscosity , pre - exponential factor , activation energy and heat release , respectively .
we find that this reduced asymptotic model captures , at both the qualitative and quantitative levels , not only the znd structure , but also the linear stability spectrum , the pulsating non - linear dynamics of one - dimensional detonations and the cellular dynamics of two - dimensional detonations .
we also mention the attempts at understanding the nonlinear dynamics of detonations via the use of qualitative models such as those introduced by fickett @xcite and majda @xcite .
the idea behind these models is to produce simplified systems that , although not derived from first principles , are capable of reproducing the observed complexity of the solutions of reactive euler equations while considerably simplifying the analysis .
the qualitative models of fickett and majda are closely related to the weakly nonlinear theory of detonations developed in @xcite .
although these models and the asymptotic theory of rosales and majda have enjoyed some success in explaining certain features of different types of traveling wave solutions of reactive euler / navier - stokes equations ( i.e. , weak and strong detonations ) , they were shown to lack the necessary complexity needed to reproduce the dynamical properties of real detonations with the rate functions used in prior work @xcite . however , simple _ ad hoc _ modifications of these models can reproduce much of the complexity of one - dimensional detonations as shown in recent work @xcite and also explained in the present work .
the remainder of this paper is organized as follows . in section
[ sec : governing - equations - of - reactive - flow ] , we state the main governing equations together with the modeling assumptions regarding the medium and the chemical reactions . in section [ sec : weakly - nonlinear - approximation ] , we develop an asymptotic approximation of the governing equations and obtain the weakly nonlinear reduced system .
we then investigate in section [ sec : traveling - wave - solutions - and - stability ] the possible traveling wave solutions of the asymptotic equations and their linear stability properties .
both the traveling wave solutions and stability spectrum of the asymptotic model are compared with their corresponding results in the reactive euler system .
for the case with no dissipative effects , predictions of the asymptotic model are calculated numerically in section [ sec : nonlinear - dynamics ] , and a quantitative comparison with the predictions of the reactive euler equations is presented as well . finally , in section
[ sec : discussion - and - conclusions ] , we discuss the results as well as point out some remaining open problems .
we assume that the medium is described by the following system of equations expressing , respectively , the laws of conservation of mass , momentum and energy , and the chemical heat release : @xmath9 here , @xmath10 is the material derivative , @xmath11 is the specific volume , @xmath12 is the density , @xmath13 is the velocity , @xmath14 is the cauchy stress tensor , @xmath15 is the deformation tensor , @xmath16 is the double contraction of tensors , @xmath17 is the total internal energy per unit mass , @xmath18 is the heat release per unit mass , @xmath19 and @xmath20 represent the flux of energy and species , respectively , @xmath21 is the rate of reaction and @xmath22 is the reaction - progress variable that changes from @xmath23 in the fresh mixture to @xmath24 in the fully burnt products .
we make the following standard modeling assumptions ( e.g. , @xcite ) : 1 .
the fluid is newtonian , with the stokes assumption on the bulk viscosity , so that + @xmath25 , where @xmath26 is the dynamic viscosity , @xmath27 is the pressure and @xmath28 is the identity tensor .
the species flux is given by fick s law , @xmath29 , with @xmath30 denoting the diffusion coefficient .
the energy flux has contributions from both the heat conduction ( given by fourier s law ) and the species diffusion , so that @xmath31 , where @xmath32 is the heat diffusion coefficient .
the medium is a perfect gas described by the ideal - gas equation of state , @xmath33 , with the internal energy given by @xmath34 , where @xmath35 is the universal gas constant divided by the molecular weight and @xmath1 is the ratio of specific heats , assumed to be constant .
5 . for simplicity , we take the rate of reaction to be @xmath36 , with the added ignition temperature assumption that @xmath37 for @xmath38 for some temperature , @xmath39 . here , @xmath40 is the rate constant and @xmath41 is the activation energy .
more general rate functions can be considered , in principle , as long as appropriate sensitivity to temperature is preserved . with these assumptions ,
we can then rewrite ( [ eq : continuity]-[eq : reaction ] ) as @xmath42 for the analysis that follows , it is convenient to use @xmath43 to express the energy equation as @xmath44 we shall focus on the two - dimensional case for simplicity . consider a localized wave moving into an equilibrium , quiescent state and let @xmath45 , @xmath46 , @xmath47 and @xmath48 denote , respectively , the density , pressure , temperature and newtonian sound speed in the fresh mixture ahead of the wave .
we rescale the dependent variables with respect to this reference state .
the independent variables are rescaled as follows : @xmath49 where @xmath50 and @xmath51 are the original spatial variables and @xmath52 is a typical wave speed , which is to be determined in the process of deriving the asymptotic model by requiring non - triviality of the leading - order corrections .
the length scales , @xmath53 , @xmath54 and the time scale , @xmath55 , are chosen to reflect the appropriate physics of weakly non - linear waves .
we assume that @xmath56 is small , which means that the spatial scale of interest in the @xmath5-direction , which is * * related to * * the size of the reaction zone , is small compared with the typical distance covered by acoustic waves in time @xmath55 . for the transverse dimension , we assume the scaling @xmath57 .
this follows from the fact that , along a weakly curved front , a distance @xmath58 in the normal direction corresponds to a distance @xmath59 in the transverse direction .
several dimensionless groups appear upon rescaling of the governing equations .
we define the reynolds , pradtl and lewis numbers , respectively , as follows : @xmath60 where @xmath61 .
writing @xmath62 , it is convenient to introduce the differential operator : @xmath63 introducing the dimensionless parameters , @xmath64 the non - dimensional governing equations become : @xmath65+\rho\left(\frac{1}{\epsilon}u_{x}+\frac{1}{\sqrt{\epsilon}}v_{y}\right ) & = & 0,\label{eq : ns - continuity - dimensionless}\\ \rho l[u]+\frac{1}{\epsilon}p_{x } & = & \frac{1}{3\epsilon\mathrm{re}}\left(u_{xx}+\sqrt{\epsilon}v_{xy}\right)+\frac{1}{\mathrm{\epsilon re}}\left(u_{xx}+\epsilon u_{yy}\right),\label{eq : ns - x - momentum - dimensionless}\\ \rho l[v]+\frac{1}{\sqrt{\epsilon}}p_{y } & = & \frac{1}{3\epsilon\mathrm{re}}\left(\sqrt{\epsilon}u_{xy}+\epsilon v_{yy}\right)+\frac{1}{\epsilon\mathrm{re}}\left(v_{xx}+\epsilon v_{yy}\right),\label{eq : ns - y - momentum - dimensionless}\\ \rho l[t]-\frac{\left(\gamma-1\right)}{\gamma}l[p ] & = & \frac{\gamma-1}{\gamma}\left(q\rho w-\frac{2}{3\epsilon\mathrm{re}}\left(u_{x}+\sqrt{\epsilon}v_{y}\right)^{2}+\frac{1}{\epsilon\mathrm{re}}\left(u_{x}^{2}+\epsilon u_{y}^{2}+v_{x}^{2}+\epsilon v_{y}^{2}\right)\right)\nonumber \\ & & + \frac{\gamma-1}{\gamma}\frac{1}{\mathrm{\epsilon re}}\left(u_{x}^{2}+\sqrt{\epsilon}u_{y}v_{x}+\sqrt{\epsilon}v_{x}u_{y}+\epsilon v_{y}^{2}\right)+\frac{1}{\mathrm{\epsilon re}\mathrm{pr}}\left(t_{xx}+\epsilon t_{yy}\right),\quad \qquad\label{eq : ns - temperature - dimensionless}\\ \rho l[\lambda ] & = & \rho w+\frac{1}{\mathrm{\epsilon re}\mathrm{pr}\mathrm{le}}\left(\left(\rho\lambda_{x}\right)_{x}+\epsilon\left(\rho\lambda_{y}\right)_{y}\right),\label{eq : ns - reaction - dimensionless}\end{aligned}\ ] ] where @xmath66 is defined as @xmath67.\label{eq : w - dimensionless}\ ] ]
we develop an asymptotic simplification of the above general formulation by considering a weakly nonlinear detonation wave , for which we assume that the heat release is small , the activation energy is large and the newtonian limit , @xmath0 , holds . * * to be precise , we start from ( [ eq : ns - continuity - dimensionless]-[eq : ns - reaction - dimensionless ] ) and make the following assumptions : 1 .
@xmath68 , @xmath69 .
this assumption is chosen to ensure that the reaction rate affects the leading order expansion of @xmath22 . since @xmath70
, this assumption implies that the characteristic time scale , @xmath55 , of weakly nonlinear detonations is large compared to the collision time , @xmath71 , i.e. , @xmath72 .
@xmath73 , @xmath74 .
this assumption implies that the heat release does not play a role at the linear level .
it does not mean that the chemistry is unimportant , but that the heat release must have the appropriate size to balance the nonlinear effects .
the extra factor of @xmath75 in front of @xmath76 arises naturally in the governing equations ( see ( [ eq : ns - temperature - dimensionless ] ) ) and is retained in the definition of @xmath8 . with the further assumption below of small @xmath77 , this implies that @xmath76 is @xmath78 .
3 . @xmath79 @xmath80 .
this ensures that small temperature deviations which are @xmath81 for weak shocks in the newtonian limit have an @xmath82 influence on the reaction rate .
@xmath83 , @xmath84 .
this assumption is needed to balance the temperature fluctuations with both the acoustics and chemistry at the same order . without this assumption ,
the leading order corrections for density , velocity and temperature all behave the same way , as in a weakly nonlinear inert shock @xcite . as we show later
, having a temperature profile that is different from density / velocity profiles is crucial , as it allows the model derived here to incorporate the dynamical instabilities of detonation waves .
@xmath85 and @xmath86 are @xmath82 , while @xmath87 is @xmath88 .
other scalings that highlight different transport effects are of course possible , but they are not considered in this work . to understand some of the intuition behind the asymptotic ordering above , recall the following well - known fact for weak shocks @xcite . if the shock strength is measured by the relative jump in pressure across the shock , @xmath89 ( subscripts @xmath90 and @xmath91 denoting post- and pre - shock states , respectively ) , then the shock mach number is @xmath92\,\delta p+o\left(\left(\delta p\right)^{2}\right)$ ] and the shock temperature is @xmath93\,\delta p+o\left(\left(\delta p\right)^{2}\right)$ ] .
therefore , for weak shocks , with @xmath94 , in the newtonian limit , @xmath95 , the leading - order temperature correction is @xmath81 .
that is , all variables have an @xmath78 jump across the shock , but the temperature jump is smaller , only @xmath81 . in the chosen asymptotic approximation
, we therefore expect similar temperature behavior in the reaction zone as well , at least with inviscid detonations .
+ now , we assume the following expansions in the reaction zone : @xmath96 the fractional powers appear here because we aim at capturing weak curvature effects in the detonation front .
these effects induce a flow velocity transverse to the front , which is @xmath59 smaller than the longitudinal velocity .
expansions of this type are standard for waves incorporating the weak - curvature effect ( e.g. , @xcite ) . expanding @xmath97
, we find that @xmath98 , @xmath99 and @xmath100 .
using these relations to eliminate pressure perturbations , ( [ eq : ns - continuity - dimensionless]-[eq : ns - reaction - dimensionless ] ) yield @xmath101 because of the weak heat release assumption , we need to expand only the reaction progress variable and the reaction rate to the leading order .
as we shall see later , the leading - order corrections to temperature are of order @xmath102 , i.e. , @xmath103 , and therefore the leading - order reaction rate is given by @xmath104 balancing @xmath105 terms , we find that @xmath106\left[\begin{array}{r } \rho\\ u\\ v\\ t \end{array}\right]_{1x}=\left[\begin{array}{r } 0\\ 0\\ 0\\ 0 \end{array}\right].\label{eq : o1-system}\ ] ] this homogeneous system has non - trivial solutions if and only if the coefficient matrix ( denoted by @xmath107 from now on ) is singular .
therefore , we must have @xmath108 .
we focus on a right - going wave for which @xmath109 . after substituting @xmath110
we integrate the first equation in ( [ eq : o1-system ] ) to obtain @xmath111 for some , so far arbitrary , functions @xmath112 and @xmath113 . * * in order for our asymptotic expansions to hold as we approach the upstream state , which is assumed to be quiescent and uniform at all times , we require @xmath114 as @xmath115 ( upstream of the wave ) . as a consequence
, it follows that @xmath116 .
had we considered the case where the wave moves into a non - uniform background , the functions @xmath117 and @xmath118 would provide the freedom needed to match the asymptotic expansion in the wave zone to the flow ahead ( see , e.g. , @xcite ) .
a similar reasoning can be used to deduce from the third and fourth equation in ( [ eq : o1-system ] ) that @xmath119 . at @xmath120 , we obtain @xmath121\left[\begin{array}{r } \rho\\ u\\ v\\ t \end{array}\right]_{2x}=\left[\begin{array}{r } 0\\ 0\\ -u_{y}\\ 0 \end{array}\right],\label{eq : o - half - system}\ ] ] with the same matrix of coefficients , * @xmath107 * , as before . since @xmath107 is singular , this system has solutions if and only if the right - hand side is orthogonal to @xmath122 $ ] , which spans the left null - space of @xmath107 .
clearly , this condition is always satisfied here .
note that from the third and fourth equation in [ eq : o - half - system ] , we obtain that @xmath123 and @xmath124 , i.e. , letting @xmath125 such that @xmath126 finally , at @xmath78 , we obtain @xmath121\left[\begin{array}{r } \rho\\ u\\ v\\ t \end{array}\right]_{3x}=\left[\begin{array}{r } -u_{\tau}-2uu_{x}-v_{y}\\ -u_{\tau}+\frac{4}{3}\frac{1}{\mathrm{\epsilon re}}u_{xx}\\ -\rho_{2y}\\ q\omega-\frac{\gamma_{1}}{\gamma}u_{x } \end{array}\right].\label{eq : o - epsilon - system}\ ] ] notice that since @xmath127 , all terms in ( [ eq : o - epsilon - system ] ) are @xmath82 . the solvability condition for this system is that the right - hand side be orthogonal to @xmath122 $ ] , which yields : @xmath128 this equation together with @xmath129 forms a closed system of equations .
the temperature dependence in the leading order rate function , ( [ eq : omega ] ) , can be eliminated by integrating the last equation in ( [ eq : o - epsilon - system ] ) , @xmath130 giving @xmath131 therefore , we obtain @xmath132 as can be seen from ( [ eq : o - epsilon - system ] ) , species and heat diffusion play no role in the asymptotic equations up to @xmath78 .
had we considered different asymptotic orderings for the lewis and prandtl numbers , these effects would introduce diffusive terms in the energy and the reaction - rate equations . *
* we consider no such effects in this work and hence the only effective diffusion in the asymptotic model comes from the viscous dissipation .
it is convenient to further rescale the variables as : @xmath133 where the scale for @xmath134 is chosen so that the traveling wave solution found later in section [ sub : travelling - wave - solutions ] has speed @xmath135 .
we also choose @xmath136 , which means that @xmath137 .
other choices of @xmath58 are possible as long as @xmath58 and @xmath77 are of the same asymptotic order when @xmath138 . these choices ,
although equivalent in the limit @xmath139 , will have some effect on the quantitative numerical predictions for finite @xmath58 . with this rescaling and the choice of @xmath58 , we obtain our final asymptotic system of equations of weakly nonlinear detonation : @xmath140,\label{eq : asymptotic - model - lambda}\end{aligned}\ ] ] where @xmath141 is the dimensionless viscosity coefficient . in the inviscid case , ( [ eq : asymptotic - model - u]-[eq : asymptotic - model - lambda ] ) must be supplemented by the appropriate jump conditions across shocks .
if the shock locus is defined by @xmath142 and @xmath143 $ ] denotes the jump of @xmath144 across the shock , the rankine - hugoniot conditions are : @xmath145-\frac{1}{2}\left[u^{2}\right]+s_{y}\left[v\right ] & = & 0,\label{eq : asymptotic - jump - condition1}\\ s_{y}\left[u\right]+\left[v\right ] & = & 0,\label{eq : asymptotic - jump - condition2}\\ \left[\lambda\right ] & = & 0.\label{eq : asymptotic - jump - condition3}\end{aligned}\ ] ] these equations follow from the conservation form of ( [ eq : asymptotic - model - u]-[eq : asymptotic - model - lambda ] ) .
implicit in the definition ( [ eq : w - dimensionless ] ) of the rate of reaction , @xmath66 , lies the ignition temperature assumption , such that @xmath146 for @xmath147 .
thus , the leading order reaction rate , @xmath148 , given by ( [ eq : omega ] ) also satisfies @xmath149 for @xmath150 or , equivalently , for @xmath151 . in order to prevent reactions from occurring in the ambient state , a reasonable constraint on the ignition temperature
is that it be larger than the ambient temperature , @xmath47 .
furthermore , for the znd solution to exist , it is necessary that the ignition temperature , @xmath39 , be smaller than the temperature at the von neumann state of the znd solution . in the weakly nonlinear regime considered in this paper , we must therefore have @xmath152 .
this should be viewed as a modeling assumption about the chemical kinetics .
notice that in the inviscid case , the ignition temperature , if taken to be @xmath153 , has no effect on the znd profiles and their stability properties .
simply , it states that the reactions occur only after the shock and therefore , @xmath154 ahead of the wave .
when considering viscous effects , however , the traveling wave profiles and their dynamical evolution will depend on the ignition temperature and @xmath39 should therefore be considered as another parameter that affects the properties of the solutions . for the convenience of the reader , in table [ tab : summary - of - parameters ] , we collect the relations between various dimensionless and dimensional quantities .
.[tab : summary - of - parameters]summary of scaling relationships . [ cols="^,^,^",options="header " , ]
in the previous section , we showed that the asymptotic model exhibits the same linear stability behavior as the reactive euler equations .
the question of what happens after the onset of instabilities can be investigated through direct numerical simulations of the model equations .
we show below that the traveling wave solutions of the asymptotic model reproduce , in both one and two spatial dimensions , the complexity observed in solutions of the reactive euler equations .
we focus here on the ability of the asymptotic model to predict the complex nonlinear dynamics of pulsating ( galloping ) detonations .
we perform a detailed numerical investigation of the large time asymptotic behaviour of oscillatory solutions of the model . in the one - dimensional inviscid case ,
the system given by ( [ eq : asymptotic - model - u]-[eq : asymptotic - model - lambda ] ) reduces to @xmath155 this system resembles the one derived in @xcite with one crucial difference as a consequence of the @xmath156 assumption , the reaction rate function in ( [ eq:1d - asymptotic-2 ] ) has a more complicated @xmath4-dependence , which is in fact at the heart of the complexity of the solutions obtained here . system ( [ eq:1d - asymptotic-1]-[eq:1d - asymptotic-2 ] ) is also the same as in @xcite , where the @xmath156 assumption is used .
we solve ( [ eq:1d - asymptotic-1]-[eq:1d - asymptotic-2 ] ) numerically in a shock - attached frame @xcite , using a second - order finite volume scheme with a second - order total variation diminishing ( tvd ) runge - kutta temporal discretization @xcite .
because no differentiation across the shock is performed , true second - order convergence is obtained for the cases tested , which include the convergence to various stable steady - state znd solutions .
we further verified the numerical algorithm by performing a cross - validation between the linear stability solver and the direct numerical solver .
that is , we made sure that for several values of @xmath8 , the linear stability prediction of the neutral boundary agrees with the neutral stability boundary of the numerical scheme for the nonlinear model . for example , when @xmath157 , the linear stability curve in fig .
[ fig : neutral - stability - curves](a ) indicates that @xmath158 is the neutral value of the activation energy , i.e. , the znd wave is unstable for @xmath159 and stable for @xmath160 .
numerical simulations of ( [ eq:1d - asymptotic-1]-[eq:1d - asymptotic-2 ] ) with @xmath161 slightly above and slightly below @xmath162 confirm this prediction , as shown in fig .
[ fig : nonlinear - dynamics ] , where the shock state , @xmath163 , is plotted as a function of time .
we also compute solutions of ( [ eq:1d - asymptotic-1]-[eq:1d - asymptotic-2 ] ) further away from the neutral boundary in order to check if the model captures a sequence of bifurcations leading to chaos as occurs in the reactive euler equations @xcite .
such a sequence of bifurcations is indeed present in the model .
long time simulations show that the solutions tend to either a fixed point , a limit cycle or ( what appears to be ) a chaotic attractor .
we run the simulations at @xmath164 and plot the post - shock state , @xmath163 ( by the rankine - hugoniot conditions , @xmath165 , where @xmath166 is the shock speed ) , as a function of time for several different types of solution , as shown in fig .
[ fig : the - shock - state ] . beyond the stability boundary ,
the shock velocity becomes oscillatory . near the neutral boundary ,
the oscillations have a small amplitude and are periodic ( fig .
[ fig : the - shock - state](a ) ) , but the structure of each period becomes more complex as we move away from the neutral boundary by increasing the activation energy ( fig .
[ fig : the - shock - state](b , c ) ) .
eventually , a value of @xmath161 is reached at which no obvious period is evident as seen in fig .
[ fig : the - shock - state](d ) .
+ the behaviour described in the previous paragraph can be understood in terms of a period doubling sequence of bifurcations where the period becomes longer and more complex with each bifurcation . in order to construct a bifurcation diagram illustrating this process
, we proceed as follows : for each value of @xmath161 , we follow the evolution of the solution until it settles on the attractor .
then , we extract the set of local minima for @xmath167 a finite set for any periodic solution .
the values of these minima are then plotted versus the bifurcation parameter , @xmath161 .
the result is shown in fig .
[ fig : convergence - bif - diagram ] , which is reminiscent of the standard feigenbaum period doubling cascade leading to chaos @xcite .
it is important to note that the further we move into the unstable region , the harder it is to numerically capture the wave dynamics with good accuracy .
that is , in the highly unstable regime ( @xmath168 in fig .
[ fig : convergence - bif - diagram ] ) , the quantitative details of the bifurcation diagram are sensitively dependent on the grid resolution . in a truly chaotic regime , such sensitivity is intrinsic and reflects the nature of the system .
however , another reason , which is at play even before the apparently chaotic regime sets on , is that the wave dynamics can involve spatial scales that undergo large changes ( by orders of magnitude ) during the wave evolution .
this is a direct consequence of the arrhenius exponential dependence of the reaction rate , which can trigger large variations in the reaction rate from moderate changes in the temperature when the activation energy is large .
this issue of stiffness associated with high - activation energy detonations in unstable regimes is discussed next in more detail . in theoretical and numerical studies of detonation ,
a widely used spatial scale is taken to be the half - reaction length , @xmath169 , defined as the distance from the shock where half of the energy is released in the znd solution .
the space is then non - dimensionalized , as done in this work , so that the half - reaction happens over a unit length .
the numerical resolution is thus measured as a number of points per this unit of length .
although appropriate for stable or weakly unstable detonations , the znd half - reaction length and therefore the resolution measured on this scale become less meaningful when considering unstable detonations at high activation energies .
the reason is that the energy release can become extremely localized during the dynamical evolution of pulsating waves and therefore the actual number of grid points used to capture the heat release region can significantly decrease .
+ in order to provide a more quantitative measure of the variations of relevant spatial scales during the time evolution of a pulsating detonation , we introduce the width of the reaction zone as the smallest value , @xmath170 , such that there exists an interval , @xmath171 , of length @xmath170 , where @xmath172 percent of the energy is released . for nice enough rate functions ,
@xmath173 of the znd solution is roughly equivalent to the half - reaction length , @xmath169 .
notice , however , that for chemical reactions with large induction zones and/or localized heat release , the values of @xmath173 and @xmath169 are significantly different , with @xmath173 becoming a more relevant spatial scale . in the calculations below , we use @xmath174 such that @xmath175 represents , at a given time , the smallest width containing @xmath176 of the heat release . in fig .
[ fig : width - of - reaction - zone ] , we show @xmath175 as a function of time during the detonation evolution .
we fix the numerical resolution at @xmath177 points per half - reaction length ( @xmath178 ) , which may be considered an overkill for a steady znd wave .
we then compute @xmath175 as a function of time at different activation energies . as can be seen in fig .
[ fig : width - of - reaction - zone](a ) , in the weakly unstable regime , the width of the heat - release zone changes by a factor of about two during the time evolution of the wave ; the heat release region is still well resolved . further into the unstable regime ( fig .
[ fig : width - of - reaction - zone](b , c ) ) , we see that the relevant size of the reaction zone can shrink by more than an order of magnitude and , thus , even with @xmath177 points per half - reaction length , there are times when only about @xmath179 points are used to resolve the heat - release region .
if we increase the activation energy even further , stepping into the apparently chaotic regime , we observe very short windows of time when only about @xmath180 points are being used per heat - release zone .
the bifurcation diagram at @xmath164 showing the local minima of the attractor solution s shock value , @xmath167 , as a function of the activation energy , @xmath161 .
the simulations were carried out at @xmath181 ( red ) and @xmath182 ( black ) grid points on the computational domain of length @xmath183 .
away from the chaotic regimes , the predictions are seen to be nearly identical.,width=576,height=384 ] because of the difficulties outlined above , without resorting to adaptive mesh refinement , @xmath184 points per half - reaction length are needed to obtain a bifurcation diagram with features that are essentially grid independent away from the chaotic regimes ( see fig .
[ fig : convergence - bif - diagram ] ) . while recognizing that such sensitivity to initial conditions or discretization errors is natural for chaotic dynamics , caution is still required when interpreting the results of numerical simulations of unstable detonations with high activation energies .
multi - dimensional instability is very important in gaseous detonations and results in cellular structures involving triple - point interactions on the detonation lead shock @xcite .
it is therefore crucial to check if the asymptotic model , ( [ eq : asymptotic - model - u]-[eq : asymptotic - model - lambda ] ) , can reproduce the dynamics of not only one - dimensional , but also multi - dimensional detonations . as we have seen in section [ sub : the - spectral - stability ] , multi - dimensional instabilities can be dominant in the asymptotic model , with the dispersion relation showing a maximum growth rate for some nonzero transverse wave number @xmath185 ( see fig . [
fig : spectrum-2d ] ) . in this section ,
we calculate the long - term dynamics of detonation waves when two - dimensional effects are present .
in particular , we show that the asymptotic model retains the essential complexity required to reproduce multi - dimensional cellular patterns .
solving ( [ eq : asymptotic - model - inviscid - u]-[eq : asymptotic - model - inviscid - lambda ] ) numerically turns out to be a non - trivial task and requires special care . in the following subsection
, we present a discussion of the algorithm employed in this work . in order to appreciate the subtlety associated with ( [ eq : asymptotic - model - inviscid - u]-[eq : asymptotic - model - inviscid - lambda ] )
, we note that the equations comprise a nonlinear hyperbolic system with one of its characteristic planes orthogonal to time .
this means that : * the initial data are given on a characteristic surface , the @xmath186 plane .
the absence of a time derivative in ( [ eq : asymptotic - model - v]-[eq : asymptotic - model - lambda ] ) requires the initial conditions to satisfy @xmath187 , @xmath188 . *
evolving in @xmath189 is a nonlocal procedure and , in the presence of a shock , care has to be taken to avoid spurious numerical oscillations .
many existing numerical methods for solving ( [ eq : utsd-1]-[eq : utsd-2 ] ) are based on a formal rewriting of the system as a single equation , @xmath190 by cross - differentiation and substitution .
two concerns arise with this approach .
first , the validity of such a transformation is not obvious when @xmath134 and @xmath191 are discontinuous functions .
second , in the presence of chemical reactions , differentiation of ( [ eq : asymptotic - model - inviscid - u ] ) with respect to @xmath5 produces a delta forcing at the ignition - temperature locus due to the discontinuous nature of the reaction rate . thus , the techniques based on solving ( [ eq : utsd - cross - differentiated ] ) are inadequate for our purposes .
another common technique is to solve for the potential , @xmath192 , that satisfies @xmath193 again , when chemical reactions are present , complications arise because the reaction rate now becomes an exponential function of @xmath194 and the discretization errors in @xmath194 therefore exponentially amplify , requiring very high - order methods for good accuracy .
the method we employ here is based on a direct semi - implicit discretization of ( [ eq : asymptotic - model - u]-[eq : asymptotic - model - v ] ) * * following some of the ideas found in * * @xcite .
in this method , all terms except for @xmath195 are treated explicitly . by chosing second - order spatial and temporal discretizations
, we obtain an algorithm that is formally second order in time and space .
the general procedure is outlined here to explain our reasoning for the choice of the algorithm . assuming that the solution at time @xmath196 is known , we evolve it to @xmath197 as follows : 1 . first , we employ a semi - implicit time discretization , where the @xmath195 term is treated implicitly .
the motivation for this comes from the fact that some waves propagate infinitely fast in the @xmath186 plane . in the simple case of a forward euler time discretization
, we obtain @xmath198 where @xmath199 . a more quantitative reason for treating @xmath195 implicitly can be seen from the von neumann stability analysis of the linearized system without chemical reactions , wherein a fully explicit scheme can be unstable ( see appendix [ sec : von - neumann - stability - analysis ] for details ) .
we approximate the explicit terms @xmath200 and @xmath201 using a shock - capturing scheme , e.g. , finite volume or eno / weno algorithms @xcite .
3 . using ( [ eq : algorithm - semi - discrete-2 ] ) , we write a forward difference representation of @xmath202 in terms of @xmath203 and @xmath204 for @xmath205 .
for instance , a first - order forward difference scheme can be used , i.e. , @xmath206 where the @xmath6 derivative approximation is postponed until the next step .
the use of the forward difference here is a consequence of up - winding the infinitely fast waves propagating from right to left .
approximate the @xmath6 derivatives , e.g. , by centered differences , to obtain the fully discrete scheme : @xmath207,\label{eq : algorithm - u - ij}\\ v_{i , j}^{n+1 } & = & v_{i+1}^{n+1}-\frac{\delta x}{2\delta y}\left(u_{i , j+1}^{n+1}-u_{i , j}^{n+1}+u_{i , j-1}^{n+1}\right).\label{eq : algorithm - v - ij}\end{aligned}\ ] ] 5 . in order to solve ( [ eq : algorithm - u - ij]-[eq : algorithm - v - ij ] ) , sweep from right to
left , assuming the right boundary values of @xmath134 and @xmath191 are known at all times , in the following way : 1 . for some fixed @xmath208 , insert ( [ eq : algorithm - v - ij ] ) into ( [ eq : algorithm - u - ij ] ) and solve the linear system for the vector @xmath203 .
2 . with @xmath203
find @xmath202 using ( [ eq : algorithm - u - ij ] ) .
3 . repeat ( a),(b ) with @xmath209 until the left boundary is reached
finally , compute @xmath210 by solving ( [ eq : algorithm - semi - discrete-3 ] ) with a boundary condition on the right of the domain , which is given by @xmath211 notice that ( [ eq : algorithm - semi - discrete-3 ] ) is actually an initial value problem for @xmath210 for fixed @xmath6 wherein @xmath212 is a time - like direction .
it can be solved with any initial value solver ( e.g. , a runge - kutta method ) if desired .
it may seem counterintuitive at first that the simplified model requires a semi - implicit method while the reactive euler equations can be solved explicitly .
the reason is that the asymptotic approximation is performed in a limit where the reactive euler equations themselves would have to be treated implicitly . in order to understand this ,
we look at the three waves present in the euler equations .
since the weak heat release approximation implies that the detonation velocity is nearly acoustic , and an acoustic wave induces no flow behind it , we see that the speed of the forward acoustic characteristic , given by @xmath213 in a frame moving with the wave , is actually an @xmath78 quantity ( @xmath214 and @xmath215 ) .
the entropy and backward acoustic characteristics , on the other hand , have @xmath82 speeds . in the asymptotic model , a slow time , @xmath189 , is chosen so that the dynamics happen on @xmath82 time scales and therefore some characteristics have speeds of @xmath216 in the slow time variable . when an explicit method is used to solve the euler equations in this limit
, a typical cfl condition would require a time step , @xmath217 , and as @xmath138 , it becomes clear that the time step restriction becomes unattainable and an implicit method is needed .
the infinite characteristic velocity arises because the @xmath218 plane is a characteristic surface for the equations .
a simple rotation in space - time resolves , to some extent , this issue such that in the rotated space - time the system behaves as a standard system of conservation laws .
the trade - off for recovering a classical hyperbolic system with finite wave speeds is that one must now consider a grid with moving boundaries .
this idea can be exploited to produce fully explicit schemes as shown in @xcite for the case without reactions . using the algorithm described in the previous section , we solve ( [ eq : asymptotic - model - inviscid - u]-[eq : asymptotic - model - inviscid - lambda ] ) in a frame moving with constant speed , @xmath109 .
all simulations are initialized with the one - dimensional znd solutions obtained in section [ sub : travelling - wave - solutions ] .
we then investigate , in the context of the asymptotic equations : 1 ) the effect of the width of the channel on nonlinear stability properties of the travelling wave solutions ; 2 ) the effect of the periodic boundary conditions ; and 3 ) the effect of increasing heat relase on the wave dynamics . in fig .
[ fig : detonation - in - channel - vary - size ] , we show the profile of @xmath134 at @xmath219 for varying widths of the domain , @xmath220 , in the @xmath6 direction .
the @xmath6 boundaries are modelled as rigid walls , appropriate for a detonation in a two - dimensional channel ( i.e. , a channel with negligible depth ) .
the parameters are fixed at @xmath157 and @xmath221 , for which the linear stability calculation shows that the znd wave is unstable only to two - dimensional perturbations . in a channel of width @xmath222 ,
the znd wave remains stable , as seen in fig .
[ fig : detonation - in - channel - vary - size](a ) , as the spacing is too narrow for the two - dimensional instability to develop .
increasing the width of the channel , however , allows for a number of transverse modes to be excited , the allowed transverse wave numbers , @xmath185 , are given by @xmath223 , where @xmath224 .
therefore , only a discrete set of modes can be excited , and larger @xmath220 typically allows for more unstable modes to appear . ] , leading to the formation of a multi - dimensional cellular pattern as shown in figs .
[ fig : detonation - in - channel - vary - size](c , d ) .
the cell size appears to remain around @xmath180 in all figs .
[ fig : detonation - in - channel - vary - size](b - d ) .
when we replace the solid wall boundary conditions with periodic conditions , we observe a two - dimensional version of a spinning detonation .
there is only one family of transverse shock waves that all propagate in the same direction .
such a wave can be imagined to form in a narrow gap between two concentric cylindrical tubes , as in a rotating detonation engine @xcite . a snapshot of the solution field , @xmath134 , is shown in fig . [ fig : spinning - detonation - in - tube ] , where relatively strong transversely propagating shocks can be seen . [ fig : detonation - on - channel - vary - q ] + detonation in a channel of width @xmath225 with periodic boundary conditions .
parameters are @xmath157 and @xmath221 . , width=576,height=192 ] + in fig .
[ fig : detonation - in - channel - vary - q ] , we show the effect of the heat release on the solution structure .
we fix @xmath226 and increase the value of @xmath8 from @xmath227 to @xmath228 .
consistent with the behaviour of detonations in the reactive euler equations , regular cells are observed at small @xmath8 , as in figs .
[ fig : detonation - in - channel - vary - q](a , b ) , but with increasing @xmath8 , the structure of the detonation front becomes more complex with the formation of irregular cells as in fig .
[ fig : detonation - in - channel - vary - q](d ) . +
all of the previous results show that , at least at a qualitative level , the asymptotic model captures many important characteristics of multi - dimensional detonations .
next , we investigate how _ quantitatively _ close the predictions are to the solutions of the reactive euler equations .
the numerical simulations of the reactive euler equations were carried out using pyclaw @xcite , which provides , among other things , a python wrapper for the classic routines in clawpack .
the software requires the user to provide a riemann solver .
we use a roe - linearized riemann solver with a harten - hyman entropy fix @xcite .
the classic package of pyclaw employs a second - order finite volume algorithm with a fractional step method for the time evolution @xcite . for the numerical simulations , the reactive euler equations were non - dimensionalized in the conventional form by the ambient state with velocities scaled by @xmath229 spatial variables scaled by the half - reaction length , @xmath169 , and time by @xmath230 .
[ fig : nonlinear - dynamics-2d ] , we show a comparison between the asymptotic solutions and the solutions of the reactive euler equations .
the parameters chosen are the same as in fig .
[ fig : spectrum-2d - euler - vs - asymptotic](b ) , i.e. , @xmath231 , @xmath232 and @xmath233 , which in the asymptotic variables are given by @xmath234 and @xmath235 .
we start both simulations with the znd solution and solve for a time interval large enough such that instabilities have already fully developed .
notice that since the asymptotic model uses the slow time variable , @xmath189 , we only need to solve it for a relatively short time interval , @xmath236 .
the reactive euler system , on the other hand , was non - dimensionalized in the conventional way using the regular time variable , @xmath237 , and thus we must solve the system up to @xmath238 .
we perform all necessary scaling conversions between the variables so that the lengths and amplitudes shown in fig . [
fig : nonlinear - dynamics-2d ] are consistent .
the length of the @xmath6 domain is fixed in the asymptotic model to be @xmath239 , which corresponds to a length @xmath240 in the regular dimensionless @xmath6 , while the @xmath5 domain is of length @xmath241 .
+ we see from fig .
[ fig : nonlinear - dynamics-2d ] that the asymptotic model captures the salient features of multi - dimensional detonations with good quantitative agreement .
the characteristic scales of the detonation cells are seen to be close .
a small difference in the transverse propagation velocity of the triple points is apparent , in the asymptotic case the speed being higher .
the transverse shocks in the asymptotic solutions far from the lead shock are seen to be smoother than in the full solutions , which is due to a larger numerical diffusion in the algorithm for the asymptotic model . finally , we remark that weakly nonlinear detonations operate in the same fluid dynamic regime as weak shock wave focusing or weak shock reflection at near grazing incidence .
as pointed out by von neumann @xcite , in this limit the classical mach triple point shock structure is impossible . yet
, both experiments and numerical calculations @xcite consistently exhibit triple shock structures in this regime .
this apparent contradiction is known as the von neumann paradox . although the structures behind the lead shock in our calculations resemble the triple - points observed in detonations ( fig .
( [ fig : triple - point - like - structures ] ) ) , a simple algebraic argument can show that the asymptotic equations do not admit `` classical '' triple points , where three shocks separated by smooth flows meet @xcite .
thus , much like in the problem of weak shock reflection , the observed cells in the weak heat release detonations considered in this work present yet another instance of the `` von - neumann paradox '' . here , we make no attempt to address this problem and simply note that triple - point - like structures appear in our numerical simulations behind the lead shock ; whether they are true discontinuities , sharp waves or some singularities remains an open problem .
triple - point - like structures in the asymptotic solution of the weakly nonlinear detonation as shown by the magnitude of the density gradient.,width=384,height=288 ]
in this work , we develop an asymptotic theory of multi - dimensional detonations within the framework of the compressible navier - stokes equations for a perfect gas reacting with a one - step heat - release law . the main outcome is a reduced model that consists , in two spatial dimensions , of a forced burgers - like equation coupled with a heat release equation and an equation that enforces zero vorticity . after specializing the model to the case without dissipative effects ,
we have analyzed it in detail and have shown that it captures many of the dynamical features of detonations in reactive euler equations .
these include : 1 ) existence of steady - state travelling waves ( i.e. , znd solutions ) , 2 ) linear instability of the znd solutions , 3 ) existence of a period - doubling bifurcation road to chaos in the one - dimensional case , and 4 ) onset of cellular detonations in the two - dimensional case .
our theory considers weakly nonlinear detonations in the distinguished limit of small heat release , large activation energy and small @xmath77 . in this limit
, the dynamics occur on time scales that are long compared with the scale of a characteristic chemical reaction time .
our reduced evolution system builds on the weakly nonlinear theory in @xcite by adding the newtonian approximation to it . in the one - dimensional and non - dissipative special case
, our theory contains the model in @xcite . as a consequence of the @xmath156 assumption , the temperature expansion in our analysis starts with an @xmath81 correction to the leading - order term , as opposed to @xmath78 correction in @xcite .
this is precisely what allows us to escape the fact that in @xcite there must be leading - order corrections to temperature , velocity and density that behave exactly the same way .
the present scaling highlights an important difference between a reactive and a non - reactive shock , namely , that the temperature in a reactive shock can increase at some distance from the shock front because of heat release .
this increase in the post - shock temperature means that the energy release can possess a maximum inside the reaction zone . as we previously elucidated with a reactive burgers model @xcite
, the presence of such an internal maximum appears to be a key factor for the mechanism that generates resonances and subsequent amplification of the waves in the reaction zone .
note that all one - dimensional weakly nonlinear asymptotic models as well as the _ ad hoc _ models devised to capture detonation dynamics @xcite are extensions of the burgers equation modified by adding a reaction forcing term , with an extra equation for the reaction progress variable .
the reason is that the dynamics of any one - dimensional genuinely nonlinear hyperbolic system reduces to that of an inviscid burgers equation in the weakly nonlinear limit @xcite . while the reduced weakly nonlinear asymptotic theory has clear predictive power , as demonstrated in this work , recognizing its limitations is important and helps highlight open problems that require further investigation .
for example , at parameter values ( especially of the heat release ) corresponding to strong cellular and pulsating detonations typical in experiments and numerical simulations , the asymptotic predictions are not sufficiently accurate when compared to the reactive euler equations .
while it is not clear how to extend the present approach directly to strongly nonlinear detonations , the ability of the reduced system obtained in this work to capture essential dynamical characteristics of unsteady and multi - dimensional detonations serves as a strong indication that a theory of comparable simplicity may be possible for strongly nonlinear detonations .
recent work @xcite on the role of dissipation in the compressible reactive navier - stokes system in one spatial dimension indicates that dissipation can have non - trivial effects on the stability of detonation waves .
elucidation of the role played by the transport effects in the asymptotic model is therefore of interest and we believe that the model retaining these effects , as in ( [ eq : asymptotic - model - u]-[eq : asymptotic - model - lambda ] ) , should be investigated further .
a problem that merits exploration , in our opinion , is that of the evolution of a small - amplitude , localized initial perturbation to a detonation wave , considered in the same weakly nonlinear asymptotic regime as in the present work .
all of the underlying assumptions in the theory remain valid for this different initial - value problem , which is relevant to the problem of detonation initiation . from a mathematical point of view , the reduced system obtained in this work and its connection with the issue of the non - existence of triple - point shock structures for the zabolotskaya - khokhlov equation pose some puzzling and challenging problems . finally , recall that the qualitative models introduced by fickett and majda do not contain instabilities with the rate functions used in prior work with these models ( e.g. , see @xcite ) .
however , using the rate function derived in this paper ( and , for that matter , the two - step rate function used in @xcite ) can be shown to lead to the same complexity of the solutions as occurs in the reactive euler equations .
a modified fickett s analog can be written as @xmath242 with the `` activation energy '' and `` heat release '' parameters , @xmath243 and @xmath244 , respectively .
clearly , one can treat more general right - hand sides for the rate function in ( [ eq : new - fickett - majda-2 ] ) , such as @xmath245 , as long as @xmath246 and @xmath192 retain the same qualitative properties as their corresponding expressions in ( [ eq : new - fickett - majda-2 ] ) .
this modification , even though simple , reflects the physics of the phenomenon responsible for the dynamics of pulsating detonations by allowing a local maximum in the reaction rate to exist behind the precursor shock in the detonation wave .
mathematical study of the modified analog model is of interest from the point of view of the theory of hyperbolic balance laws and the dynamics of their solutions .
l.f . and a.k .
gratefully acknowledge research support by king abdullah university of science and technology ( kaust ) .
the research by r. r. rosales was partially supported by nsf grants dms-1007967 , dms-1115278 , dms-1318942 , and by kaust during his research visit in november 2013 .
l. f. would like to thank slava korneev and david ketcheson for their help with numerical computations .
in order to motivate the use of a semi - implicit scheme to solve ( [ eq : asymptotic - model - u]-[eq : asymptotic - model - lambda ] ) , here we perform a von - neumann stability analysis of a natural / reasonable explicit scheme and show that it leads to instabilities .
since such an analysis requires a constant coefficient linear system , we consider here the linearized , unsteady , transonic small disturbance equations @xmath247 we discretize these equations using forward finite differences in @xmath237 and @xmath5 , and centered finite differences in @xmath6 .
this leads to the scheme then we compute the ( periodic ) discrete eigenfunctions for the scheme using the ansatz @xmath249 where @xmath250 and @xmath185 are the discrete wave numbers , @xmath251 is the growth factor , @xmath252 and @xmath253 are constants , @xmath254 , @xmath255 , and @xmath256 in the standard fashion of the von neumann stability analysis , this leads to an eigenvalue problem for the vector with components @xmath252 and @xmath253 , with eigenvalue @xmath251 . solving this problem yields @xmath257 note
that , in ( [ eq : von - neumann - growth - factor ] ) , the term @xmath258 can be traced back to the explicit treatment of @xmath195 .
this term causes instability , since it can become arbitrarily large for @xmath259 small and @xmath260 away from zero , independently of the size of @xmath261 .
hence , the scheme is unstable .
instabilities like this one are inevitable in explicit finite difference schemes , regardless of the choice .
the reason is that the wave speeds in the @xmath6-direction are unbounded .
no explicit scheme can hence satisfy the cfl ( courant - friedrichs - lewy ) condition .
d. s. stewart , t. d. aslam , and j. yao . on the evolution of cellular detonation . in _ proceedings of the combustion institute _ ,
volume 26 , pages 29812989 , pittsburgh , pa , 1996 . the combustion institute . | we propose a theory of weakly nonlinear multi - dimensional self sustained detonations based on asymptotic analysis of the reactive compressible navier - stokes equations .
we show that these equations can be reduced to a model consisting of a forced , unsteady , small disturbance , transonic equation and a rate equation for the heat release .
in one spatial dimension , the model simplifies to a forced burgers equation . through analysis , numerical calculations and comparison with the reactive euler equations ,
the model is demonstrated to capture such essential dynamical characteristics of detonations as the steady - state structure , the linear stability spectrum , the period - doubling sequence of bifurcations and chaos in one - dimensional detonations and cellular structures in multi - dimensional detonations . |
we are witnessing an explosion in visual content .
significant recent advances in machine learning and computer vision , especially via deep neural networks , have relied on supervised learning and availability of copious annotated data .
however , manually labelling data is a time - consuming , laborious , and often expensive process . in order to make better use of available unlabeled images , clustering and/or unsupervised learning is a promising direction . in this work
, we aim to address image clustering and representation learning on unlabeled images in a unified framework .
it is a natural idea to leverage cluster ids of images as supervisory signals to learn representations and in turn the representations would be beneficial to image clustering . at a high - level view , given a collection of @xmath0 unlabeled images @xmath1 , the global objective function for learning image representations and clusters can be written as : @xmath2 where @xmath3 is a loss function , @xmath4 denotes the cluster ids for all images , and @xmath5 denotes the parameters for representations .
if we hold one in @xmath6 to be fixed , the optimization can be decomposed into two alternating steps : @xmath7 @xmath8 0.32 0.32 0.32 intuitively , can be cast as a conventional clustering problem based on fixed representations , while is a standard supervised representation learning process . in this paper , we propose an approach that alternates between the two steps updating the cluster ids given the current representation parameters and updating the representation parameters given the current clustering result .
specifically , we cluster images using agglomerative clustering@xcite and represent images via activations of a convolutional neural network ( cnn ) . the reason to choose agglomerative
clustering is three - fold : 1 ) it begins with an over - clustering , which is more reliable in the beginning when a good representation has not yet been learned .
intuitively , clustering with representations from a cnn initialized with random weights are not reliable , but nearest neighbors and over - clusterings are often acceptable ; 2 ) these over - clusterings can be merged as better representations are learned ; 3 ) agglomerative clustering is a recurrent process and can naturally be interpreted in a recurrent framework .
our final algorithm is farily intuitive .
we start with an intial over - clustering , update cnn parameters ( 2b ) using image cluster labels as supervisory signals , then merge clusters ( 2a ) and iterate until we reach a stopping criterion .
an outcome of the proposed framework is illustrated in fig .
[ fig_introduction ] .
initially , there are 1,762 clusters for mnist test set ( 10k samples ) , and the representations ( image intensities ) are not that discriminative . after several iterations
, we obtain 17 clusters and more discriminative representations .
finally , we obtain 10 clusters which are well - separated by the learned representations and interestingly correspond primarily to the groundtruth category labels in the dataset , even though the representation is learnt in an unsupervised manner . to summarize , the major contributions of our work are : we propose a simple but effective end - to - end learning framework to jointly learn deep representations and image clusters from an unlabeled image set ; we formulate the joint learning in a recurrent framework , where merging operations of agglomerative clustering
are expressed as a forward pass , and representation learning of cnn as a backward pass ; we derive _ a single loss function _ to guide agglomerative clustering and deep representation learning , which makes optimization over the two tasks seamless ; our experimental results show that the proposed framework outperforms previous methods on image clustering and learns deep representations that can be transferred to other tasks and datasets .
* clustering * clustering algorithms can be broadly categorized into hierarchical and partitional approaches @xcite .
agglomerative clustering is a hierarchical clustering algorithm that begins with many small clusters , and then merges clusters gradually @xcite . as for partitional clustering methods ,
the most well - known is k - means @xcite , which minimizes the sum of square errors between data points and their nearest cluster centers .
related ideas form the basis of a number of methods , such as expectation maximization ( em ) @xcite , spectral clustering @xcite , and non - negative matrix factorization ( nmf ) based clustering @xcite . * deep representation learning * many works use raw image intensity or hand - crafted features @xcite combined with conventional clustering methods .
recently , representations learned using deep neural networks have presented significant improvements over hand - designed features on many computer vision tasks , such as image classification @xcite , object detection @xcite , etc .
however , these approaches rely on supervised learning with large amounts of labeled data to learn rich representations .
a number of works have focused on learning representations from unlabled image data .
one class of approaches cater to reconstruction tasks , such as auto - encoders @xcite , deep belief networks ( dbn ) @xcite , etc .
another group of techniques learn discriminative representations after fabricating supervisory signals for images , and then finetune them supervisedly for downstream applications @xcite . unlike our approach ,
the fabricated supervisory signal in these previous works is not updated during representation learning .
* combination * a number of works have explored combining image clustering with representation learning . in @xcite , the authors proposed to learn a non - linear embedding of the undirected affinity graph using stacked autoencoder , and then ran k - means in the embedding space to obtain clusters . in @xcite , a deep semi - nmf model was used to factorize the input into multiple stacking factors which are initialized and updated layer by layer . using the representations on the top layer ,
k - means was implemented to get the final results .
unlike our work , they do not jointly optimize for the representation learning and clustering . to connect image clustering and representation learning more closely , @xcite conducted image clustering and codebook learning iteratively .
however , they learned codebook over sift feature @xcite , and did not learn deep representations . instead of using hand - crafted features ,
chen @xcite used dbn to learn representations , and then conducted a nonparametric maximum margin clustering upon the outputs of dbn . afterwards , they fine - tuned the top layer of dbn based on clustering results . a more recent
work on jointly optimizing two tasks is found in @xcite , where the authors trained a task - specific deep architecture for clustering .
the deep architecture is composed of sparse coding modules which can be jointly trained through back propagation from a cluster - oriented loss .
however , they used sparse coding to extract representations for images , while we use a cnn .
instead of fixing the number of clusters to be the number of categories and predicted labels based on softmax outputs , we predict the labels using agglomerative clustering based on the learned representations . in our experiments
we show that our approach outperforms @xcite .
we denote an image set with @xmath0 images by @xmath9 .
the cluster labels for this image set are @xmath10 .
@xmath5 are the cnn parameters , based on which we obtain deep representations @xmath11 from @xmath12 .
given the predicted image cluster labels , we organize them into @xmath13 clusters @xmath14 , where @xmath15 .
@xmath16 are the @xmath17 nearest neighbours of @xmath18 , and @xmath19 is the set of @xmath20 nearest neighbour clusters of @xmath21 . for convenience ,
we sort clusters in @xmath19 in descending order of affinity with @xmath21 so that the nearest neighbour @xmath22 is the first entry @xmath23 $ ] . here
, @xmath24 is a function to measure the affinity ( or similarity ) between two clusters .
we add a superscript @xmath25 to \{@xmath5 , @xmath26 , @xmath4 , @xmath27 } to refer to their states at timestep @xmath25 .
we use @xmath28 to denote the sequence @xmath29 with @xmath30 timesteps . as background
, we first briefly describe conventional agglomerative clustering @xcite .
the core idea in agglomerative clustering is to merge two clusters at each step until some stopping conditions .
mathematically , it tries to find two clusters @xmath31 and @xmath32 by @xmath33 there are many methods to compute the affinity between two clusters @xcite .
more details can be found in @xcite .
we now describe how the affinity is measured by @xmath24 in our approach .
first , we build a directed graph @xmath34 , where @xmath35 is the set of vertices corresponding to deep representations @xmath26 for @xmath12 , and @xmath36 is the set of edges connecting vertices .
we define an affinity matrix @xmath37 corresponding to the edge set . the weight from vertex @xmath18 to @xmath38 is defined by @xmath39 where @xmath40 .
this way to build up a directed graph can be found in many previous works such as @xcite . here ,
@xmath41 and @xmath17 are two predefined parameters ( their values are listed in table [ tb_ac_param ] ) . after constructing a directed graph for samples
, we then adopt the graph degree linkage in @xcite to measure the affinity between cluster @xmath21 and @xmath42 , denoted by @xmath43 .
our key insight is that agglomerative clustering can be interpreted as a recurrent process in the sense that it merges clusters over multiple timesteps .
based on this insight , we propose a recurrent framework to combine the image clustering and representation learning processes . as shown in fig .
[ fig_framework ] , at the timestep @xmath25 , images @xmath12 are first fed into the cnn to get representations @xmath44 and then used in conjunction with previous hidden state @xmath45 to predict current hidden state @xmath46 , i.e , the image cluster labels at timestep @xmath25 . in our context , the output at timestep @xmath25 is @xmath47 .
hence , at timestep @xmath25 @xmath48 @xmath49 @xmath50 where @xmath51 is a function to extract deep representations @xmath44 for input @xmath12 using the cnn parameterized by @xmath52 , and @xmath53 is a merging process for generating @xmath54 based on @xmath44 and @xmath55 . in a typical recurrent neural network
, one would unroll all timesteps at each training iteration . in our case , that would involve performing agglomerative clustering until we obtain the desired number of clusters , and then update the cnn parameters by back - propagation . in this work
, we introduce a _ partial unrolling _ strategy , , we split the overall @xmath30 timesteps into multiple periods , and unroll one period at a time .
the intuitive reason we unroll partially is that the representation of the cnn at the beginning is not reliable .
we need to update cnn parameters to obtain more discriminative representations for the following merging processes . in each period
, we merge a number of clusters and update cnn parameters for a fixed number of iterations at the end of the period .
an extreme case would be one timestep per period , but it involves updating the cnn parameters too frequently and is thus time - consuming .
therefore , the number of timesteps per period ( and thus the number of clusters merged per period ) is determined by a parameter in our approach .
we elaborate on this more in sec .
[ sec_optimization ] . in our recurrent framework
, we accumulate the losses from all timesteps , which is formulated as @xmath56 here , @xmath57 takes each image as a cluster . at timestep @xmath25 , we find two clusters to merge given @xmath58 .
in conventional agglomerative clustering , the two clusters are determined by finding the maximal affinity over all pairs of clusters . in this paper
, we introduce a criterion that considers not only the affinity between two clusters but also the local structure surrounding the clusters .
assume from @xmath58 to @xmath59 , we merged a cluster @xmath60 and its nearest neighbour .
then the loss at timestep @xmath25 is a combination of negative affinities , that is , @xmath61 ) & & & & & & & & & & & & & & \end{aligned } \label{eq_similarity_term_a_timestep_t}\ ] ] @xmath62 ) - \bm{\mathcal{a}}(\mathcal{c}^t_i , \mathcal{n}_{\mathcal{c}^t_i}^{k_c}[k])\right ) \end{aligned } \label{eq_similarity_term_b_timestep_t}\ ] ] where @xmath63 weighs and .
note that @xmath59 , @xmath58 and @xmath52 are not explicitly presented at the right side , but they determine the loss via the image cluster labels and affinities among clusters . on the right side of the above equation , there are two terms : 1 ) measures the affinity between cluster @xmath21 and its nearest neighbour , which follows conventional agglomerative clustering ; 2 ) measures the difference between affinity of @xmath21 to its nearest neighbour cluster and affinities of @xmath21 to its other neighbour clusters .
this term takes the local structure into account .
see sec .
[ sec_forward_pass ] for detailed explanation .
it is hard to simultaneously derive the optimal @xmath29 and @xmath64 that minimize the overall loss in eq . .
as aforementioned , we optimize iteratively in a recurrent process .
we divide @xmath30 timesteps into @xmath65 partially unrolled periods . in each period , we fix @xmath5 and search optimal @xmath4 in the forward pass , and then in the backward pass we derive optimal @xmath5 given the optimal @xmath4 .
details will be explained in the following sections .
in forward pass of the @xmath66-th ( @xmath67 ) partially unrolled period , we update the cluster labels with @xmath5 fixed to @xmath68 , and the overall loss in period @xmath66 is @xmath69 where @xmath70 is the sequence of image labels in period @xmath66 , and @xmath71 $ ] is the corresponding timesteps in period @xmath66 . for optimization , we follow a greedy search similar to conventional agglomerative clustering . starting from the time step @xmath72
, it finds one cluster and its nearest neighbour to merge so that @xmath73 is minimized over all possible cluster pairs . in fig .
[ fig_approach_toyexample ] , we present a toy example to explain the reason why we employ the term .
as shown , it is often the case that the clusters are densely populated in some regions while sparse in some other regions . in conventional agglomerative clustering
, it will choose two clusters with largest affinity ( or smallest loss ) at each time no mater where the clusters are located .
in this specific case , it will choose cluster @xmath32 and its nearest neighbour to merge .
in contrast , as shown in fig . [ fig_approach_toyexample](b ) , our algorithm by adding will find cluster @xmath74 , because it is not only close to it nearest neighbour , but also relatively far away from its other neighbours , i.e. , the local structure is considered around one cluster .
another merit of introducing is that it will allow us to write the loss in terms of triplets as explained next .
0.48 0.48 in forward pass of the @xmath66-th partially unrolled period , we have merged a number of clusters .
let the sequence of optimal image cluster labels be given by @xmath75 , and clusters merged in forward pass are denoted by @xmath76\}$ ] , @xmath77 . in the backward pass ,
we aim to derive the optimal @xmath5 to minimize the losses generated in forward pass .
because the clustering in current period is conditioned on the clustering results of all previous periods , we accumulate the losses of all @xmath66 periods , i.e. , @xmath78 minimizing w.r.t @xmath5 leads to representation learning on @xmath12 supervised by @xmath79 or @xmath80 . based on and , the loss in eq .
[ eq_loss_overall_wst_theta ] is reformulated to @xmath81 ) - \bm{\mathcal{a}}(\mathcal{c}^t _ * , \mathcal{n}_{\mathcal{c}^t_*}^{k_c}[k])\right ) \end{aligned } \label{eq_loss_triplet_time_step}\ ] ] where @xmath82 .
is a loss defined on clusters of points , which needs the entire dataset to estimate , making it difficult to use batch - based optimization .
however , we show that this loss can be approximated by a sample - based loss , enabling us to compute unbiased estimators for the gradients using batch - statistics .
the intuition behind reformulation of the loss is that agglomerative clustering starts with each datapoint as a cluster , and clusters at a higher level in the hierarchy are formed by merging lower level clusters .
thus , affinities between clusters can be expressed in terms of affinities between datapoints .
we show in the supplement that the loss in can be approximately reformulated as @xmath83 where @xmath84 is a weight whose value depends on @xmath85 and how clusters are merged during the forward pass .
@xmath18 and @xmath38 are from the same cluster , while @xmath86 is from the neighbouring clusters , and their cluster labels are merely determined by the final clustering result @xmath87 . to further simplify the optimization , we instead search @xmath86 in at most @xmath20 neighbour samples of @xmath88 from other clusters in a training batch . hence , the batch - wise optimization can be performed using conventional stochastic gradient descent method .
note that such triplet losses have appeared in other works @xcite . because it is associated with a weight
, we call the weighted triplet loss
. + @xmath12 : = collection of image data ; + @xmath89 : = target number of clusters ; + + @xmath90 : = final image labels and cnn parameters ; + @xmath91 ; @xmath92 initialize @xmath5 and @xmath4 update @xmath59 to @xmath93 by merging two clusters @xmath94 @xmath95 cluster number reaches @xmath96 @xmath97 ; @xmath98 [ alg_optimization ] given an image dataset with @xmath0 samples , we assume the number of desired clusters @xmath96 is given to us as is standard in clustering
. then we can build up a recurrent process with @xmath99 timesteps , starting by regarding each sample as a cluster .
however , such initialization makes the optimization time - consuming , especially when datasets contain a large number of samples . to address this problem
, we can first run a fast clustering algorithm to get the initial clusters . here , we adopt the initialization algorithm proposed in @xcite for fair comparison with their experiment results .
note that other kind of initializations can also be used , e.g. k - means .
based on the algorithm in @xcite , we obtain a number of clusters which contain a few samples for each ( average is about 4 in our experiments ) .
given these initial clusters , our optimization algorithm learns deep representations and clusters .
the algorithm is outlined in alg .
[ alg_optimization ] . in each partially unrolled period , we perform forward and backward passes to update @xmath4 and @xmath5 , respectively . specifically , in the forward pass , we merge two clusters at each timestep . in the backward pass ,
we run about 20 epochs to update @xmath5 , and the affinity matrix @xmath100 is also updated based on the new representation .
the duration of the @xmath66-th period is @xmath101 timesteps , where @xmath102 is the number of clusters at the beginning of current period , and @xmath103 is a parameter called _ unrolling rate _ to control the number of timesteps .
the less @xmath103 is , the more frequently we update @xmath5 .
we compare our approach with 12 clustering algorithms , including k - means @xcite , njw spectral clustering ( sc - njw ) @xcite , self - tuning spectral clustering ( sc - st)@xcite , large - scale spectral clustering ( sc - ls ) @xcite , agglomerative clustering with average linkage ( ac - link)@xcite , zeta function based agglomerative clustering ( ac - zell ) @xcite , graph degree linkage - based agglomerative clustering ( ac - gdl ) @xcite , agglomerative clustering via path integral ( ac - pic ) @xcite , normalized cuts ( n - cuts ) @xcite , locality preserving non - negative matrix factorization ( nmf - lp ) @xcite , nmf with deep model ( nmf - d ) @xcite , task - specific clustering with deep model ( tsc - d ) @xcite .
we show the first three principle components of learned representations in fig .
[ fig_pca_display_5 ] and fig .
[ fig_pca_display_4 ] at different stages . for comparison , we show the image intensities at the first column .
we use different colors for representing different clusters that we predict during the algorithm . at the bottom of each plot
, we give the number of clusters at the corresponding stage . at the final stage ,
the number of cluster is same to the number of categories in the dataset . after a number of iterations
, we can learn more discriminative representations for the datasets , and thus facilitate more precise clustering results . | in this paper , we propose a recurrent framework for * * j**oint * * u**nsupervised * * le**arning ( * jule * ) of deep representations and image clusters . in our framework ,
successive operations in a clustering algorithm are expressed as _ steps in a recurrent process _ , stacked on top of representations output by a convolutional neural network ( cnn ) . during training ,
image clusters and representations are updated jointly : image clustering is conducted in the forward pass , while representation learning in the backward pass .
our key idea behind this framework is that good representations are beneficial to image clustering and clustering results provide supervisory signals to representation learning . by integrating two processes into a single model with a unified weighted triplet loss and optimizing it
end - to - end , we can obtain not only more powerful representations , but also more precise image clusters .
extensive experiments show that our method outperforms the state - of - the - art on image clustering across a variety of image datasets .
moreover , the learned representations generalize well when transferred to other tasks .
the source code can be downloaded from https://github.com/jwyang/joint-unsupervised-learning . |
figures [ fig : power]a and b show the temperature dependence of the resonance frequency and quality factor .
these values are measured at large bias currents where the backaction is negligible .
the frequency change due to temperature is small compared to the observed backaction ( see main text ) .
the intrinsic quality factor decreases significantly with increasing temperature .
this rules out that the observed frequency shift and q - factor change are caused by heating of the resonator due to joule heating in the junctions : we observe an _ increase _ in quality factor with increasing bias current and voltage setpoint ( i.e. dissipation ) , but a _ decrease _ in quality factor with increasing temperature .
an increased damping at higher temperatures is seen more often in micro- or nanomechanical resonators @xcite . the observed frequency shift and change in damping
do not depend on the driving power .
as shown in figure [ fig : power]c , the measured resonator response stays the same in all panels .
although the driving power is changed by three orders of magnitude , the only effect is that the signal - to - noise ratio becomes better when increasing the power . for the highest driving power ( @xmath76 ) and highest q - factor ( @xmath77 ) the amplitude of the resonator motion is @xmath78 , as determined using the calibrated displacement responsivity @xcite . the flux through the dc squid is then modulated with an amplitude @xmath79 .
so even for the largest resonator motion , the change in flux is much smaller than a flux quantum .
exactly on resonance , the piezo motion is @xmath18 times smaller than @xmath80 , about @xmath81 .
the driving force @xmath30 is then given by the resonator mass @xmath27 times the acceleration of the piezo element , @xmath82 .
the measurements shown in the main text are done with a driving power of @xmath83 ( @xmath84 ) .
( a ) and resonator frequency @xmath58 ( b ) .
( c ) colorscale plot of the oscillator response at @xmath10 at different driving powers @xmath85 .
this measurement is done with a voltage setpoint halfway between @xmath6 and @xmath7 .
[ fig : power],width=325 ]
all effects that we have observed in the device that we discuss in the main text have also been measured in a second device .
the resonator in this dc squid operates around @xmath86 .
the maximum critical current of device b is @xmath87 at @xmath88 and @xmath89 at @xmath90 . in the measurements on device
b , the feedback loop could not maintain the squid voltage for low voltage setpoints .
also , a less sensitive , room temperature amplifier was used for the resonator signal .
its lower gain resulted in an increased scatter in the data and the inability to explore the region with the highest backaction , i.e. , at low bias current and low voltage setpoints .
the measurements in fig .
[ fig : devb ] show qualitatively the same backaction as the data presented in the main text .
( circles ) and @xmath91 ( triangles ) .
the inset depicts the voltage setpoint for these measurements with respect to @xmath6 and @xmath7 .
bias current and setpoint dependence of the frequency shift ( c ) and damping ( d ) at @xmath92 mt .
the solid lines indicate the measured @xmath6 and @xmath7 .
[ fig : devb],width=325 ]
to calculate the backaction , the dc squid is modelled using the resistively- and capacitively - shunted junction ( rcsj ) model .
this widely - used model is discussed in detail in ref .
the introduction to this model presented here is largely based on this review .
a current @xmath1 is sent through the squid and the circulating current @xmath25 redistributes the current over the two junctions , which we assume to be identical . in the rcsj model , the two junctions ( labelled with @xmath93 ) are modelled as a resistor ( @xmath46 ) , capacitor ( @xmath94 ) and an `` ideal '' josephson junction with critical current @xmath95 in parallel .
the voltage over each junction is related to the time derivative of the phase difference @xmath96 of the superconducting wave function : @xmath97 , where @xmath98 is the flux quantum .
current conservation yields two second - order differential equations , governing the time - dependence of the phase differences @xmath99 of the junctions : @xmath100 these equations are coupled to each other by the amount of flux piercing the loop : @xmath101 the total flux @xmath102 has two contributions : the externally applied flux @xmath2 ( which also includes the flux due to the resonator displacement ) and the flux due to the circulating current flowing through the inductance of the loop @xmath103 , i.e. , @xmath104 .
the equations are scaled to simplify their numerical integration .
this yields : @xmath105 the bias current and circulating current are normalized using the critical current : @xmath106 and @xmath107 .
furthermore , time is scaled using the characteristic frequency @xmath108 , fluxes using the flux quantum , i.e. , @xmath109 , and voltages using the characteristic voltage @xmath110 so that @xmath111
. the parameter @xmath112 and @xmath113 are the inductive screening parameter and the stewart - mccumber number respectively .
the inductive screening parameter indicates how much a change in flux is screened by the circulating current @xmath25 flowing through the self - inductance of the loop @xmath103 , whereas @xmath114 indicates the importance of inertial terms due to the junction capacitance @xmath94 .
the three equations are integrated numerically for different bias conditions , i.e. , different values for the bias current @xmath1 and for the flux @xmath2 through the squid .
figure [ fig : simulation]a shows typical examples of calculated time - traces of the circulating current @xmath115 and voltage @xmath116 .
both are rapidly oscillating at a frequency of @xmath117 , which is the josephson frequency that equals the average value of the voltage @xmath118 @xcite .
the maximum current that the dc squid can carry without generating a voltage equals the sum of the critical currents of the two junctions : @xmath119 . and
voltage @xmath120 in the absence of noise .
( b ) circulating current for a small ( @xmath121 ) modulation of the flux with @xmath122 .
the time - averaged value of the circulating current @xmath123 has a phase shift with respect to the modulation @xmath124 as indicated by the orange lines .
( c ) absolute value of the fourier transform of the timetraces of the squid voltage and circulating current shown in ( b ) .
( d ) the modulation frequency dependence of the real ( dark gray ) and imaginary ( light gray ) parts of the transfer function .
the lines are a guide to the eye .
these simulations were done for the dc squid parameters from the main text ( @xmath74 and @xmath75 ) at @xmath125 and @xmath126 .
[ fig : simulation ] ] the critical current ( @xmath127 ) and the normal - state resistance ( @xmath128 ) of the junctions are estimated from the iv - characteristics ( fig .
1b of the main text ) . using finite - element simulations we estimate @xmath129 for our device .
finally , the capacitance @xmath130 is obtained from the position of the lc resonance in the dc squid @xcite .
when the resonator moves , the flux through the dc squid loop is altered , which in turn changes the average circulating current @xmath25 . in principle , @xmath131 could depend on the all the past displacements , @xmath132 for @xmath133 .
however , the dc squid reacts at a frequency ( @xmath134 ) that is much faster than that of the resonator ( @xmath135 ) , so the circulating current @xmath131 is expected to depends on the _ instantaneous _ displacement @xmath136 . for small amplitudes ( @xmath137 ) the response is linear and gives a contribution @xmath138 .
another contribution comes from the velocity of the resonator , @xmath139 , which causes a time - varying flux .
this generates an electromotive force in the squid loop ( faraday s induction law ) , which also changes the circulating current by @xmath140 .
combining these two effects gives @xmath141 .
the values of the parameters @xmath142 and @xmath143 depend on the dynamics of the dc squid and are @xmath144 and @xmath145 .
as discussed in the main text , these quantities are related to the intrinsic flux - to - current transfer functions , @xmath146 and @xmath147 . in principle @xmath148
can be obtained by calculating the average circulating current at different fluxes and then numerically differentiating this to obtain @xmath149 . however , for the velocity - dependent transfer function this is not possible .
our method for calculating these transfer functions @xmath52 and @xmath53 works as follows : we calculate the steady - state response of the circulating current with a small modulation added to the applied flux , @xmath150 . figure [ fig : simulation]b shows that this modulates the circulating current @xmath151 .
figure [ fig : simulation]c shows the fourier transform of the circulating current and the voltage . in the spectrum of both @xmath115 and @xmath116
a peak appears at the modulation frequency .
the real part of the peak corresponds to the derivative @xmath152 $ ] , while @xmath153 = - \omega_\mathrm{mod } \jmath_{\dot \phi}$ ] and similar for @xmath154 . the frequency dependence in fig [ fig : simulation]d , shows a constant @xmath155 $ ] and a linearly increasing @xmath156 $ ] as indicated by the dotted lines .
the transfer functions do not depend on the modulation frequency , provided that it is sufficiently low .
this confirms that the circulating current only depends on the instantaneous displacement and velocity as was postulated at the beginning of this section . | we have measured the backaction of a dc superconducting quantum interference device ( squid ) position detector on an integrated 1 mhz flexural resonator .
the frequency and quality factor of the micromechanical resonator can be tuned with bias current and applied magnetic flux .
the backaction is caused by the lorentz force due to the change in circulating current when the resonator displaces .
the experimental features are reproduced by numerical calculations using the resistively and capacitively shunted junction ( rcsj ) model .
it has recently been demonstrated that a macroscopic mechanical resonator can be put in a quantum state @xcite by coupling it to another quantum system . at the same time , linear detectors coupled to mechanical resonators are rapidly approaching the quantum limit on position detection .
this limit implies that the resonator position can not be measured with arbitrary accuracy , as the detector itself affects the resonator position @xcite .
this is an example of backaction .
backaction does not just impose limits , it can also work to one s advantage : backaction can cool the resonator , squeeze its motion , and couple and synchronize multiple resonators .
different backaction mechanisms have been identified : when using optical interferometers @xcite or electronic resonant circuits @xcite , backaction results from radiation pressure . in single - electron transistors ( set )
@xcite , cooper - pair boxes @xcite , carbon nanotube quantum dots @xcite , or atomic and quantum point contacts @xcite backaction is due to the tunneling of electrons .
recently , we have used a dc squid as a sensitive detector of the position of an integrated mechanical resonator @xcite .
this embedded resonator - squid geometry enables the experimental realization of a growing number of theoretical proposals for which a good understanding of the backaction is required @xcite .
transduces a beam displacement @xmath0 into a change in magnetic flux .
a bias current @xmath1 is sent through the squid and its output voltage is measured .
the flux @xmath2 is fine - tuned with a stripline current @xmath3 that is controlled by a feedback circuit ( dashed ) that keeps the output voltage @xmath4 at @xmath5 .
( b ) the bias current dependence of the measured @xmath6 and @xmath7 .
( c ) the amplitude ( bottom ) and phase ( top ) response .
the line is a fitted harmonic oscillator response @xcite .
[ fig : overview ] ] in this letter , we present experiments that show that the dc squid detector exerts backaction on the resonator . by adjusting the bias conditions of the dc squid the frequency and damping of the mechanical resonator change .
the backaction by the dc squid has a different origin than in the experiments mentioned above : it is due to the lorentz force generated by the circulating current .
numerical calculations using the rcsj model for the dc squid @xcite reproduce the experimental features .
the device ( fig .
[ fig : overview]a ) consists of a dc squid with proximity - effect - based junctions @xcite .
a part of one arm is underetched , forming a @xmath8 flexural resonator with length @xmath9 . in this letter
we present data on a device in an in - plane magnetic field of @xmath10 .
measurements have been performed at several magnetic fields and on an additional device ; the observed backaction is similar @xcite .
first the dc squid is characterized .
the output voltage of a dc squid depends on the magnetic flux through its loop @xmath2 @xcite and we measure the minimum and maximum voltage ( @xmath6 and @xmath11 ) by sweeping the flux over a few flux quanta @xmath12 with a nearby stripline ( fig .
1a ) .
this is repeated for different bias currents to obtain the current - voltage curves shown in fig . [
fig : overview]b .
the maximum critical current is @xmath13 and the normal - state resistance of the junctions is @xmath14 .
after this characterization , the dc squid is operated at a given setpoint voltage @xmath5 using a feedback loop that adjusts the flux via the stripline current @xcite .
the feedback loop is used to reduce low - frequency flux noise and flux drift and has a bandwidth of @xmath15 , i.e. , it does not respond to the @xmath8 resonator signal .
the fundamental mode of the flexural resonator is excited using a piezo element underneath the sample and the displacement of the beam is detected as follows : the in - plane magnetic field transduces a displacement of the beam @xmath0 into a flux change @xmath16 , which in turn changes the voltage over the dc squid .
this voltage is amplified using a cryogenic high - electron mobility transistor followed by a room temperature amplifier and then recorded using a network analyzer .
figure [ fig : overview]c shows the amplitude and phase of the measured response , from which the resonance frequency @xmath17 and quality factor @xmath18 are obtained . to observe backaction of the dc squid detector on the resonator
, the frequency response is measured for different bias conditions of the squid .
figure [ fig : backaction_traces ] shows that both @xmath17 and @xmath18 depend on the bias current @xmath1 .
the resonance frequency saturates at @xmath19 for large positive and negative bias currents . however , when decreasing the @xmath1 , @xmath17 first goes up by a few hundred hz around @xmath20 and then it decreases rapidly with about -2000 hz at the lowest stable setpoint voltage .
this is more than @xmath21 the linewidth @xmath22 of the resonance shown in fig .
[ fig : overview]c .
figure [ fig : backaction_traces]b shows that the quality factor of the resonator changes from @xmath23 to less than 2000 .
similar to the resonance frequency , first an increase and then a stronger decrease in @xmath18 is observed when lowering the bias current . and
@xmath7 in these measurements .
[ fig : backaction_traces ] ] figure [ fig : backaction_maps ] shows that the frequency and damping can also be changed by adjusting @xmath5 , i.e. , the flux through the squid loop .
the shifts are largest for low setpoints and low bias currents ( dark regions ) .
the regions with a lower frequency coincide with the regions where the damping has increased .
bias points with positive frequency shifts and increases in @xmath18 are indicated in white .
finally , by varying the driving power we confirm that the observed effects are not due to nonlinearities in the squid or in the resonator @xcite . ] unlike for position detectors such as the set where the backaction originates from the coulomb force , the backward coupling between the squid and the beam is the lorentz force @xmath24 @xcite .
this force is due to the current that flows through the beam in the presence of the magnetic field that couples the resonator and the squid .
a displacement changes the flux , and this in turn changes the circulating current in the loop @xmath25 @xcite , giving a different force on the beam .
in addition , resonator motion yields a time - varying flux through the loop , which induces an electro - motive force and thereby also generates currents that change the lorentz force @xcite .
the displacement of the fundamental out - of - plane flexural mode @xmath0 is given by @xcite : @xmath26 the resonator has a mass @xmath27 , ( intrinsic ) frequency @xmath28 and quality factor @xmath29 .
@xmath30 is the driving force and @xmath31 is the lorentz force . here , @xmath32 for the fundamental mode , so that also @xmath33 @xcite . for small amplitudes and low resonator frequencies ( much smaller than the characteristic squid frequency @xmath34 ) ,
the average circulating current can be expanded in the displacement and velocity @xmath35 @xcite : @xmath36 inserting eq .
( [ eq : jtransfer ] ) into eq .
( [ eq : ho ] ) shows that the @xmath37 term affects the spring constant @xmath38 and thus @xmath17 , whereas the @xmath39 term renormalizes the damping .
the shifted resonance frequency and quality factor are : @xmath40 here , @xmath41 and @xmath42 are the scaled flux - to - current transfer functions @xcite .
the former indicates how much the circulating current changes when the flux through the ring is altered , whereas the latter quantifies the effect of a time - dependent flux on the circulating current .
these functions were first studied in the analysis of the dynamic input impedance of tuned squid amplifiers @xcite and are intrinsic properties of the squid . before looking in more detail at the transfer functions , we first focus on the coupling . the dimensionless parameters @xmath43 @xcite and @xmath44 characterize the backaction strength .
they contain the term @xmath45 squared as both the flux change and the lorentz force are proportional to the magnetic field .
this implies that the backaction remains the same when the direction of the magnetic field is reversed and this is what we observe experimentally .
@xmath43 is proportional to @xmath20 , whereas the damping induced by the squid depends on @xmath46 . by a careful design of the resonator and squid ,
the backaction strengths can be tuned over a wide range .
eqs .
( [ eq : fr ] ) and ( [ eq : q ] ) show that the largest backaction occurs for large flux changes @xmath45 , low spring constants @xmath47 , and large circulating currents , i.e. , large @xmath48 and low @xmath46 . for the device studied in this letter
, we estimate @xmath49 and @xmath50 . finally , note that the two coupling parameters are related by @xmath51 .
so far , the analysis did not assume anything about the number of junctions , nor about their microscopic details . to obtain the transfer functions @xmath52 and @xmath53
, we model the junctions in the dc squid using the rcsj model @xcite .
the transfer functions can be calculated analytically in certain limits @xcite . however , to obtain their full bias - condition dependence , @xmath52 and @xmath53 must be calculated numerically .
this is done by simulating the dynamics of the squid in the presence of a time - varying flux @xcite .
figure [ fig : transfer_function]a shows the bias - dependence of @xmath52 . in the region where @xmath54 , the circulating current redistributes the bias current between the two junctions such that no voltage develops . here
the circulating current is of the order of @xmath55 , which gives @xmath56 ( blue ) . in the dissipative region ( @xmath57 ) , the circulating current is suppressed .
therefore , the circulating current changes rapidly close to the edge of the dissipative region . the orange color in fig . [ fig : transfer_function]a indicates that @xmath52 is large and positive near the critical current .
the largest downward frequency shift is expected near a half - integer number of flux quanta , whereas @xmath52 vanishes for integer flux . with the value of the coupling parameter @xmath43 and the resonance frequency @xmath58 the frequency shift is calculated as shown in fig .
[ fig : transfer_function]c .
the maximum value @xmath59 gives a frequency shift of @xmath60 in the lower - left corner , which has to be compared with the experimental value of @xmath61 .
increasing the bias current above the critical current results in a smaller @xmath52 ( light yellow and light blue ) that depends linearly on the inductive screening parameter @xmath62 @xcite for the experimental conditions . in this region
the simulations predict both positive ( blue ) and negative ( yellow ) value for @xmath52 .
positive and negative shifts are also observed in the experiment ( figs .
[ fig : backaction_traces ] and [ fig : backaction_maps ] ) .
the largest negative value found in the simulations of that region is @xmath63 , which results in an increase in @xmath17 of @xmath64 , which is in agreement with the observed value of @xmath65 ( fig . [
fig : backaction_traces]a ) . for even larger bias currents ,
the frequency shift vanishes @xmath66 .
this corresponds to the flat regions in fig .
[ fig : backaction_traces]a . the change in damping is determined by @xmath53 as indicated by eq .
( [ eq : q ] ) .
its dependence on the bias conditions is shown in fig .
[ fig : transfer_function]b . well inside the experimentally inaccessible non - dissipative region @xmath67 and the backaction results in a small increase in @xmath18 . in the opposite limit of large bias currents
@xmath68 ( light blue ) .
this value combined with the small value of @xmath44 implies that the quality factor in the flat region in fig .
[ fig : backaction_traces]b is close to the intrinsic q - factor , @xmath29 . in this region
the small additional damping is due to the current induced by the time - varying flux @xmath69 , which is dissipated in the junction resistances @xcite .
this contribution is well - known from magnetomotive readout of mechanical resonators @xcite .
when lowering the bias current , @xmath53 changes sign and rises to about @xmath70 .
this reduces the damping and might even lead to instability ( @xmath71 ) if @xmath44 is large enough .
this decrease of damping corresponds to the bumps in fig .
[ fig : backaction_traces]b .
the largest observed quality factor @xmath72 corresponds to @xmath73 , which is in reasonable agreement with the simulations .
close to the critical current , @xmath53 goes to large _ negative _ values leading to an enhanced dissipation . figure [ fig : transfer_function]d shows that the calculated q - factor is indeed lowest near the critical current . in summary , our model shows that although the coupling strength is small , the dynamics of the dc squid greatly enhances the backaction .
( a ) and @xmath53 ( b ) . with the values for @xmath43 and @xmath44 , the frequency shift ( c ) and quality factor change ( d ) are calculated .
the simulation is done for the experimental conditions where the inductive screening parameter @xmath74 and the stewart - mccumber parameter @xmath75 @xcite .
[ fig : transfer_function ] ] various interesting effects can be observed when the backaction is strong .
if the resonator and squid are strongly coupled , the resonator temperature is set by the effective bath temperature @xcite of the squid .
the increased damping cools the resonator , but the shot noise in the bias current leads to an increase in the force noise on the resonator , heating it .
the question whether the resonator temperature is above or below the environmental temperature should be addressed in future research .
furthermore , the dependence of @xmath17 and @xmath18 on the bias conditions allows parametric excitation of the mechanical resonator by either modulating the flux or the bias current .
this enables squeezing of the thermomechanical noise of the resonator @xcite .
finally , if the squid contains multiple , nearly identical mechanical resonators , these are coupled to each other by the backaction .
this , in turn , can synchronize their motion and might lead to frequency entrainment if higher order terms in eq .
( [ eq : jtransfer ] ) become significant @xcite .
these examples are only a few intriguing possibilities of the rich physics connected to the backaction that we have described in this letter .
we thank t. akazaki , h. okamoto and k. yamazaki for help with fabrication and r. schouten for the measurement electronics .
this work was supported in part by jsps kakenhi ( 20246064 and 18201018 ) , fom , nwo ( vici grant ) , and a eu fp7 strep project ( qnems ) .
33 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) .
, * * , ( pages ) ( ) .
, * * , ( pages ) ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( pages ) ( ) . , * * , ( ) . , * * , ( ) . , * * , ( pages ) ( ) . , * * , ( pages ) ( ) . , * * , ( pages ) ( ) . ,
* * , ( pages ) ( ) .
, , , * * , ( pages ) ( ) . ,
* * , ( pages ) ( ) . , * * , ( ) . , _ _ ( , ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) .
, , , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( pages ) ( ) . |
due to their importance in theory and applications , fine potential theoretical properties of lvy processes have been under intense study recently .
the transition density @xmath8 of a lvy process is the heat kernel of the generator of the process .
however , the transition density ( if it exists ) of a general lvy process rarely admits an explicit expression . thus obtaining sharp estimates on @xmath8
is a fundamental problem both in probability theory and in analysis .
the generator of a discontinuous lvy process is an integro - differential operator and so it is a non - local operator .
recently , quite a few people in pde are interested in problems related to non - local operators ; see , for example , @xcite and the references therein .
when @xmath6 is a symmetric diffusion on @xmath7 whose infinitesimal generator is a uniformly elliptic and bounded divergence form operator , it is well - known that @xmath8 enjoys the celebrated aronson s gaussian type estimates .
when @xmath6 is a pure jump symmetric process on @xmath7 , sharp estimates on @xmath8 have been studied in @xcite recently , which can be viewed as the counterpart of aronson s estimates for non - local operators . due to the complication near the boundary
, two - sided estimates for the transition densities of discontinuous lvy processes killed upon leaving an open set @xmath0 ( equivalently , the dirichlet heat kernels ) have been established very recently for a few particular processes only .
the first of such estimates is obtained in @xcite , where we succeeded in establishing sharp two - sided estimates for the heat kernel of the fractional laplacian @xmath9 with zero exterior condition on @xmath10 ( or equivalently , the transition density of the killed symmetric @xmath11-stable process ) in any @xmath2 open set @xmath0 .
the approach developed in @xcite provides a road map for establishing sharp two - sided heat kernel estimates of other jump processes in open subsets of @xmath7 .
the ideas of @xcite were adapted to establish sharp two - sided heat kernel estimates of relativistic stable processes and mixed stable processes in @xmath2 open subsets of @xmath7 in @xcite respectively . in all these cases ,
the characteristic exponents of these lvy processes admit explicit expressions , the boundary decay rates of the dirichlet heat kernels are suitable powers of the distance to the boundary . on the other hand , a varopoulos type two - sided dirichlet heat kernel estimate of symmetric stable processes in @xmath1-fat open sets was derived in @xcite ; this type of estimates is expressed in terms of surviving probabilities and the global transition density of the symmetric stable process .
the objective of this paper is to establish sharp two - sided estimates on the transition density @xmath12 for a large class of purely discontinuous rotationally symmetric lvy processes . unlike the cases considered in @xcite , the characteristic exponents of the symmetric lvy processes considered in this paper are quite general , satisfying only certain mild growth condition at @xmath13 .
moreover , the boundary decay rate of @xmath12 is no longer some power of the distance to the boundary .
the analysis of the precise boundary behavior of @xmath12 is quite challenging and delicate .
the main tools to obtain the precise boundary behavior of @xmath12 are two versions of the boundary harnack principle obtained in @xcite . in this paper
we combine the approaches developed in @xcite with these boundary harnack principles to obtain sharp two - sided estimates for @xmath12 , which cover the main results in @xcite and much more .
suppose that @xmath14 is a subordinator with laplace exponent @xmath4 , that is , @xmath15 is a nonnegative lvy process with @xmath16 and @xmath17=e^{-t\phi(\lambda)}$ ] for every @xmath18 .
the function @xmath4 can be written in the form @xmath19 where @xmath20 and @xmath21 is a measure on @xmath22 satisfying @xmath23 .
the constant @xmath24 is called the drift of the subordinator and @xmath21 the lvy measure of the subordinator ( or of @xmath4 ) .
the function @xmath4 is a bernstein function , i.e. , it is @xmath25 , positive and @xmath26 for all @xmath27 . in particular , since @xmath28 and @xmath29 , the bernstein function @xmath4 has the property that @xmath30 the laplace exponent @xmath4 is said to be a complete bernstein function if the lvy measure @xmath21 of @xmath4 has a completely monotone density @xmath31 , i.e. , @xmath32 for every non - negative integer @xmath33 . for basic results on complete bernstein functions , we refer the reader to @xcite . throughout this paper ,
we assume that @xmath4 is a complete bernstein function satisfying the following growth condition at infinity ( see @xcite ) : * ( a ) : * there exist constants @xmath34 , @xmath35 , @xmath36 and @xmath37 such that @xmath38 note that it follows from the upper bound condition in * ( a ) * that @xmath4 has no drift .
let @xmath39 be a brownian motion in @xmath7 independent of the subordinator @xmath15 .
the subordinate brownian motion @xmath40 is defined by @xmath41 , which is a rotationally symmetric lvy process with lvy exponent @xmath42 .
the infinitesimal generator of @xmath43 is @xmath44 . here and below for a function @xmath45 on @xmath46 , @xmath47 is defined as a pseudo differential operator in terms of fourier transform ; that is , @xmath48 , where @xmath49 is the fourier transform of a function @xmath50 on @xmath7 .
it is known that the lvy measure of the process @xmath43 has a density given by @xmath51 where @xmath52 note that the function @xmath53 is continuous and decreasing on @xmath22 .
we will assume that @xmath6 is a purely discontinuous rotationally symmetric lvy process with lvy exponent @xmath54 .
because of rotational symmetry , the function @xmath5 depends on @xmath55 only , and by a slight abuse of notation we write @xmath56 .
the infinitesimal generator of @xmath6 is @xmath57 .
we further assume that the lvy measure of @xmath6 has a density with respect to the lebesgue measure on @xmath7 , which is denoted by @xmath58 .
that is , @xmath59=e^{-t\psi(|\xi| ) } \quad \quad \mbox { for every } x\in \r^d \mbox { and } \xi\in \r^d,\ ] ] with @xmath60 we assume that @xmath61 is continuous on @xmath22 and that there is a constant @xmath62 such that @xmath63 this implies that @xmath64 and @xmath65 are comparable . clearly also implies that @xmath66 we remark that under the above assumptions , @xmath6 does not need to be a subordinate brownian motion because @xmath67 does not need to be monotone .
for example , choose @xmath68 such that @xmath69 and a continuous function @xmath70 with @xmath71 , @xmath72 and @xmath73 for all @xmath74 .
then @xmath75 is not monotone and its corresponding lvy process @xmath6 ( through lvy exponent ) is not a subordinate brownian motion . under the above setup
, @xmath6 has a continuous transition density @xmath8 with respect to the lebesgue measure on @xmath7 ( see @xcite ) .
clearly , @xmath8 is a function depending only on @xmath76 and @xmath77 , and so , by an abuse of notation , we also denote @xmath8 by @xmath78 . for every open subset @xmath79 ,
we denote by @xmath80 the subprocess of @xmath6 killed upon exiting @xmath0 .
it is known ( see @xcite ) that @xmath80 has a transition density @xmath12 , with respect to the lebesgue measure , which is jointly locally hlder continuous .
note that @xmath81 is the fundamental solution for @xmath82 in @xmath0 with zero exterior condition and so it can also be called the dirichlet heat kernel of @xmath83 in @xmath0 .
the purpose of this paper is to establish sharp two - sided estimates on @xmath12 .
the following two conditions will be needed for some of the results in this paper when @xmath0 is unbounded . * ( b ) * : there exist constants @xmath84 and @xmath85 $ ] such that @xmath86 \hbox { and } u\geq r>0.\ ] ] * ( c ) * : there exist constants @xmath87 and @xmath88 $ ] such that @xmath89 \hbox { and } r>0.\ ] ] throughout this paper we will use @xmath90 to denote the function [ e:1.9 ] ( r)=1(r^-2 ) , r>0 . note that in particular it follows from that [ e:1.8 ] ( 2r)= = 4(r ) r>0 .
the inverse function of @xmath90 will be denoted by the usual notation @xmath91 . here and in the following , for @xmath92 , @xmath93 and @xmath94 .
[ r:1.1 ] \(i ) the condition * ( b ) * is pretty mild .
when @xmath6 is a rotationally symmetric lvy process such that @xmath95 is decreasing , condition * ( b ) * holds for all @xmath96 ( see ( * ? ? ?
* proposition ) ) .
in particular it holds for all subordinate brownian motions with @xmath97 . in this special case , we can also see this using the following elementary argument : when @xmath6 is a subordinate brownian motion , @xmath98 $ ] , where @xmath99 is the transition density of the brownian motion @xmath100 .
it follows immediately that @xmath101 is decreasing in @xmath102 and so * ( b ) * holds with @xmath97 .
\(ii ) under condition * ( a ) * , condition * ( b ) * is weaker than * ( c)*. under condition * ( a ) * , we will show in this paper that there exists @xmath103 such that @xmath104 \hbox { and } r>0\ ] ] ( see proposition [ p : new1 ] and theorem [ t : globalhke ] below ) .
thus condition * ( c ) * amounts to say that there exist constants @xmath105 and @xmath106 $ ] such that for @xmath107 $ ] and @xmath74 , @xmath108 since @xmath109 is a decreasing function in @xmath102 , implies that condition * ( b ) * holds with @xmath110 and @xmath111 .
\(iii ) assume that condition * ( a ) * holds .
it follows from @xcite that , for every @xmath112 , there is a constant @xmath113 so that holds for @xmath114\times ( 0 , r]$ ] with @xmath115 .
( see proposition [ p : new1 ] below . )
so the assertions in conditions * ( b ) * and * ( c ) * are always satisfied for @xmath116 .
\(iv ) by @xcite , @xmath117\times ( 0 , \infty),\ ] ] and consequently conditions * ( b ) * and * ( c ) * , hold for a large class of discontinuous processes including mixed stable - like processes ( with @xmath115 ) and relativistic stable - like processes ( with @xmath115 , see ( * ? ? ?
* theorem 4.1 ) ) . before stating the main results of this paper , we need first to set up some notations .
let @xmath118 .
we denote the euclidean distance between @xmath119 and @xmath120 in @xmath7 by @xmath77 and denote by @xmath121 the open ball centered at @xmath122 with radius @xmath74 ; for any two positive functions @xmath50 and @xmath123 , @xmath124 means that there is a positive constant @xmath105 so that @xmath125 on their common domain of definition ; for any open @xmath126 and @xmath127 , @xmath128 stands for the diameter of @xmath0 and @xmath129 stands for the euclidean distance between @xmath119 and @xmath10 .
[ def : ub ] let @xmath130 .
we say that a open set @xmath0 is @xmath1-fat if there is @xmath131 such that for all @xmath132 and all @xmath133 $ ] , there is a ball @xmath134 .
the pair @xmath135 is called the characteristics of the @xmath1-fat open set @xmath0 .
the following factorization of the dirichlet heat kernel is the first main result of this paper .
recall that @xmath136 and @xmath137 are the constants in condition * ( b)*. [ thm : oppz ] let @xmath6 be a purely discontinuous rotationally symmetric lvy process with lvy exponent @xmath5 and lvy density @xmath65 satisfying and respectively , where the complete bernstein function @xmath4 satisfies * ( a)*. suppose that @xmath0 is a @xmath1-fat open set with characteristics @xmath135 .
\(i ) for every @xmath138 , there exists @xmath139 such that for @xmath140 , @xmath141 , @xmath142 \(ii ) if @xmath0 is unbounded , we assume in addition that condition * ( b ) * holds . for every @xmath138
, there exists @xmath143 such that for @xmath140 , @xmath141 , @xmath144 where @xmath145 .
\(iii ) suppose in addition that @xmath0 is bounded .
then there exists @xmath146 so that for all @xmath147 , @xmath148 where @xmath149 is the largest eigenvalue of the generator of @xmath80 . when @xmath6 is a rotationally symmetric @xmath11-stable process in @xmath7 , that is , when @xmath150 for some @xmath151 , parts ( i ) and ( ii ) of theorem [ thm : oppz ] are proved in @xcite .
recall that @xmath152 and @xmath153 are the constants in condition * ( c)*. combining theorem [ thm : oppz](i)(ii ) and remark [ r:1.1](ii ) , we have the following corollary .
[ c : main1 ] let @xmath6 be a purely discontinuous rotationally symmetric lvy process with lvy exponent @xmath5 and lvy density @xmath65 satisfying and respectively , where the complete bernstein function @xmath4 satisfies * ( a)*. suppose that @xmath0 is a @xmath1-fat open set with characteristics @xmath135 .
if @xmath0 is unbounded , we assume in addition that condition * ( c ) * holds .
for every @xmath138 , there exist @xmath154 and @xmath155 such that for @xmath140 , @xmath141 , @xmath156 where @xmath157 . the second main result of this paper is on explicit sharp dirichlet heat kernel estimates for subordinate brownian motions in @xmath158 open sets .
so in the remainder of this section , we assume that @xmath159 , a subordinate brownian motion with lvy exponent @xmath160 .
recall that an open set @xmath0 in @xmath161 ( when @xmath162 ) is said to be a ( uniform ) @xmath158 open set if there exist a localization radius @xmath163 and a constant @xmath164 such that for every @xmath165 , there exist a @xmath158-function @xmath166 satisfying @xmath167 , @xmath168 , @xmath169 , @xmath170 such that @xmath171 the pair @xmath172 is called the characteristics of the @xmath158 open set @xmath0 . note that a @xmath158 open set @xmath0 with characteristics @xmath172 can be unbounded and disconnected ; the distance between two distinct components of @xmath0 is at least @xmath173 . by a @xmath158
open set in @xmath174 we mean an open set which can be written as the union of disjoint intervals so that the minimum of the lengths of all these intervals is positive and the minimum of the distances between these intervals is positive .
here is the second main result of this paper , which gives an affirmative answer to the conjecture posed in @xcite . in view of remark [ r:1.1 ]
, it extends the main results of @xcite .
recall that condition * ( b ) * holds with @xmath97 for any subordinate brownian motion .
[ t : main ] suppose that @xmath6 is a subordinate brownian motion with lvy exponent @xmath160 with @xmath4 being a complete bernstein function satisfying condition * ( a)*. let @xmath0 be a @xmath158 open subset of @xmath161 with characteristics @xmath172 .
\(i ) for every @xmath138 , there exists @xmath175 such that for all @xmath176\times d\times d$ ] , @xmath177 \(ii ) for every @xmath138 , there exists @xmath178 such that for all @xmath176\times d\times d$ ] , @xmath179 \(iii ) suppose in addition that @xmath0 is bounded . for every @xmath138
, there exists @xmath180 depending only on @xmath181 and @xmath182 so that for all @xmath183 , @xmath184 where @xmath149 is the largest eigenvalue of the generator of @xmath80 .
recall that @xmath152 and @xmath153 are the constants in condition * ( c)*. combining theorem [ t : main](i)(ii ) and remark [ r:1.1](ii ) , we have the following corollary .
[ c : main2 ] suppose that @xmath6 is a subordinate brownian motion with lvy exponent @xmath160 with @xmath4 being a complete bernstein function satisfying condition * ( a)*. let @xmath0 be a @xmath158 open subset of @xmath161 with characteristics @xmath172 .
if @xmath0 is unbounded , we assume in addition that condition * ( c ) * holds .
for every @xmath138 , there exist @xmath175 and @xmath185 such that for @xmath140 , @xmath141 , @xmath186 when @xmath6 is a rotationally symmetric @xmath11-stable process in @xmath7 , theorem [ t : main ] is first established in @xcite .
sharp two - sided dirichlet heat kernel estimates in @xmath158 open sets are subsequently established in @xcite for censored stable processes , relativistic stable processes , mixed stable processes , and mixed brownian motion and stable processes , respectively . by integrating the two - sided heat kernel estimates in theorem [ t : main ] with respect to @xmath76
, we obtain the two - sided estimates on the green function @xmath187 ( see theorem [ t : gfe ] below ) , which extend ( * ? ? ?
* theorem 1.1 ) .
condition * ( a ) * is a very weak condition on the behavior of @xmath4 near infinity .
using the tables at the end of @xcite , one can come up plenty of explicit examples of complete bernstein functions satisfying condition * ( a)*. here are a few of them .
\(1 ) @xmath188 , @xmath189 $ ] ( symmetric @xmath11-stable process ) ; \(2 ) @xmath190 , @xmath191 and @xmath192 ( relativistic @xmath11-stable process ) ; \(3 ) @xmath193 , @xmath194 ( mixed symmetric @xmath11- and @xmath195-stable processes ) ; \(4 ) @xmath196 , @xmath191 , @xmath197 $ ] .
now we give a way of constructing less explicit complete bernstein functions that have very general asymptotic behavior at infinity .
suppose that @xmath191 and @xmath198 is a positive function on @xmath22 which is slowly varying at infinity .
we further assume that @xmath199 is a right continuous increasing function with @xmath200 ( so @xmath201 ) .
then the function @xmath202 is a stieltjes function , and so the function @xmath203 is a complete bernstein function ( see ( * ? ? ?
* theorem 7.3 ) ) .
it follows from ( * ? ? ?
* lemma 6.2 ) that @xmath204 when @xmath205 .
the rest of the paper is organized as follows .
section [ s:2 ] recalls and collects some preliminary results that will be used in the sequel , including on - diagonal heat kernel estimates and the boundary harnack principle .
section [ s:3 ] presents the interior lower bound heat kernel estimates , including an explicit lower bound estimate for symmetric lvy processes on @xmath7 .
the proof of the short time factorization result for @xmath12 ( that is , theorem [ thm : oppz](i ) and ( ii ) ) is given in section [ s:4 ] , while the proof of theorem [ t : main](i ) and ( ii ) is given in section [ s:5 ] .
the large time heat kernel estimates are proved in section [ s:6 ] .
the green function estimates for subordinate brownian motions in bounded @xmath158 open sets are derived in section [ s:7 ] from the two - sided dirichlet heat kernel estimates in theorem [ t : main ] .
the derivation , however , requires quite some effort . throughout this paper , @xmath206 and
the constants @xmath207 , @xmath208 , @xmath173 , @xmath209 , @xmath1 , @xmath210 , @xmath211 , @xmath136 , @xmath137 , @xmath152 , @xmath153 , @xmath212 and @xmath213 will be fixed .
we use @xmath214 to denote generic constants , whose exact values are not important and can change from one appearance to another .
the labeling of the constants @xmath214 starts anew in the statement of each result .
the dependence of the constant @xmath215 on the dimension @xmath216 will not be mentioned explicitly
. we will use @xmath217 " to denote a definition , which is read as is defined to be " .
we will use @xmath218 to denote a cemetery point and for every function @xmath50 , we extend its definition to @xmath218 by setting @xmath219 .
we will use @xmath220 to denote the lebesgue measure in @xmath161 . for a borel set @xmath221
, we also use @xmath222 to denote its lebesgue measure .
in the first part of this section , we assume that @xmath6 is a purely discontinuous rotationally symmetric lvy process with lvy exponent @xmath5 and lvy density @xmath65 satisfying and respectively , where the complete bernstein function @xmath4 satisfies * ( a)*. a function @xmath223 is said to be harmonic in an open set @xmath79 with respect to @xmath6 if for every open set @xmath224 whose closure is a compact subset of @xmath0 , @xmath225
\qquad \hbox{for every } x\in b.\ ] ] a function @xmath223 is said to be regular harmonic in an open set @xmath79 with respect to @xmath6 if @xmath226 \qquad \hbox{for every } x\in d.\ ] ] clearly , a regular harmonic function in @xmath0 is harmonic in @xmath0 .
very recently the following form of the boundary harnack principle is established in @xcite .
[ ubhp ] there exists @xmath227 such that for any @xmath228 , any open set @xmath229 , any @xmath230 and any nonnegative functions @xmath231 in @xmath7 which are regular harmonic in @xmath232 with respect to @xmath6 and vanish in @xmath233 , we have @xmath234 recall also from @xcite that @xmath235 enjoys the following properties : for every @xmath112 , @xmath236,\ ] ] and there exists @xmath237 such that @xmath238 it follows from and that both the function @xmath90 defined by and the function @xmath235 satisfy a doubling property ; that is , for every constant @xmath112 , there is a constant @xmath103 so that @xmath239.\ ] ] moreover , under condition * ( a ) * , for any given @xmath112 , @xmath109 satisfies all the conditions in @xcite for @xmath240 .
thus we have the following two - sided estimates for @xmath8 from @xcite : [ p : new1 ] for any @xmath138 , there exists @xmath241 such that @xmath242\times \r^d \times \r^d.\ ] ] for any @xmath243 , there exists @xmath244 such that for all @xmath245\times \r^d\times \r^d$ ] with @xmath246 @xmath247 is given in the first display on page 1073 of @xcite . by (
* theorem 2.4 ) and , there exist @xmath248 and @xmath249 such that for all @xmath250\times \r^d\times \r^d$ ] with @xmath251 , holds .
now we assume @xmath252 .
we construct @xmath253 from @xmath6 by removing jumps of size larger than @xmath254 via meyer s construction ( see @xcite ) .
let @xmath255 be the transition density of @xmath253 . by ( * ? ?
* lemma 3.6 ) and ( * ? ? ?
* lemma 3.1(c ) ) we have for every @xmath96 and @xmath256 , @xmath257 where @xmath258 applying ( * ? ? ?
* theorem 1.4 ) and its proof to , we get @xmath259 now follows from .
the function @xmath260 gives rise to a lvy system for @xmath6 , which describes the jumps of the process @xmath6 : for any non - negative measurable function @xmath50 on @xmath261 with @xmath262 for all @xmath263 and stopping time @xmath182 ( with respect to the filtration of @xmath6 ) , @xmath264 = \e_x \left [ \int_0^t \left ( \int_{\br^d } f(s , x_s , y ) j_x(x_s , y ) dy \right ) ds \right].\ ] ] ( see , for example , ( * ? ? ?
* proof of lemma 4.7 ) and ( * ? ? ? * appendix a ) . ) when @xmath265 and @xmath4 is a complete bernstein function satisfying condition * ( a ) * , the following boundary harnack principle on a @xmath158 open subset with explicit decay rate is established in @xcite ( see also @xcite ) .
[ t : bhp ] suppose that @xmath6 is a rotationally symmetric lvy process with lvy exponent @xmath160 with @xmath4 being a complete bernstein function satisfying condition * ( a)*. assume that @xmath0 is a ( possibly unbounded ) @xmath2 open set in @xmath161 with characteristics @xmath172
. then there exists @xmath266 such that for @xmath267 $ ] , @xmath268 and any nonnegative function @xmath269 in @xmath7 that is harmonic in @xmath270 with respect to @xmath6 and vanishes continuously on @xmath271 , we have @xmath272
in this section , we assume that @xmath6 is a purely discontinuous rotationally symmetric lvy process with lvy exponent @xmath5 and lvy density @xmath65 satisfying and respectively , where the complete bernstein function @xmath4 satisfies * ( a)*. we will give some preliminary lower bounds on @xmath12 and @xmath8 .
we first recall the following from @xcite . recall that @xmath273 was defined in .
[ l : exit ] for any positive constant @xmath274 , there exist @xmath275 and @xmath276 such that for all @xmath277 $ ] and @xmath278 , @xmath279 and @xmath280\ , \leq\ , c \,\phi ( r).\ ] ] follows directly from ( * ? ? ?
* lemma 2.5 ) .
we then have @xmath281 \geq \eps \phi ( r ) \p_z ( \tau_{b(z , r ) } > \eps \ , \phi ( r ) ) > 2^{-1 } e^{-\|\sj_1\|_{\infty}}\eps \phi ( r ) .\ ] ] on the other hand , by the lvy system for @xmath6 in , , and , we have @xmath282\\ & \geq & c_1\e_z \left [ \int_0^{\tau_{b(z , r ) } } \int_{b(z , 3r)\setminus b(z , 2r ) } j(x_s , y ) dy ds\right ] \geq c_2 \frac{|b(z , 3r)\setminus b(z , 2r)|}{r^d \phi ( r ) } \ , \e_z \left [ \tau_{b(z , r ) } \right ] , \end{aligned}\ ] ] which yields @xmath283\leq c \phi ( r)$ ] .
[ l : cks2lemma3.1 ] for any positive constants @xmath284 and @xmath254 , there exists @xmath285 such that for all @xmath286 and @xmath287 $ ] , @xmath288 by and , there exists @xmath289 such that for all @xmath290 $ ] , @xmath291 thus it suffices to prove the lemma for @xmath292 . applying the parabolic harnack inequality ( * ? ? ?
* theorem 5.2 ) at most @xmath293 $ ] times , we conclude that there exists @xmath294 such that for every @xmath295 , @xmath296 thus @xmath297 this proves the lemma . for the next four results ,
@xmath0 is an arbitrary nonempty open set and we use the convention that @xmath298 when @xmath299 . [ p : cks2prop3.2 ] let @xmath138 and @xmath300 be constants .
there exists @xmath301 such that [ e : lb1 ] p_d(t , x , y ) c ( ^-1(t))^-d for every @xmath302\times d\times d$ ] with @xmath303 .
we fix @xmath302\times d\times d$ ] satisfying @xmath304 .
note that @xmath305 and that @xmath306
so by the parabolic harnack inequality ( * ? ? ?
* theorem 5.2 ) , there exists @xmath307 such that @xmath308 this together with lemma [ l : cks2lemma3.1 ] and yields that @xmath309 where @xmath310 for @xmath311 .
[ l : cks2lemma3.3 ] let @xmath138 and @xmath300 be constants .
there exists @xmath312 so that @xmath313 for every @xmath302\times d\times d$ ] with @xmath314 and @xmath315 .
it follows from lemma [ l : cks2lemma3.1 ] that , starting at @xmath316 , with probability at least @xmath317 the process @xmath6 does not move more than @xmath318 by time @xmath319 . thus
, it suffices to show that there exists @xmath320 such that for @xmath302\times d\times d$ ] with @xmath321 and @xmath315 , @xmath322 let @xmath323 , @xmath324 and @xmath325 .
it follows from lemma [ l : cks2lemma3.1 ] and that there exists @xmath326 such that @xmath327 \,\ge\,t \ , \p_x \left(\tau^t_x \ge t \right ) \,\ge\ , c_3\,t \qquad \hbox { for all } t\in ( 0 , t].\ ] ] since @xmath328 , by the lvy system of @xmath6 and , @xmath329\nn\\ & \ge & c_4\e_x \left[\int_0^{t\wedge \tau^t_x } \int_{b^t_y } j ( x_s , u ) duds \right ] .
\label{e : dew}\end{aligned}\ ] ] we consider two cases separately .
\(i ) suppose @xmath330 .
since @xmath331 , we have for @xmath332 and @xmath333 , @xmath334 thus by and , @xmath335 \ , | b^t_y| \ , j(2|x - y| ) \,\ge\ , c_5\ , ( \phi^{-1}(t))^d \ , t j(2|x - y|)\end{aligned}\ ] ] for some positive constant @xmath336 .
therefore , in view of , the assertion of the lemma holds when @xmath330 .
\(ii ) suppose @xmath337 . in this case , for @xmath332 and @xmath333 , @xmath338 thus from and then , @xmath339 \int_{b^t_y } j\left(|x - y| + a\phi^{-1}(t)/4 \right ) \ ,
du\\ & \ge & c_7 \ , t\ , | b^t_y| \ , j\left(|x - y| + a\phi^{-1}(t)/4 \right ) \\ & \ge & c_8 \ , t(\phi^{-1}(t))^d j\left(|x - y| + a\phi^{-1}(t)/4 \right)\end{aligned}\ ] ] for some constants @xmath340 , @xmath341 .
thus we conclude from and that the assertion of the lemma holds for @xmath337 as well .
[ p : cks2prop3.4 ] let @xmath182 and @xmath284 be positive constants .
there exists @xmath342 such that @xmath343 for every @xmath344\times d\times d$ ] with @xmath345 and @xmath315 . by the semigroup property ,
proposition [ p : cks2prop3.2 ] , lemma [ l : cks2lemma3.3 ] and , there exist positive constants @xmath346 and @xmath347 such that @xmath348 combining propositions [ p : cks2prop3.2 ] and [ p : cks2prop3.4 ] , we have the following preliminary lower bound for @xmath349 .
[ step31 ] let @xmath182 and @xmath284 be positive constants .
there exists @xmath342 such that @xmath350 for every @xmath344\times d\times d$ ] with @xmath351 . in particular
, proposition [ step31 ] with @xmath352 gives [ t : globalhke ] for any constant @xmath138 , there exists @xmath353 such that for all @xmath245\times \r^d\times \r^d$ ] , @xmath354
in this section , we continue assuming that @xmath6 is a purely discontinuous rotationally symmetric lvy process with lvy exponent @xmath5 and lvy density @xmath65 satisfying and respectively , where the complete bernstein function @xmath4 satisfies the growth condition * ( a)*. throughout this section , @xmath138 is a fixed constant and @xmath0 is a fixed @xmath1-fat open set with characteristics @xmath135 . recall that @xmath355 is defined in definition [ def : ub ] . for @xmath356\times
\overline d$ ] , set @xmath357 and define @xmath358 let @xmath359 note that @xmath360 and @xmath361 .
see figure [ pic : ts ] .
[ lem : etd1r ] for every @xmath138 and @xmath362 , we have that , for @xmath363\times d$ ] , @xmath364 , \end{aligned}\ ] ] where @xmath365 and @xmath366 are the sets defined in and the ( implicit ) comparison constants in depend only on @xmath367 and @xmath4 . without loss of generality
we assume @xmath368 .
fix @xmath363\times d$ ] , and set @xmath369 .
recall that @xmath365 and @xmath366 are the sets defined in .
observe that @xmath370 note that by , and ( [ e : levy ] ) , we have @xmath371 \asymp \
, \phi ( r)^{-1 } \ , \e _ { x}\big [ \tau_{u(x , t)}\big ] \ , \asymp \ , t^{-1}\ , \e _ { x}[\tau_{u(x , t ) } ] .
\label{eq : cesp } \end{aligned}\ ] ] if @xmath372 , then @xmath373 and so @xmath374 which is greater than or equal to a positive constant depending only on @xmath375 by lemma [ l : cks2lemma3.1 ] .
thus we have , in view of , @xmath376 moreover , when @xmath372 , we have @xmath377 and so by lemma [ l : exit ] and , @xmath378 \leq \e_x[\tau_{u(x , t ) } ] \leq \e_x[\tau_{b(x , r ) } ] \le c_3\phi(r ) \leq c_4 t.\ ] ] combining the last two displays with and the fact that @xmath379 we arrive at the assertion of the lemma when @xmath380 .
now we assume that @xmath381 .
we note that @xmath382 for @xmath383 $ ] , by theorem [ ubhp ] , we have @xmath384 note that when @xmath385 , @xmath386 and so @xmath387 .
thus , by , and lemma [ l : exit ] , @xmath388 \geq c_7 \phi ( r)^{-1}\ , \e_0\big[\tau_{b(0,\kappa r/3)}\big ] \ge c_8 \phi ( r)^{-1}\,\phi(\kappa r/3 ) \ge c_9,\end{aligned}\ ] ] where the last inequality is due to .
therefore , @xmath389 since @xmath390,$ ] we have by ( [ eq : rsp ] ) , and , @xmath391 .\label{eq : cesp00 } \end{aligned}\ ] ] on the other hand , by the strong markov property of @xmath6 , @xmath392\\ & & + \e_x\left [ \p_{x_{\tau_{u(x , t)}}}\left(\tau_{b(x_{\tau_{u(x , t)}},\kappa r/6)}\le mt \right ) : x_{\tau_{u(x , t ) } } \in b(a'_r(x ) , \kappa r/6)\right]\\ & \le & \e_x\left [ \p_{x_{\tau_{u(x , t)}}}\left(\tau_{v(x , t ) } > mt\right ) ; x_{\tau_{u(x , t ) } } \in b(a'_r(x ) , \kappa r/6)\right]\\ & & + \p_{0}\left(\tau_{b(0,\kappa r/6 ) } \le mt\right ) \p_x\left ( x_{\tau_{u(x , t ) } } \in b(a'_r(x ) , \kappa r/6)\right)\\ & = & \p_x\left ( \tau_{v(x , t ) } > mt , x_{\tau_{u(x , t ) } } \in b(a'_r(x ) , \kappa r/6)\right)\\ & & + \left(1-\p_{0}\left(\tau_{b(0,\kappa r/6)}>mt\right)\right ) \p_x\left ( x_{\tau_{u(x , t ) } } \in b(a'_r(x ) , \kappa r/6)\right)\\ & \le & \p_x\left ( \tau_{v(x , t ) } > mt \right)+\left(1-\p_{0}\left(\tau_{b(0,\kappa r/6)}>mt\right)\right ) \p_x\left ( x_{\tau_{u(x , t ) } } \in b(a'_r(x ) , \kappa r/6)\right ) .
\end{aligned}\ ] ] thus @xmath393 since @xmath394 by and lemma [ l : cks2lemma3.1 ] , combining with , we get @xmath395 \leq c_{13}\p_x(x_{\tau_{u(x , t)}}\in b(a'_r(x ) , \kappa r/6 ) ) \leq c_{14 } \p_x(\tau_{v(x , t)}>mt).\end{aligned}\ ] ] this together with and completes the proof of this lemma .
[ lemppu100 ] suppose that @xmath396 are open subsets of @xmath161 with @xmath397 and @xmath398 .
let @xmath399 . if @xmath400 and @xmath401 , then for all @xmath402 , @xmath403 \right)\left(\sup_{u\in u_1,\ , z\in u_3 }
j_x(u , z)\right)\ ] ] and @xmath404 for , see the proof of ( * ? ? ?
* lemma 3.4 ) . for , see the proof of ( * ? ? ?
* lemma 3.3 ) .
* proof of theorem [ thm : oppz](ii ) . *
it follows from lemma [ lem : etd1r ] and the semigroup property that it suffices to establish the assertion for @xmath405 .
so in the remainder of this proof , we assume that @xmath405 .
fix @xmath406 $ ] and set @xmath407 . by , , and theorem [ t : globalhke ] , @xmath408 by the semigroup property , , and lemma [ lem : etd1r ] , when @xmath409 , @xmath410 thus by and , @xmath411 now we assume @xmath412 . let @xmath413 be the set defined in , @xmath414 and @xmath415 then by condition * ( b ) * we have @xmath416 extend the definition of @xmath101 by setting @xmath417 for @xmath418 and @xmath419 . for each fixed @xmath256 and @xmath96 with @xmath420 ,
one can easily check , in view of , that @xmath421 is a parabolic function in @xmath422\times b(x , 2r)$ ] .
so by the parabolic harnack inequality from ( * ? ? ?
* theorem 5.2 ) , there is a constant @xmath423 so that for every @xmath406 $ ] , @xmath424 hence we have @xmath425 where in the last inequality , we applied condition * ( b ) * again . if @xmath426 , then @xmath427 . thus if @xmath428 , then @xmath429 thus by , , and theorem [ t : globalhke ] , @xmath430 let @xmath431 .
it follows then from , lemmas [ lem : etd1r ] and [ lemppu100 ] that for @xmath412 , @xmath432\right)\nn\\ & \leq & c_6\p_x(\tau_{d}>t)\,p(t/2 , c_5x , c_5y ) .
\label{eq : ppun3 } \end{aligned}\ ] ] hence by condition * ( b ) * , , , symmetry , the semigroup property and lemma [ lem : etd1r ] , we conclude that for all @xmath302\times d\times d$ ] , @xmath433 * proof of theorem [ thm : oppz](i ) .
* fix @xmath302\times d \times d$ ] and set @xmath434 .
let @xmath365 be the set defined in , @xmath355 and @xmath435 be the points defined in definition [ def : ub ] and respectively .
note that @xmath436 for @xmath437 , @xmath438 where in the first inequality we used proposition [ step31 ] , in the second inequality we used and respectively in the cases @xmath439 and @xmath440 .
on the other hand , for @xmath441 , by lemma [ lemppu100 ] with @xmath442 and @xmath443 , we have @xmath444 since @xmath445 we have by lemma [ l : cks2lemma3.1 ] , and , @xmath446 it follows from condition * ( a ) * that there exists @xmath447 such that @xmath448 for every @xmath449 and @xmath450 $ ] .
thus with @xmath451 we have that @xmath452 for every @xmath449 , and hence @xmath453 . thus lemma [ lem : etd1r ] implies that @xmath454 consequently , by , @xmath455 similarly , using the symmetry we also have @xmath456 .
hence we have by and , @xmath457
in this section we assume that @xmath0 is a @xmath158 open set in @xmath161 with characteristics @xmath172 and that @xmath6 is a subordinate brownian motion with lvy exponent @xmath160 , where @xmath4 is a complete bernstein function satisfying condition * ( a)*. * proof of theorem [ t : main](ii ) . * without loss of generality we assume @xmath458 . in view of theorem [ thm : oppz](ii ) , it suffices to show that @xmath459\times d.\ ] ] fix @xmath356\times d$ ] and set @xmath460 . by theorem [ t : globalhke ] , we only need to show the theorem for @xmath461 .
take @xmath462 such that @xmath463 .
let @xmath464 and @xmath465 be the unit inward normal of @xmath466 at the point @xmath467 .
put @xmath468 note that @xmath469 . applying the boundary harnack principle ( theorem [ t : bhp ] ) and we get @xmath470 take @xmath471 so that @xmath472 then , by , and , we have @xmath473\\ & \ge & c_3 \e_x [ \tau_{u_1}]\\ & \ge & c_4 \phi(\phi^{-1}(t))^{-1 } \e_x [ \tau_{u_1 } ] = c_4 t^{-1 } \e_x [ \tau_{u_1 } ] . \end{aligned}\ ] ] now by the same argument as that in , we get , @xmath474 \le c_4^{-1 } t \p_x ( x_{\tau_{u_1 } } \in b(x_2 , r ) ) \le c_5 t \p_{x_1}(x_{\tau_{u_1 } } \in b(x_2 , r ) ) \sqrt{\frac { \phi(\delta_d(x ) ) } { t } } \le c_5\sqrt{t\ , \phi(\delta_d(x ) ) } .
\label{e : nnw2 } \end{aligned}\ ] ] thus , by and , we have @xmath475 + \p_x \left ( x_{\tau_{u_1 } } \in d\setminus u_1\right ) \le c_6 \sqrt{\frac { \phi(\delta_d(x))}{t}}.\end{aligned}\ ] ] since @xmath476 , follows immediately .
let @xmath477 be the euclidean distance between @xmath119 and @xmath478 .
it is well - known that any @xmath2 open set @xmath0 with @xmath158-characteristics @xmath172 satisfies both the _ uniform interior ball condition _ and the _ uniform exterior ball condition _ : there exists @xmath479 such that for every @xmath127 with @xmath480 and @xmath481 with @xmath482 , there are @xmath483 so that @xmath484 , @xmath485 and that @xmath486 and @xmath487 for @xmath488 and @xmath489 . in the remainder of this section
, we fix such an @xmath490 and set @xmath491 . for any @xmath492 with @xmath493 , let @xmath494 be a point on @xmath478 such that @xmath495 and @xmath496 .
[ l : cks2lemma4.6 ] let @xmath497 and @xmath300 .
there exists a constant @xmath498 such that for every @xmath499\times d$ ] with @xmath500 and @xmath501 , @xmath502 where @xmath503 .
let @xmath504 and assume first that @xmath505 .
as in the proof of lemma [ l : cks2lemma3.3 ] , we get that , in this case , using the fact that @xmath506 $ ] , there exist constants @xmath507 , @xmath508 such that for all @xmath509 , we have @xmath510 by taking @xmath511 , this shows that holds for all @xmath300 in the case when @xmath512 .
so it suffices to consider the case that @xmath513 .
we now show that there is some @xmath514 so that holds for every @xmath515 and @xmath516 . for simplicity , we assume without loss of generality that @xmath517 and let @xmath518 .
let @xmath519 . by the strong markov property of @xmath520 at the first exit time @xmath521 from @xmath522 and lemma
[ l : cks2lemma3.1 ] , there exists @xmath523 such that @xmath524\right ) \nonumber \\ & \geq & c_3 \ , \p_{x}\left ( \tau_u < at \hbox { and } x_{\tau_u}\in b(0 , 2^{-1 } \kappa_0 \phi^{-1}(t ) ) \right ) \label{e:4.8}.\end{aligned}\ ] ] let @xmath525 and @xmath526 .
it follows from the boundary harnack principle ( theorem [ t : bhp ] ) and that there exist @xmath527 , @xmath528 , such that for all @xmath529 $ ] , @xmath530 by , lemma [ l : exit ] and - , we have @xmath531 \\ & \geq & c_6 j(\kappa_0 \phi^{-1}(t ) ) \ , |b(0 , \kappa_0 \phi^{-1}(t ) /2)| \ , \e_x \left [ \tau_{b_1 } \right ] \\ & \geq & \frac{c_7 } { ( \kappa_0 \phi^{-1}(t))^d \phi ( \kappa_0 \phi^{-1}(t ) ) } \ , ( \kappa_0 \phi^{-1}(t))^d \ , \phi ( \kappa_0 \phi^{-1}(t ) ) = c_7.\end{aligned}\ ] ] thus @xmath532 it follows from that there exists @xmath533 such that @xmath534 \le \ , a^{-1 } c_9\ , \sqrt{\frac{\phi(\delta_d(x))}{t}}.\end{aligned}\ ] ] define @xmath535 .
we have by and the display above that for @xmath515 , @xmath536 and show that holds for every @xmath515 and for every @xmath127 with @xmath537 .
now we deal with the case @xmath538 and @xmath513 . if @xmath539 , we have from for the case of @xmath540 that there exist @xmath541 and @xmath542 such that @xmath543 if @xmath544 ( in this case @xmath545 ) , we get from by taking @xmath546 . the proof of the lemma is now complete .
* proof of theorem [ t : main](i)*. by theorem [ thm : oppz](i ) , it suffices to show that @xmath547\times d.\ ] ] assume @xmath356\times d$ ] . since @xmath0 satisfies the uniform interior ball condition with radius @xmath490 and @xmath548
, we can choose a point @xmath549 as follows : if @xmath550 , let @xmath551 so that @xmath552 and @xmath553 for every @xmath554 ) . if @xmath555 , choose @xmath556 so that @xmath557 .
note that in this case , @xmath558 and @xmath559 for every @xmath560 .
we also define @xmath561 the same way . if @xmath562 , by lemma [ lem : etd1r ] ( with @xmath563 when @xmath564 ) and lemma [ l : cks2lemma4.6 ] ( with @xmath565 ) , @xmath566 if @xmath567 , by lemma [ lem : etd1r ] , proposition [ step31 ] and , @xmath568
in this section , we first give the proofs of theorems [ thm : oppz](iii ) and [ t : main](iii ) . *
proof of theorem [ thm : oppz](iii ) . * since @xmath0 is bounded , in view of , the transition semigroup @xmath569 of @xmath80 consists of hilbert - schmidt operators , and hence compact operators , in @xmath570 .
so @xmath571 has discrete spectrum @xmath572 , arranged in decreasing order and repeated according to their multiplicity .
let @xmath573 be the corresponding eigenfunctions with unit @xmath574-norm ( @xmath575 ) which forms an orthonormal basis for @xmath570 . clearly , for every @xmath576 @xmath577 by using the eigenfunction expansion of @xmath578 we get @xmath579 noting that @xmath580 is increasing and @xmath581 , we have for all @xmath96 , @xmath582 on the other hand , by theorem [ thm : oppz](ii ) , remark [ r:1.1](iii ) and we have that there is a constant @xmath583 so that for every @xmath127 , @xmath584 it now follows from that for every that for every @xmath96 , @xmath585 for @xmath586 and @xmath587 , we have that @xmath588 by theorem [ thm : oppz](ii ) , remark [ r:1.1](iii ) , and we have that there are constants @xmath589 , @xmath311 , so that for every @xmath586 and @xmath587 , @xmath590 by theorem [ thm : oppz](i ) , theorem [ t : globalhke ] , the boundedness of @xmath0 and we have that there are constants @xmath589 , @xmath591 , so that for every @xmath586 and @xmath587 , @xmath592 this combined with establishes theorem [ thm : oppz](iii ) . *
proof of theorem [ t : main](iii ) .
* by theorem [ t : main](i ) , it suffices to prove the theorem for @xmath593 . by , and the boundedness of @xmath0 , @xmath594 .
this and theorem [ thm : oppz](iii ) imply theorem [ t : main](iii ) .
in this section , we use theorem [ t : main ] to get sharp two - sided estimates on the green functions of subordinate brownian motions in bounded @xmath2 open sets .
we first establish the following two lemmas .
the second inequality holds trivially . without loss of generality ,
we assume @xmath597 . if both @xmath598 and @xmath599 are less than 1 or if both are larger than one , @xmath600 so we only need to consider the case when @xmath601 .
note that @xmath602 . if @xmath603 , then by , @xmath604 and so @xmath605 when @xmath606 , by again , @xmath607 and so @xmath608 where the assumption @xmath609 is used in the last inequality .
this establishes the first inequality of . by condition * ( a ) * , we have that for every @xmath138 , there exist @xmath610 such that @xmath611 moreover , for every @xmath612 , we have [ e:6.20 ] r(r ) ( r ) r(0 , m ] ( see the paragraph after ( * ? ? ? * lemma 1.3 ) ) . on the other hand , for @xmath618 with @xmath623 , using the change of variable @xmath624 and then applying integration by parts for the first integral below , we have @xmath625 since @xmath626 , by and the fact that @xmath627 is increasing , @xmath628 for some @xmath629 . using and in the second integral in , we get that for @xmath618 with @xmath623 , @xmath630 [ t : gfe ] suppose that @xmath6 is a subordinate brownian motion with lvy exponent @xmath160 with @xmath4 being a complete bernstein function satisfying condition * ( a)*. let @xmath0 be a bounded @xmath158 open subset of @xmath161 with characteristics @xmath172 and @xmath638 , @xmath639 . using the boundedness of @xmath0 , remark [ r:1.1](iii ) and , the results of theorem [ t : main](i)(ii )
can be rewritten as follows : there exists @xmath655 such that for @xmath302\times d\times d$ ] , @xmath656 [ c : gfe ] suppose that @xmath6 is a one - dimensional subordinate brownian motion with lvy exponent @xmath160 with @xmath4 being a complete bernstein function satisfying condition * ( a)*. let @xmath0 be a bounded @xmath158 open subset of @xmath174 with characteristics @xmath172 and @xmath638 , @xmath668 . [ r:6.4 ] let @xmath678 .
note that @xmath679 let @xmath680 then @xmath681 the upper and lower matuszewska indices can be computed as @xmath682 thus we have @xmath683 ( see @xcite ) .
let @xmath684 , @xmath685 . with the change of variable @xmath686 , we see that is equivalent to [ e:6.9n ] ( r)=_t^-2^r1(t ) t^1/2 dt = 2_r^-1/2^t ds c_1 ( r^-1/2 ) r^1/2 = ( r ) r t^-2 .
let @xmath687 . is equivalent to [ e:6.10n ] ( r)= ^_r 1(t ) t^1/2 dt = 2_0^r^-1/2 ds c_2 ( r^-1/2 ) r^1/2=(r ) r t^-2 .
in fact , by ( * ? ? ? * corollaries 2.6.2 and 2.6.4 ) , @xmath688 for every @xmath689 if and only if @xmath690 , and @xmath691 for every @xmath689 if and only if @xmath692 . by following the same proof in (
* section 6 ) , one can construct @xmath4 whose upper and lower matuszewska indices are @xmath693 and @xmath694 so that @xmath695 and @xmath696 . for such @xmath4 , neither nor hold .
* proof of corollary [ c : gfe ] .
* put @xmath697 , where @xmath128 is the diameter of @xmath0 and @xmath616 is the constant in . for notational simplicity ,
we let @xmath698 and @xmath699 . recall that @xmath700 is defined by . in view of theorem [ t : gfe](iii ) , we only need to consider @xmath701 , which we will assume from now on in this proof .
\(ii ) since @xmath704 , by we have that for @xmath701 , @xmath705 thus by theorem [ t : gfe](iii ) , @xmath706 since @xmath707 is increasing in @xmath102 , the above together with theorem [ t : gfe](iii ) implies that @xmath708 we next give an example of a one - dimensional subordinate brownian motion with lvy exponent @xmath709 that satisfies condition * ( a ) * but its associated function @xmath710 satisfies neither condition nor .
we can get its explicit green function estimates on bounded @xmath158 open sets by using theorem [ t : gfe](iii ) but not by corollary [ c : gfe ] .
[ e : g1 ] by the discussion at the end of the introduction , we know that for any @xmath711 , the function @xmath712 is a complete bernstein function .
moreover , @xmath713 when @xmath714 .
when @xmath715 $ ] , @xmath716 is an explicit example of a complete bernstein function such that @xmath713 when @xmath714 .
let @xmath6 be a one - dimensional subordinate brownian motion with lvy exponent @xmath709 .
note that [ e:7.16 ] ( r)=1/ ( 1/r^2 ) r ( ( 1/r ) ) ^-p 0<r1/2 .
let @xmath717 be the lambert - w function , that is , @xmath717 is the unique solution to @xmath718 .
suppose @xmath719 .
then @xmath720 so @xmath721 , which implies that @xmath722 .
so in view of , there is a constant @xmath723 so that @xmath724.\ ] ] let @xmath734 .
we have by theorem [ t : gfe](iii ) , and - that for @xmath728 , @xmath735 while for @xmath730 , @xmath736 where @xmath737 .
it is elementary to check that for @xmath738 and @xmath739 , @xmath740 when @xmath728 and @xmath741 thus we have from the four displays above together with that for @xmath728 , @xmath742 while for @xmath730 , @xmath743 * acknowledgement . * the main results of this paper ,
in particular theorems [ thm : oppz ] and corollary [ c : main2 ] , were reported at the _ the sixth international conference on stochastic analysis and its applications _ held at bedlewo , poland , from september 9 to 14 , 2012 . at the same meeting
, k. bogdan , t. grzywny and m. ryznar announced that they had also obtained the factorization form estimates in terms of surviving probabilities and global heat kernel @xmath744 as in our theorem [ thm : oppz](i)(ii ) for the dirichlet heat kernels of a similar class of purely discontinuous subordinate brownian motions considered in this paper but only for bounded @xmath158 open sets .
chen , p. kim and t. kumagai . on heat kernel estimates and parabolic harnack inequality for jump processes on metric measure spaces .
_ acta mathematica sinica , english series * 25 * _ ( 2009 ) , 10671086 . | in this paper , we consider a large class of purely discontinuous rotationally symmetric lvy processes .
we establish sharp two - sided estimates for the transition densities of such processes killed upon leaving an open set @xmath0 .
when @xmath0 is a @xmath1-fat open set , the sharp two - sided estimates are given in terms of surviving probabilities and the global transition density of the lvy process .
when @xmath0 is a @xmath2 open set and the lvy exponent of the process is given by @xmath3 with @xmath4 being a complete bernstein function satisfying a mild growth condition at infinity , our two - sided estimates are explicit in terms of @xmath5 , the distance function to the boundary of @xmath0 and the jumping kernel of @xmath6 , which give an affirmative answer to the conjecture posted in @xcite .
our results are the first sharp two - sided dirichlet heat kernel estimates for a large class of symmetric lvy processes with general lvy exponents .
we also derive an explicit lower bound estimate for symmetric lvy processes on @xmath7 in terms of their lvy exponents . *
ams 2000 mathematics subject classification * : primary 60j35 , 47g20 , 60j75 ; secondary 47d07 * keywords and phrases * : lvy processes , subordinate brownian motion , heat kernel , transition density , dirichlet transition density , green function , exit time , lvy system , boundary harnack inequality , parabolic harnack inequality |
magnetic fields is a key ingredient in present - day star formation .
for example , protostellar jets , which are ubiquitous in star - forming regions , are considered to be driven from protostars by the lorentz force .
protostellar jets influence gas accretion onto protostars and disturb the ambient medium .
in addition , the angular momentum of the cloud is removed by magnetic braking and protostellar jets .
the removal of angular momentum makes protostar formation possible in a parent cloud that has a much larger specific angular momentum than the protostar .
so far , magnetic effects in primordial gas clouds have been ignored in many studies because magnetic fields in the early universe are supposed to be extremely weak .
however , recent studies indicate magnetic fields of moderate strength can exist even in the early universe .
cosmological fluctuations produced magnetic fields before the epoch of recombination @xcite .
these fields were sufficiently large to seed the magnetic fields in galaxies . a generation mechanism for magnetic fields at the epoch of reionization
are also proposed @xcite , in which magnetic fields in intergalactic matter are amplified up to @xmath1 .
these fields can therefore increase up to @xmath2 in the first collapsed object having number density of @xmath3 .
these fields may influence the evolution of primordial gas clouds and formation of population iii stars . under
spherical symmetry including hydrodynamical radiative transfer , many authors have carefully investigated both the present - day ( e.g. , * ? ? ? * ) and primordial ( e.g. , * ? ? ?
* ) star formation processes . a significant difference between present - day and
primordial star formation exists in the thermal evolution of the collapsing gas cloud because of differences in the abundance of dust grains and metals . in present - day star formation , the gas temperature in molecular clouds is @xmath4k .
these clouds collapse isothermally for @xmath5 .
then the gas becomes adiabatic at @xmath6 , and an adiabatic core ( or the first core ) forms .
after the dissociation of molecular hydrogen ( @xmath7 ) , the protostar forms at @xmath8 .
on the other hand , primordial gas clouds have temperatures of @xmath9k at @xmath10 @xcite . these clouds collapse keeping polytropic index @xmath11 for a long range of @xmath12 .
thus , the first core does not appear in the primordial collapsing cloud .
after the central density reaches @xmath7 , the thermal evolution of the primordial collapsing cloud begins to coincide with that of a present - day cloud @xcite .
the difference in thermal evolution between present - day and primordial clouds arises when @xmath13 .
another major difference between present - day and primordial star formation exists in their magnetic evolution . in present - day star formation
, neutral gas is well - coupled with ions for @xmath14 and @xmath15 , while the magnetic field dissipates by ohmic dissipation in the range of @xmath16 @xcite . in the collapsing cloud , @xmath17% of the magnetic field
is dissipated for @xmath16 @xcite . on the other hand , in a primordial gas cloud ,
the magnetic field couples strongly with the primordial gas during all phases of the star formation , as long as the initial field strength is weaker than @xmath18 g @xcite . in summary ,
the magnetic field is largely dissipated by ohmic dissipation before protostar formation in present - day clouds , while the magnetic field can continue to be amplified without dissipation in primordial clouds . the magnetic field in the collapsing cloud is closely related to the fragmentation or formation of binary or multiple stellar systems . in present - day star formation
, the magnetic field strongly suppresses rotation - driven fragmentation @xcite . for primordial clouds , magnetic effects on fragmentation are still unknown . in this study
, we investigate the evolution of weakly magnetized primordial clouds and the formation of population iii stars using three - dimensional simulations .
to calculate the evolution of primordial star - forming clouds in a large dynamic range , we use the three - dimensional ideal mhd nested grid method @xcite . for gas pressure , we use a barotropic relation that approximates the result of one - zone calculation where thermal and chemical processes for primordial gas are solved in detail @xcite . as the initial state ,
we take a spherical cloud whose density is twice higher than that for hydrostatic equilibrium with external pressure ( bonnor - ebert sphere ) .
the central ( number ) density of the bonnor - ebert sphere is set at @xmath19 .
the initial temperature is set at 250k , and the radius of a bonnor - ebert sphere is @xmath20pc . to promote fragmentation , a small @xmath21-mode density perturbation is imposed on the spherical cloud .
the total mass contained inside @xmath22 is @xmath23 .
the initial cloud rotates around the @xmath24-axis at a uniform angular velocity of @xmath25 , and has a uniform magnetic field @xmath26 parallel to the @xmath24-axis ( or rotation axis ) .
the initial model is characterized by two non - dimensional parameters : the ratios of the rotational energy to the gravitational energy @xmath27 ( @xmath28 ) , and of the magnetic energy to the gravitational energy @xmath29 ( @xmath30 ) , where @xmath31 , @xmath32 and @xmath33 are the rotational , magnetic and gravitational energies , respectively .
we made 36 models using different values for these two parameters .
we adopted the nested grid method ( for details , see * ? ? ?
* ; * ? ? ?
* ) to obtain high spatial resolution near the center .
each level of a rectangular grid has the same number of cells ( @xmath34 ) , with the cell width @xmath35 depending on the grid level @xmath36 .
the cell width is halved with every increment of the grid level .
the highest level of the grid changes dynamically : a new finer grid is generated whenever the minimum local jeans length @xmath37 falls below @xmath38 , where @xmath39 is the cell width . the maximum level of grids is restricted to @xmath40 . since the density is highest in the finest grid , generation of a new grid ensures the jeans condition with a safety factor of 2 .
we begin our calculations with three grid levels ( @xmath41 ) .
the box size of the initial finest grid @xmath42 is chosen to be @xmath43 , where @xmath22 is the radius of the critical bonnor - ebert sphere . the coarsest grid ( @xmath44 ) then has a box size of @xmath45 .
figure [ fig:1 ] shows the cloud evolution before protostar formation ( @xmath46 ) from the initial stage for magnetically - dominated model that has a larger magnetic energy than the rotation energy ( @xmath47 ) .
this model has parameters of ( @xmath29 , @xmath27 ) = ( @xmath48 , @xmath49 ) , and the initial cloud has a magnetic field strength of @xmath50 g , and an angular velocity of @xmath51s@xmath52 .
figure [ fig:1]_a _ shows the initial spherical cloud threaded by a uniform magnetic field .
figure [ fig:1]_b__d _ shows the cloud structure around the center of cloud when the central density reaches @xmath53 ( _ b _ ) @xmath54 , ( _ c _ ) @xmath55 , and ( _ d _ ) @xmath56 , respectively .
the density contours projected on the sidewall in these figures indicate that the central region becomes oblate as the cloud collapses because of the magnetic field and rotation , both of which are amplified as the cloud collapses .
figure [ fig:1]_a__d _ also shows that the magnetic field lines gradually converge toward the center as the central density increases .
figure [ fig:1]_g__i _ shows the evolution after the protostar formation , in which the strong jet is driven by the circumstellar disk .
the speed of jet is @xmath57 that corresponds to the kepler speed near the protostar .
figure [ fig:2 ] shows a configuration of magnetic field lines ( black - white streamlines ) , a structure of a jet ( transparent iso - velocity surface ) and fragments for some models against @xmath27 - @xmath29 plane .
the magnetically - dominated models ( @xmath47 ) show a powerful jet after the proto - population iii star formation without fragmentation ( or binary formation ) , while the rotation - dominated models ( @xmath58 ) show fragmentation and binary formation without jet driving . in the jet - driving models ,
the magnetic field lines are strongly twisted inside the jet region .
these jets have hourglass - like configurations of the magnetic field lines , where the poloidal component is more dominant than the toroidal component .
these configurations of the magnetic field lines can easily drive a strong jet by the disk wind mechanism . in fragmentation models ,
the configurations of the magnetic field lines are disturbed , and the toroidal field tends to be more dominant than the poloidal filed .
such configuration of the magnetic field lines means the weak magnetic field , which is insufficient to drive a jet .
the weak poloidal field is attributed to the oblique shocks on the surface of the disk envelope , and the disturbance of the field lines are attributed to the orbital motion of the fragments and the spin of the fragments .
the spin of the fragment winds the magnetic fields around its rotation axis , and the toroidal component of the magnetic field is amplified .
thus , in the further stages , the magnetic pressure may drive a jet .
in this study , we calculated cloud evolution from the stage of @xmath59 until the protostar is formed ( @xmath60 ) for 36 models , parameterizing the initial magnetic field strength and rotation , to investigate effects of magnetic fields in collapsing primordial clouds .
our calculations showed that fragmentation occurs but no jet appears when @xmath58 , and jet appears after the protostar formation without fragmentation when @xmath61 .
thus , in the collapsing primordial cloud , the cloud evolution is mainly controlled by the centrifugal force than the lorentz force when @xmath58 , while the lorenz force is more dominant than the centrifugal force when @xmath62 .
a jet is driven when the initial cloud has magnetic field of @xmath63 if the cloud rotates slowly as @xmath64s@xmath52 .
the power of a jet , e.g. , a mass ejection rate , is considered to be controlled by the accretion rate as indicated in present - day star formation ; the mass ejection rate of a jet is 1/10 of the mass accretion onto the protostar .
the accretion rate of primordial star formation is expected to be considerably larger than that at present day , and it produces a stronger jet .
the life time of the jet also seems to be controlled by accretion in the present - day ; a jet stops when mass accretion stops . for population iii stars ,
the gas accretion does not halt within their lifetimes @xcite .
therefore , a jet also may continue during the all lifetime of the protostar , and the strong jet propagates to disturb a surrounding medium significantly .
the disturbance of the medium could trigger the subsequent star formation as frequently observed in present - day star formation . assuming the power law growth of @xmath65 , the critical strength of the magnetic field , @xmath66 g at @xmath67 corresponds to @xmath68 g at @xmath69 [ see , eq .
( [ eq : bcrit ] ) ] , which is much stronger than the background magnetic field derived by @xcite . however , when the magnetic field is amplified to @xmath70 g by some mechanisms ( e.g. , * ? ? ? * ) , the magnetic field can affect the collapse of the primordial cloud . even if a cloud has a magnetic field weaker than the critical strength @xmath66 g
, the magnetic field may play an important role after the protostar formation . in analytical study of the evolution of accretion disks around the first stars @xcite ,
it is suggested that magnetic fields amplified in the circumstellar disk eventually give rise to protostellar jets during the protostellar accretion phase .
rotation promotes fragmentation when the first collapsed objects has the angular velocity of @xmath71s@xmath52 .
the fragmentation is expected to produces binary or multiple stellar system .
when a multiple stellar system is formed , some stars can be ejected by close encounters . at the protostar formation epoch , the protostar has a mass of @xmath72 .
the ejected proto - population iii stars may evolve to metal - free brown dwarfs or low - mass stars .
when a binary component in a multiple stellar system is ejected from the parent cloud by protostellar interaction , a low - mass metal free binary may also appear in the early universe .
it is considered that the extremely metal - poor ( [ fe / h]@xmath73 - 5 ) stars @xcite are formed as binary members from metal - free gas , and then have been polluted by the companion stars during the stellar evolution @xcite .
in addition , a binary frequency in population iii star may be comparable to or larger than that at present day @xcite . in order to confirm the ejection scenario ,
the further long - term calculations are required . ,
@xmath29 ) = ( @xmath49 , @xmath48 ) ] before protostar formation @xmath74 . in each panel , the structures of the high - density region ( _ isosurface _ ) and magnetic field lines ( _ black and white streamlines _ ) are plotted in three dimensions , while the density contours ( _ color _ and _ contour lines _ ) and velocity vectors ( _ thin arrows _ ) are projected on each wall surface . the central number density @xmath75 , elapsed time @xmath76 , grid level @xmath36 , and grid size are also shown in each panel .
the transparent red surface in panels ( _ g_)-(_i _ ) means the border between inflow and outflow inside which the mass ejected by the protostellar jet . , width=529 ] and @xmath29 .
the magnetic field lines ( _ black and white streamlines _ ) , high - density regions ( _ isosurface _ ) , and jet ( _ transparent isosurface _ ) are plotted in each panel .
the density contours ( _ color _ and _ contour lines _ ) and velocity vectors ( _ thin arrows _ ) are also projected on each wall surface . the grid level ( @xmath36 ) , and grid scale
are shown in each panel .
, width=566 ] | the evolution of primordial collapsing clouds and formation of proto - population iii stars are investigated using three - dimensional ideal mhd simulations .
we calculated the collapse of magnetized primordial clouds from the prestellar stage until the epoch after the proto - population iii star formation , spatially resolving both parsec - scale clouds and sub - au scale protostars .
the formation process of proto - population iii star is characterized by the ratio of rotational to magnetic energy of the natal cloud . when the rotational energy is larger than the magnetic energy , fragmentation occurs in the collapsing primordial cloud before the proto - population iii star formation and binary or multiple system appears . instead ,
when the magnetic energy is larger than the rotational energy , strong jet with @xmath0 is driven by circumstellar disk around the proto - population iii star without fragmentation .
thus , even in the early universe , the magnetic field plays an important role in the star formation process . |
the depinning model with relaxation corresponds to the mechanical circuit sketched in fig .
1b of the main text .
we first describe the one - dimensional case .
the sample is decomposed in blocks of mass @xmath107 , labeled @xmath6 and moving along horizontal rails @xmath34 .
the action of the dashpot is to resist the change in @xmath108 via viscous friction , with a resulting force on @xmath34 given by @xmath109 .
the blocks move in a medium with an effective viscosity @xmath4 , we will study the overdamped regime @xmath110.as each block is described by two degrees of freedom @xmath34 and @xmath111 , its time evolution is governed by two equations .
the first equation comes from the force balance on @xmath34 : @xmath112 the second equation is derived from the force balance on @xmath111 : @xmath113 where we assumed that the internal degree of freedom @xmath111 has no mass .
we inject this second equation into the first , and subtract the force balance on @xmath114 to the second equation .
it is convenient to let the laplacian term @xmath115 appear by defining the variable @xmath116 : @xmath117 to generalize this to higher dimensions ( on a square lattice ) , one simply has to connect each block @xmath34 to its neighbours via symmetrically arranged viscoelastic elements .
the equations obtained are exactly the same , with the label @xmath6 now referring to @xmath0-dimensional space , the @xmath42 discrete laplacian replaced with the @xmath0-dimensional laplacian denoted @xmath118 , and the @xmath19 variable redefined as : @xmath119 in this way , one obtains eq .
( 2 ) of the main text .
it is worth to notice that the dynamics of the model without relaxation satisfies the middleton theorem @xcite that guarantees that the interface moves only forward .
however , in presence of viscoelastic elements the middleton theorem does not apply , and backward movements of the interface are possible .
fortunately , these movements are not frequent , due to the biased driving term @xmath120 and we observed numerically that the real dynamics yields the same statistical results as the dynamics that allows only forward movements .
thus , we restrain the dynamics to forward movements . to efficiently study eqs .
( 1 ) and ( 2 ) of the main text , we adopted the so - called `` narrow wells '' approximation , which does not affect the universal properties of the dynamics , and can be extended to the model with relaxation .
this approach greatly simplifies the numerical simulations and allows an analytical analysis in the mean field case . in this scheme ,
the disorder is modeled as a collection of narrow pinning wells representing impurities ( see fig .
along the @xmath26 direction , the pinning wells are separated by random intervals ( spacings ) @xmath54 with distribution @xmath55 and mean length @xmath56 .
a natural choice for @xmath55 is the exponential law , which corresponds to the case where impurities are uncorrelated in space .
if the spatial extension of the wells is negligible compared to @xmath24 , we can safely assume that each block is located in one of those wells , so that its coordinate @xmath34 evolves only by discrete jumps @xmath54 . to exit from a well
, a block needs to be pulled by a force larger than a threshold @xmath121 related to the well s depth . within this approximation
, the continuous dynamics of the blocks can be re - written in terms of the variable @xmath58 , which measures the remaining stability range of the block @xmath6 . in the depinning model
it reads : @xmath59 increasing the load @xmath13 , all @xmath58 s decrease , until a block becomes unstable ( @xmath122 ) and moves to the next pinning well ( @xmath123 ) , characterized by a new random threshold @xmath124 .
the unstable block can be the seed of an avalanche because its motion produces a drop @xmath125 in the variables @xmath126 of the neighbouring blocks .
the avalanche event is exhausted when all blocks are stable .
in general , the mean field limit can be studied using models which are much simpler than their finite dimension counterpart .
for example , the mean field depinning can be mapped @xcite onto the problem of a single particle driven in a brownian force landscape , the so - called abbm model @xcite .
many results on the avalanche statistics of the mean field interface can be obtained from this latter model @xcite .
unfortunately , such a mapping does not hold in presence of relaxation ( recently , a version of the abbm model `` with retardation '' was studied @xcite , displaying aftershocks but no oscillatory behavior ) .
a different strategy , which can be generalized to that case , is to consider the fully connected model , where each site interacts with all the others . in the fully connected model , all sites are equivalent and the @xmath58 s are independent and identically distributed variables , characterized by their probability distribution @xmath65 which in general depends on the initial condition @xmath66 and on the value of @xmath13 .
the aim of this section is to write down the evolution equation for @xmath65 when @xmath13 increases .
we will show that in the depinning case the distribution reaches a stationary form , while the viscoelastic depinning displays a periodic solution . by replacing the local term @xmath127 in eq .
( 4 ) of main text with its fully connected version @xmath128 , we obtain : @xmath129 let us set the threshold force @xmath57 to be constant , a choice that , for the mean field analysis , does not alter the results .
when the external driving is increased by a small positive quantity @xmath130 , the distribution evolves from its initial shape @xmath65 , to a new shape @xmath131 . in order to compute the latter , it is useful to decompose the dynamical evolution in different steps . in a first step ,
the center of the parabolic potential moves from @xmath13 to @xmath132 and all @xmath58 s decrease by @xmath133 , moreover a fraction @xmath134 of sites becomes unstable and moves to the next wells .
the new @xmath58 are given by @xmath135 , with @xmath54 s drawn from the distribution @xmath55 .
this writes : @xmath136 the redistribution of @xmath126 s changes @xmath137 of a quantity @xmath138 , so that all blocks are subject to a shift in their @xmath58 .
this can induce a second step which acts on @xmath139 exactly as @xmath140 did on @xmath65 , but with a shift @xmath141 .
these steps go on until there are no more unstable sites , so that the distribution reaches the stable form @xmath142 .
let us remark that eq.[[eqpdelta ] ] has a fixed point @xmath143 found when @xmath144 this equation can be easily integrated and @xmath145 determined by the normalization condition .
this gives : @xmath146 where @xmath69 .
a stability analysis shows that the fixed point is attractive , so that any initial condition converges to it .
moreover , it is possible to prove that for a given initial condition , there exists a finite @xmath147 at which the distribution reaches the fixed point and remains there for @xmath148 .
this indicates that the large time dynamics is stationary , and that the applied stress in the system is constant in time : @xmath149 this result becomes @xmath150 for an exponentially distributed @xmath54 . for the depinning case
, we can also compute the probability distribution of the avalanche sizes @xmath151 in the fully connected approximation , for finite values of the parameters @xmath152 .
let us first consider the case where @xmath153 . for a finite system with @xmath154 sites , the typical configuration @xmath155 corresponds to a set of @xmath154 independent and identically distributed random variables drawn from @xmath46 .
let us sort the set : @xmath156 .
when the system becomes unstable we have by definition @xmath157 .
this site jumps to the next well at distance @xmath158 , so that all @xmath58 s are decreased by @xmath159 . this will produce at least another jump if @xmath160 .
more generally , the avalanche size @xmath15 corresponds to the first time that the relation : @xmath161 is fulfilled .
it is thus important to study the statistics of the @xmath58 with @xmath162 .
let us observe that when @xmath154 is very large the distribution of these @xmath58 s can be approximated with a uniform distribution : @xmath163 . within this approximation ,
the spacings @xmath164 are independent exponential variables of mean @xmath165 and variance @xmath166 .
we conclude that the sequence @xmath167 is a random walk with diffusion constant @xmath168 and drift @xmath169 .
the statistics of @xmath15 thus corresponds to the problem of first crossing with @xmath170 of a random walk with diffusion constant @xmath171 and drift @xmath172 . for a positive drift
, there is a finite probability that this random walk never crosses @xmath170 , which corresponds to a _
global _ event . for a negative drift ,
the time of zero crossing is always finite , and has been computed for the brownian motion in @xcite .
the distribution of the avalanche sizes thus reads : @xmath173 where for simplicity we have neglected the short - scale regularization in the expression of @xmath151 . if now we replace the choice @xmath153 with a broader function @xmath55 , the results of eq . [
[ smax ] ] still hold but with a different diffusion constant .
finally let us remark that the results we obtain here by focusing on @xmath58 coincide with the results obtained using the mapping to the abbm model .
the fully connected approximation for the model with viscoelastic elements is obtained by replacing @xmath127 in eq .
( 5 ) of main text with @xmath128 , as for the depinning model : @xmath174 it is useful to split @xmath126 in a _ fast _
part , @xmath175 , and a _ relaxed _ one , @xmath176 : @xmath177 under a small increase of @xmath178 , two dynamical regimes are observed . on the short time
scales @xmath76 the dashpots are blocked , so that all @xmath179 s remain constant while all @xmath180 are shifted by @xmath181 .
sites that become unstable ( i.e. with @xmath182 ) move to their next pinning well , following the rules of the rigid depinning interface , with stiffness @xmath32 .
on longer time scales @xmath77 the dashpots relax , so that @xmath183 s slowly evolve and can possibly trigger new fast events called aftershocks . within the approximation @xmath184 ,
no aftershocks are observed .
this allows the blocks to fully relax after each event , so that the system s state just before any event is always fully relaxed ( @xmath185 ) .
this corresponds to a relaxed @xmath179 : @xmath186 where the last equality is obtained by inverting eqs .
[ [ defdeltaf ] ] .
the effect of an event is to modify the distribution of @xmath187 which , just before an event , is related to @xmath65 ( using eq .
[ [ fullyrelaxed ] ] ) : @xmath188 where @xmath189 and @xmath190 . at the first step of the event ,
the unstable sites are those with @xmath191 , so that @xmath192 evolves from @xmath193 to @xmath194 via : @xmath195 this equation has a depinning - type fixed point @xmath196 , which translates for the variable @xmath126 in the fixed point @xmath197 of the depinning model for the more flexible interface ( with stiffness @xmath17 ) .
we conclude that when the dynamics consists in avalanches of very few _ steps _ ( here we assume just one ) , the distribution @xmath65 converges to this attractive fixed point .
the dynamics actually consists in avalanches of very few steps ( which are furthermore of decreasing amplitude ) , so that the distribution @xmath65 indeed goes `` up '' towards this attractive fixed point . this convergence , observed in fig .
2a of the main text , stops when @xmath65 overcomes the fixed point of the rigid interface at @xmath198 , namely @xmath199 . at this stage , a global avalanche
is triggered in the system , can not be computed from eq .
[ [ pdeltaevolution ] ] because the system is not fully relaxed during the avalanche .
instead it is necessary to follow the evolution of the joint probability distribution @xmath200 .
the details of this numerical integration are given in the supplemental material sec .
3.1 ] and @xmath65 jumps to the fixed point @xmath51 of the rigid interface ( stage @xmath50 ) .
finally , at the end of this global event , relaxation takes the system far from this fixed point , and a new cycle starts ( stage @xmath53 ) . the evolution of @xmath65 for the quenched edwards - wilkinson model when @xmath13 is increased .
the initial distribution is a gaussian centred in @xmath201 , with standard deviation @xmath202 , and the weight at the left of @xmath198 cut .
@xmath46 quickly reaches its stationary form . ]
let us discretize @xmath46 with a bin of size @xmath203 .
the distribution probability is then a vector @xmath204 ( related to @xmath46 by @xmath205 ) which evolves with the following rules : * _ driving process _ : we shift @xmath204 of one bin : @xmath206 . * _ instability check _ : we compute the weight of unstable sites : @xmath207 if @xmath208 , we perform the _ avalanche process_. + else we go back to the _ driving process_. * _ avalanche process _ :
it is composed by a stress drop and a stress shift . * * stress drop : @xmath209 * * stress shift : we shift @xmath204 of @xmath210 $ ] bins . @xmath211 + then we perform the _ instability check_. this algorithm converges very quickly from any initial configuration to @xmath143 for any choice of @xmath55 , see fig . [ convergence ] of the supplemental material .
analogously to the previous case , we discretize @xmath212 with a bin @xmath203 .
the distribution probability is then a matrix @xmath213 where @xmath214 .
the matrix evolves with the following rules : * _ driving process _ : we shift @xmath213 of one bin : + @xmath215 . *
_ instability check _ :
we compute @xmath216 .
if @xmath217 , we perform the _ driving process_. + we compute the weight of unstable sites : @xmath218 + if @xmath219 , we perform the _ avalanche process_. + else we perform the _ relaxation process_. * _ avalanche process _ : it is composed by a stress drop and a stress shift . * * stress drop : @xmath220 , + if @xmath221 : @xmath222 if @xmath223 : @xmath224 where @xmath225 . * * stress shift : we shift @xmath213 of a fraction of bin : + @xmath226 , @xmath227 + then we perform the _ instability check_. * _ relaxation process _ : we compute @xmath228 , the single bin associated to @xmath229 as with two bins , @xmath230 and @xmath231 .
the contribution @xmath232 is split in the two bins using a linear interpolation . ]
@xmath233 so that the relaxation corresponds to : @xmath234 then we perform the _ driving process_. this algorithm integrates the fully connected version of the viscoelastic model , and produces the results shown in fig . 2 of the main text .
we provide here details on the integration of the dynamic equations of the viscoelastic model .
our starting point is the set of equations ( 2 ) of the main text : @xmath235 with @xmath236 . for the numerical work ,
it is convenient to introduce variables @xmath237 and @xmath238 , defined as : @xmath239 using @xmath237 and @xmath238 , the model equations can be written as @xmath240 it is thus clear that @xmath238 represents the force onto @xmath34 exerted through @xmath17 and @xmath9 springs , whereas @xmath237 is the force coming from the branches that contain the dashpots and @xmath45 springs .
we work in the case in which temporal scales are well separated : @xmath29 .
this corresponds to @xmath241 . as discussed in the main text , within the narrow well approximation the actual integration of eqs .
[ [ hdot ] ] and [ [ fdot ] ] does not need a continuous time algorithm , but can be presented in the form of a discrete set of rules . from a relaxed configuration with @xmath242 at time @xmath243 , the load increase triggers a new instability of eq . [ [ hdot ] ] when the total force from the springs , here @xmath238 , reaches @xmath244 , and this occurs after a time interval : @xmath245 thus at time @xmath246 an avalanche starts at position @xmath6 , producing the advance of @xmath34 to the next potential well @xmath247 , and a corresponding rearrangement of the forces according to ( in two dimensions ) : @xmath248 where @xmath249 are the four neighbour sites to @xmath6 , and the value of @xmath244 is renewed from its probability distribution .
all successive unstable sites are treated in the same way until there are no more unstable sites .
this defines the primary avalanche . at this point
the relaxation dynamics [ [ fdot ] ] begins to act , until some site eventually becomes unstable .
note that due to the discrete pinning potential , in this stage @xmath26 remains constant , namely the relaxation dynamics is simply : @xmath250 this means that a given site @xmath6 will trigger an avalanche due to relaxation if for some increase in time @xmath251 the total force from the springs on this site , here @xmath252 , reaches @xmath244 , i.e. , if @xmath253 ( note that in order to have a solution , @xmath237 must be negative , as the l.h.s . is lower that the r.h.s . at the starting time ) .
this leads to the determination of @xmath251 as @xmath254 \label{dtint}\ ] ] once all the secondary avalanches generated by relaxation have been produced and @xmath237 has relaxed to zero , the external driving is increased again , according to [ [ dtext ] ] .
this is the main scheme of the simulation .
we should mention however , that its efficient implementation relies on a classification scheme of all sites , in such a way that the determination of the next unstable site in [ [ dtext ] ] and [ [ dtint ] ] does not require a time consuming sweep over the whole lattice .
in fact , following grassberger @xcite we classify the sites according to their value of the r.h.s . of [ [ dtext ] ] and [ [ dtint ] ] , and bin them , in such a way that the determination of the next unstable site can be limited to the bin corresponding to the lowest values of these quantities .
when sites change their @xmath26 values along the simulation , they are reaccommodated in the bins using a linked list algorithm . from this efficient algorithm ,
we obtain distributions @xmath151 of the events sizes for various values of @xmath9 , as shown in fig .
[ ndes ] of the supplemental material .
a power law with exponent @xmath255 is consistently obtained .
the different dynamics we observe in 2d ( power law distribution of avalanches ) compared with the mean field case ( periodic system size avalanches ) is remarkable . from a theoretical point of view
an interesting question is to identify the upper critical dimension @xmath256 of the problem , beyond which its behavior is well described by the mean field results .
our qualitatively different results in 2d compared to mean field indicate that @xmath257 .
note that marchetti and dahmen proposed a mapping of the hysteretic visco - elastic model onto the non - equilibrium random field ising model ( rfim ) @xcite .
this mapping seems to suggest an upper critical dimension @xmath258 .
however there is no strong evidence for it yet .
is consistent with the exponent @xmath259 .
the dotted line separating the regions of small and large events is indicated : the size distribution does not display any strong feature around this value .
the system size is in all cases much larger than the largest avalanche observed , and reaches values of @xmath260 .
other parameters are : @xmath83 , @xmath261 . ] one may have noticed that in the our viscoelastic model , the relaxation of the variable @xmath19 is local and controlled by a single time constant .
this choice allows the fast computation we just described , but yields an unrealistic exponential decay of the aftershocks over time , which does not follow the omori law . in this respect
, it would be suitable to consider non - local relaxation mechanisms , as the laplacian relaxation studied in @xcite , which can reproduce the omori law .
for figs . 1 , 3 and 4 , the numerical method used is the one described in the supplemental material ( sm ) section 4 .
1 the number of blocks is @xmath262 ( in 2d , we consider a square - shaped interface with @xmath263 ) .
we use @xmath264 and the thresholds @xmath265 are distributed as a gaussian of mean @xmath50 and unit variance . in fig .
1c we use @xmath266 , i.e. we simulate the conventional depinning model . in fig .
1d and 1e we use @xmath267 , and the grey dotted and dashed lines represent the stress averaged in small patches of size @xmath268 ( 100 blocks ) .
the two patches are chosen to be as far as possible . in fig .
2 we perform the simulation described in sm section 4 ( and in particular in sec .
we use @xmath269 and the thresholds are set constant @xmath270 .
the discretization of @xmath175 and @xmath271 is made using a binning @xmath272 . in fig .
3 and 4 of the main text and fig . 2 of the sm , we use @xmath273 .
@xmath55 is the uniform distribution in the range @xmath274 $ ] .
the @xmath265 are distributed as in fig .
the number of blocks is @xmath275 ( with @xmath276 ) . in fig .
4 we used @xmath277 . | in many complex systems a continuous input of energy over time can be suddenly relaxed in the form of avalanches .
conventional avalanche models disregard the possibility of internal dynamical effects in the inter - avalanche periods , and thus miss basic features observed in some real systems .
we address this issue by studying a model with viscoelastic relaxation , showing how coherent oscillations of the stress field can emerge spontaneously .
remarkably , these oscillations generate avalanche patterns that are similar to those observed in seismic phenomena .
the driven dynamics of heterogeneous systems often proceeds by random jumps called avalanches , which display scale - free statistics .
this critical out - of - equilibrium behaviour emerges from the competition between internal elastic interactions and interactions with heterogeneities , and is understood in the framework of the depinning transition @xcite .
remarkably , one can often disregard the precise details of the microscopic dynamics when considering the large scale behavior . as a result ,
various phenomena such as barkhausen noise in ferromagnets @xcite , crack propagation in brittle materials @xcite or wetting fronts moving on rough substrates @xcite may display similar avalanche statistics . in this description of avalanches a trivial dynamics in the inter - avalanche periods is usually assumed @xcite .
however , the inclusion of viscoelastic effects with their own characteristic time scales brings about novel dynamical features .
the existence of this kind of relaxation may have drastic consequences on the macroscopic behaviour of the system , as in the context of friction where it generates the time increase of static friction during the contact between two surfaces at rest @xcite . here
we show how these relaxation processes generically induce a novel avalanche dynamics characterized by new critical exponents and bursts of aftershocks strongly correlated in time and space . due to its simplicity , the model allows for analytic treatment in mean field , and for extensive numerical simulations in finite dimensions .
our main observations are twofold .
first , in mean field the time scale of viscoelastic relaxation is associated with a dynamical instability , which we prove to be responsible for periodic oscillations of the stress in the entire system .
this instability , named _
avalanche oscillator _ , was observed in numerical simulations and experiments of compression of nickel micro crystals @xcite .
note that viscoelastic interactions are also at the root of the _ hysteretic depinning _ emerging in mean field periodic systems like vortex lattice or charge density waves @xcite .
second , in two dimensions the global oscillations found in mean field remain coherent only on small regions . in each region the oscillations of the local stress have roughly the same amplitude and period but different phases , so that at a given time the stress map has a terraced structure .
we claim that the relaxation processes studied in our model are essential to capture the basic features of seismic dynamics .
in particular , the viscoelastic time scale is the one involved in the aftershock phenomenon @xcite .
moreover the oscillations of the stress field explain the quasi - periodic time recurrence of earthquakes that emerges from the data analysis of the seismic activity in some geographical areas @xcite .
finally we show that in two dimensions , viscoelastic relaxation produces an increase in the exponent of the avalanche size distribution compatible with the gutenberg - richter law , and the aftershock spatial correlations obtained have strong similarities with the _ migration effect _ observed in real earthquakes @xcite .
_ the models .
_ our model with relaxation is constructed upon the paradigmatic model of avalanche dynamics , describing the depinning of a @xmath0-dimensional elastic interface moving inside a @xmath1 dimensional space @xcite . in this model ,
the interface consists of a collection of blocks ( see fig . [ fig1_v0]a ) obeying the equation of motion : @xmath2 where @xmath3 is the @xmath1-dimensional coordinate of the block and @xmath4 is the viscosity of the medium .
each block feels elastic interactions via the ( discrete ) laplacian term @xmath5 ( summation is restricted to nearest neighbors of @xmath6 ) , disorder via @xmath7 and is driven towards the position @xmath8 via springs of elasticity @xmath9 .
the force per unit area applied by the drive , namely the stress is given by @xmath10 , where @xmath11 is the mean value of interface height ( @xmath12 ) .
the slow increase of @xmath13 over time induces an augmentation of the pulling force on each block . as a response ,
blocks typically adjust slightly their positions , but sometimes a block reaches a mechanically unstable state and moves far away from its position to a new local energy minimum .
this can in turn destabilize neighbouring blocks , thus triggering an avalanche event that we characterize by its size @xmath14 , which is simply the volume swept by the interface during the event . in fig .
[ fig1_v0]c we show the sizes @xmath15 of a sequence of avalanches obtained by driving @xmath13 quasi - statically ( @xmath16 ) .
the sequence displays an almost poissonian behavior , in the sense that both the sizes and the time occurrences of the events are almost uncorrelated variables .
moreover the stress is constant in time , with fluctuations due to finite system size .
our modified model consists in replacing springs @xmath17 by viscoelastic elements , built using springs and dashpots as depicted in fig .
[ fig1_v0]b .
dynamical equations become : @xmath18 where the auxiliary variables @xmath19 depend on the elongation of the neighbouring dashpots : in one dimension this variable reads @xmath20 ( see supplemental material ) .
the relaxation constant @xmath21 sets a new characteristic time @xmath22 , to be compared with the two scales : ( i ) @xmath23 which accounts for the slow increase of the external drive @xmath13 ( where @xmath24 is the typical microscopic disorder length scale , defined later ) , ( ii ) @xmath25 $ ] , which is the response time of the @xmath26 variables .
essentially , `` main '' avalanches are triggered by the drive through @xmath9 , whereas relaxation ( via @xmath27 ) triggers additional events on a time scale of order @xmath28 : the aftershocks .
in our analysis , we assume that the three scales are well separated , namely @xmath29 ( i.e. @xmath30 ) .
hence , on the time scale @xmath31 the @xmath19 s are constant in time and the dynamics is exactly the same as for the depinning model with elastic constant @xmath32 .
however , after an avalanche , and in a time scale @xmath33 , @xmath34 s are pinned ( due to the narrow wells approximation , see below ) and @xmath19 s relax exponentially : @xmath35 where @xmath36 is the time at which the last avalanche occurred .
the effect of relaxation is to suppress the term @xmath37 in eq .
( [ 2 ] ) , so that some blocks may become unstable .
this triggers secondary avalanches in the system , identified with aftershocks in the seismic context .
aftershocks occur without any additional driving : the ensemble of events that occur at a given @xmath13 will be called a _ cluster _
( fig .
[ fig1_v0]d ) .
when @xmath38 @xmath39 , the system is _ fully relaxed _ and
new instabilities can only be triggered by an increase of @xmath13 .
note that the fully relaxed configuration corresponds to a stable configuration of the depinning model with same disorder realization and elastic constants @xmath40 .
we mostly have in mind the @xmath41 case , which is the one realized in the seismic context .
we consider local elastic interactions for implementation convenience .
realistic long range interactions within the faults ( induced by the three dimensional nature of the plates ) may have effects on the results we present , that are difficult to assess without a full numerical simulation .
this remains as a prospect for future work .
sketch of the models for @xmath42 . *
( a ) * conventional depinning model : disorder and elastic interactions acting on the blocks , located in @xmath43 , etc .
the disordered force derives from the pinning potential ( grey ) : @xmath44 . *
( b ) * depinning with relaxation : we introduce dashpots with relaxation constant @xmath21 and springs of stiffness @xmath45 .
numerical results showing sequences of avalanches sizes @xmath15 and stress ( grey ) as a function of drive @xmath13 for : the elastic depinning model in @xmath41 * ( c ) * , the depinning with relaxation in @xmath41 * ( d ) * ( dashed and dotted grey lines correspond to local stress in two distant regions ) , and in mean field * ( e)*. ] evolution of @xmath46 ( solid line ) and the stress @xmath47 ( lower panel ) computed from direct integration of the evolution equations .
( @xmath48 ) driving without any avalanche , linearly increasing stress ; ( @xmath49 ) driving with elastic - depinning avalanches , slower stress increase .
( @xmath50 ) global event : @xmath46 collapses to the depinning fixed point @xmath51 ( lower dashed curve ) and the stress drops to @xmath52 ( lower dashed line ) .
( @xmath53 ) relaxation closes the cycle back to stage ( @xmath48 ) without altering average stress .
, width=325 ] _ the narrow wells approximation . _ to efficiently study eqs .
( [ depinning ] , [ 2 ] ) , we adopt the so - called `` narrow wells '' approximation . in this scheme , disorder is modeled as a collection of narrow pinning wells representing impurities ( see fig .
[ fig1_v0]a ) , with spacings @xmath54 , distributed as @xmath55 and with average @xmath56 .
the maximum value of the pinning force at each well is denoted @xmath57 ( see the supplemental material ) .
thanks to this choice , the system state can be reduced to a variable @xmath58 . in the depinning model : @xmath59 as soon as @xmath60 , this site becomes unstable : an avalanche is triggered .
the avalanche evolution translates into simple rules for @xmath58 s . when @xmath61 , all blocks are stable and the avalanche is exhausted .
driving then follows until a new event is triggered . in our model ,
s read : @xmath62 the dynamics proceeds as before , with @xmath19 s kept constant during avalanches .
when the avalanche is exhausted , a slow relaxation of @xmath19 takes place ( eq .
( [ 2 ] ) ) .
this evolution can decrease @xmath58 s and thus trigger aftershocks . _
mean field analysis .
_ we analyze the mean field , fully connected model , which corresponds to replacing @xmath63 with @xmath64 in eqs .
( [ fth1 ] ) , ( [ fth2 ] ) . in this case , all sites are equivalent and the @xmath58 s are independent and identically distributed variables , characterized by the probability distribution @xmath65 which in general depends on the initial condition @xmath66 and on @xmath13 . in the supplemental material ,
we obtain the evolution of @xmath65 under an infinitesimal increase in @xmath13 for both models , for @xmath67 const . for the elastic depinning model ,
we show that this evolution has a fixed point reached within a finite increase in @xmath13 , at which @xmath65 is given by the function : @xmath68 where @xmath69 .
this indicates that the large time dynamics is stationary , and that the applied stress in the system is constant in time : @xmath70 .
further analysis shows that as long as @xmath71 the system displays avalanches bounded by a system - size independent cutoff : @xmath72 .
for example , at the fixed point ( [ pdelta ] ) we have @xmath73 , so that @xmath74 .
however if @xmath75 the system becomes unstable , with a _ global event _ , that involves a finite fraction of the system . for the model with relaxation ,
the evolution of @xmath65 is non stationary and displays oscillations in time . under a small increase in @xmath13 ,
two dynamical regimes are observed . on short times @xmath76 , sites that become unstable
move following the rules of a rigid elastic interface , with stiffness @xmath32 . on longer times
@xmath77 , during relaxation , the interface becomes more flexible ( stiffness @xmath17 ) , thus evolving towards the fixed point @xmath78 ( _ stages _ @xmath48 and @xmath49 in fig . [ figpdelta ] ) .
however when @xmath79 becomes larger than @xmath80 , the rigid interface is unstable so that a single global avalanche drives @xmath65 to the rigid fixed point @xmath81 ( _ stage _ @xmath50 in fig .
[ figpdelta ] ) .
finally , this state is altered by relaxation and a new cycle starts ( _ stage _
@xmath53 ) .
note that cyclic behaviour is independent of the details of the mean field model : e.g. fig .
[ fig1_v0]e corresponds to the case of randomly distributed thresholds @xmath82 .
the avalanche dynamics is different and displays aftershocks , but global events and stress oscillations are also present . a similar visco - elastic model ( with @xmath83 , periodic disorder and under constant force @xmath84 )
was originally introduced @xcite to model the hysteretic depinning observed in vortex lattices and charge density waves . in the fully connected approximation
a self - consistent calculation pointed out that the average velocity is multi - valued , yielding hysteretic behavior in a wide range of external stress .
this hysteresis echoes with the oscillations in fig .
2 . yet , in two dimensions , we see that constant velocity driving produces stress distributions that are not uniform , nor constant , and we get qualitative new results that were not obtained in the constant applied force case studied in @xcite ) . ) and just after ( bottom , @xmath85 ) it takes place , as a function of the cluster size @xmath86 . thus , in e.g. a compression experiment , one expects the average local variation of stress to vanish ( with undefined values of @xmath87 ) for small avalanches , and to saturate to a constant ( nonzero ) value ( with well defined values for @xmath87 ) for large avalanches . ] stress map in @xmath41 .
colors indicate stress levels , from high ( red ) to low ( blue ) .
upper left : stress map just before a large event , with the unstable region highlighted by a dashed line . from left to right and top to bottom : expansion of the affected area is seen to mainly spread ( black arrows ) around the initial main shock and the subsequent aftershocks ( small crosses ) .
affected regions have low chance to witness new large events , due to the low value of the local stress . ] _ two dimensional results .
_ for @xmath41 we must rely on the numerical implementation of eqs .
( [ fth2 ] , [ fidet ] ) via an efficient method originally developed in @xcite ( see supplemental material ) . in fig .
[ fig1_v0]d , we see a clear distribution of events in clusters of main shocks and aftershocks , as in actual seismicity ( where , indeed , any single cluster spans a finite @xmath13 interval , due to the incomplete separation of time scales ) .
the periodicity of the mean field ( fig .
[ fig1_v0]e ) has now disappeared .
nevertheless , a careful analysis of the @xmath41 model shows an interesting reminiscence of the mean field behaviour . in fig .
[ sasb ] we compute for each cluster the local stress restricted to the cluster area , just before ( @xmath88 ) and just after ( @xmath85 ) it takes place ( the same analysis for events instead of clusters yields the same results ) .
small clusters show broad distributions of @xmath88 and @xmath85 , similar to what would be observed for the depinning case .
however for large clusters both distributions become very narrow : @xmath88 sets to a value that we denote @xmath89 , and @xmath85 sets to @xmath90 .
this is the fingerprint of the mean field behaviour , suggesting a large scale description of the @xmath41 interface as a terraced structure . indeed ,
we observe ( fig .
[ figbrokenzone ] ) that different parts of the system have different values of the stress , which range from @xmath90 to @xmath89 . in analogy with mean field ,
when the stress of a region reaches a value @xmath91 , it gets destabilized and the whole region collapses to @xmath90 . in fact the evolution of the local stress associated with a small patch of the interface is non stationary , and shows an almost periodic oscillation between @xmath90 and @xmath89 ( fig .
[ fig1_v0]d , dashed and dotted lines ) .
however this oscillation is not synchronized among different patches , so the system does not display a global oscillation .
it is remarkable that the width of the distribution of the local stress ( @xmath92 ) remains finite when @xmath93 , while in the depinning model @xcite , it vanishes as @xmath94 for very small @xmath9 ( @xmath95 is the roughness exponent of the interface which is found to be smaller than @xmath49 ) .
moreover our model supports the idea that seismic activity in some geographical regions displays quasi - periodicity ( the so - called seismic cycle @xcite ) .
this periodicity was recently studied in the context of micro - crystals deformation @xcite , where it was named avalanche oscillator " .
similar kind of oscillations were also observed in models with relaxation @xcite , granular materials @xcite and molecular dynamics of visco - elastic disordered systems @xcite .
a second important feature is the spatial distribution of aftershocks in a given cluster ( see fig .
[ figbrokenzone ] ) . after a main shock ,
many aftershocks follow , extending the slip area . the small ones ( not indicated )
are rather uniformly distributed inside the slip region ; while the epicentres of the large ones typically occur at the border , extending the slip region .
field observations report this effect as aftershock migration " @xcite . as a third point
, we discuss the size distribution of the avalanches in 2d , presented in fig . 2 of the supplemental material .
we find a consistent power law decay in all the range that we have been able to explore ( at least in a size range of @xmath96 ) with an _ anomalous _ exponent @xmath97 .
this is quite remarkable , given that in all conventional avalanche models like depinning or directed percolation , this exponent is always smaller than @xmath98 , which corresponds to the mean field limit @xcite .
in particular in the 2d depinning case one measures @xmath99 @xcite . our result can be compared with the value for actual earthquakes where where @xmath100 values in the range @xmath101 are reported .. the gutenberg - richter law ( gr ) states that @xmath102 , with reported values of @xmath103 in the range @xmath104 , so that from the definition @xmath105 , we obtain approximately @xmath106 .
] however , since we do not consider realistic long range elastic interactions , this coincidence has to be taken with caution .
we also note that a justification for the value of the gr exponent has been given recently using a forest fire model analogy @xcite .
it is worth mentioning that arguments in @xcite build upon a model that has a terraced structure of the interface compatible with the one we find here .
_ conclusions .
_ internal relaxation plays crucial roles in the dynamics of sliding objects , and becomes particularly important at large scales , relevant to seismic phenomena .
the dynamics of our model shows a strong tendency to become non - stationary .
this tendency is manifest in mean field , where sliding proceeds as a sequence of global and periodic stick - slips . in @xmath41 ,
we provide numerical evidence that periodic stick - slips occur locally , without global synchronization across the system . there
, our predictions mainly deal with spatial properties of events , thus demanding high spatial resolution in experiments .
we thank shamik gupta and mikhail b. zvonarev for useful discussions .
we acknowledge support from the france - argentina mincyt - ecos a12e05 .
e.a.j .
is financially supported by conicet ( argentina ) . partial support from grant pict-2012 - 3032 ( anpcyt , argentina ) is also acknowledged . 36 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , * * , ( ) . ,
* * , ( ) .
, , , , * * , ( ) . , , , , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , , * * , ( ) . , , , , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , , , , , * * , ( ) . , , , * * , ( ) .
( ) , . , * * , ( ) . ,
_ _ ( , ) . , * * , ( ) . , , , * * , ( ) . ,
* * , ( ) . , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , , , * * , ( ) .
( ) , . , * * , ( ) . , , ,
* * , ( ) .
, , , * * , ( ) . ,
* * , ( ) . , * * , ( ) . , , , , * * , ( ) . ,
* * , ( ) . , * * , ( ) .
* supplemental material * + |
topological objects are frequently described as the time - independent regular solutions possessing finite - energy arising from highly nonlinear euler - lagrange equations in the presence of appropriated boundary conditions @xcite . in some particular cases ,
the bogomolnyi - prasad - sommerfield ( bps ) formalism allows to show that these solutions can also satisfy a set of coupled first - order differential equations , the bps ones @xcite . in this sense , vortices are radially symmetric solitonic configurations appearing in a planar scalar scenario endowed by a gauge field , their energies being commonly proportional to the magnetic flux , both ones being quantized , i.e. proportional to an integer winding number . in the context of the classical field theory ,
these structures were firstly studied within the maxwell - higgs system , the self - interacting potential possessing no symmetric vacuum engendering topological bps vortex solutions only @xcite .
similar objects were also encountered in a chern - simons - higgs scenario , the corresponding theory giving rise to both topological and nontopological bps vortices @xcite .
furthermore , during the last years , many additional progress have been made regarding bps vortices arising within different theoretical scenarios , including generalizations of the abelian - higgs theories @xcite , lorentz - violating systems @xcite and gauged models with noncanonical kinetic terms @xcite , many of them being applied as an attempt to explain different cosmological and gravitational phenomena @xcite .
in particular , in view of the developments introduced in @xcite , the question on whether a gauged @xmath2 model with @xmath3 supports topological vortices arises in a rather natural way , mainly due to the straight relation between the @xmath2 theory and the four - dimensional yang - mills - higgs one , the first mapping some interesting properties of the second @xcite . in a very recent contribution @xcite
, the author have investigated topological vortices arising within a gauged @xmath2 model , the electromagnetic and the @xmath2 scalar fields interacting minimally , the dynamics of the gauge sector being controlled by the usual maxwell term . in that work , the author have considered only non - bps profiles , the corresponding solutions being obtained via the second - order euler - lagrange equations , from which the author have also studied interesting properties of the resulting structures , such as the way the energy and magnetic flux depend on the parameters of the overall model .
we now go a little bit further into such theme by considering the very same theoretical scenario , but now looking for a general first - order framework engendering radially symmetric bps vortices , whilst studying how these structures differ from their non - bps counterparts . in order to present our results ,
this manuscript is organized as follows : firstly , in the next section ii , we introduce the gauged @xmath2 model we study , presenting some basic definitions and conventions .
we particularize our investigation by focusing on the time - independent fields giving rise to radially symmetric configurations . in such a scenario ,
we look for a consistent first - order framework by manipulating the expression for the effective energy functional in order to establish a well - defined lower bound for the corresponding total energy ( here , it is important to point out that such construction is only possible when a particular constraint involving the self - dual potential is fulfilled ) .
we verify that this bound is saturated ( the energy being minimized ) when the profile functions describing the original fields satisfy a particular set of coupled first - order bps equations .
we also calculate general results for the energy bound and the magnetic flux inherent to the bps configurations . in the section iii ,
we investigate in detail the way the first - order expressions introduced previously generate legitimate self - dual solutions , whilst using the aforecited constraint to determine the potential term engendering self - duality , this way defining the vacuum manifold of the corresponding model .
then , we solve the bps equations numerically via convenient boundary conditions , from which we obtain regular configurations possessing finite energy , also identifying and commenting the main features these solitons engender . finally , in the sec .
iv , we present our conclusions and general perspectives regarding future contributions .
we highlight that , throughout all this manuscript , we use the natural units system together with @xmath4 for the planar metric signature .
we begin our investigation by considering the planar gauged @xmath2 model introduced in @xcite , its lagrange density reading@xmath5with @xmath6 being the usual electromagnetic field strength tensor , @xmath7 standing for a projection operator defined conveniently and @xmath8 representing the corresponding covariant derivative ( here , @xmath9 is a charge matrix , diagonal and real ) .
additionally , the @xmath2 field @xmath10 is constrained to satisfy @xmath11 . in our notation ,
greek indexes run over time - space coordinates , the latin ones counting the complex components of the @xmath2 field themselves .
it is instructive to write down the static gauss law coming from ( [ ai1 ] ) , i.e. ( here , @xmath12 runs over spatial coordinates only)@xmath13the charge density being@xmath14 \text{,}\]]where @xmath15 . in this sense , according the gauss law ( [ aig1 ] ) , one concludes that the static configurations that the model ( [ ai1 ] ) engenders have null electric charge , being then compatible with the gauge condition @xmath16 which satisfies ( [ aig1 ] ) identically . in this work , for the sake of simplicity , we consider the @xmath17 case , this way reducing our study to the @xmath0 scenario . in such a context , we look for time - independent solutions with radial symmetry . in this sense
, we guide our calculations using the standard map ( with winding numbers @xmath18 , @xmath19 and @xmath20 ) @xmath21@xmath22with @xmath23 and @xmath24 being the bidimensional levi - civita tensor ( @xmath25 ) and the position unit vector , respectively .
it is worthwhile to note that , once we are interested in regular solutions presenting no divergences , the profile functions @xmath26 and @xmath27 must obey@xmath28as we demonstrate below , we use these conditions to obtain well - behaved field solutions related to ( [ ai1 ] ) .
we highlight that , regarding the combination between the charge matrix @xmath29 and the winding numbers in ( [ li1 ] ) , there are two possibilities engendering topological solitons : ( i ) @xmath30 and @xmath31 , and ( ii ) @xmath32 and @xmath33 ( both ones with @xmath34 , @xmath35 and @xmath36 being the diagonal gell - mann matrices , i.e. @xmath37diag@xmath38 and @xmath39diag@xmath40 ) .
nevertheless , in @xcite , the author have demonstrated that these two combinations simply mimic each other , this way existing only one effective scenario .
therefore , in this work , we investigate only the case defined by @xmath31 , @xmath34 and@xmath41for convenience . the differential equation for the profile function @xmath1 reads@xmath42giving rise to two constant solutions , i.e. @xmath43with @xmath44 . a priori , these solutions define two different cases .
however , as we demonstrate later below , these cases engender the very same phenomenology , at least when concerning the bps results at the classical level .
it is important to say that , from this point on , our expressions effectively describe the scenario defined by the choices we have introduced above .
we focus our attention on those solutions satisfying a particular set of coupled first - order equations . in this work ,
we obtain these equations by following the canonical approach , i.e. by requiring the minimization of the total energy of the system , the starting - point being the expression for the energy - momentum tensor related to the model ( [ ai1 ] ) , i.e.@xmath45from which one gets the radially symmetric energy density ( @xmath46)@xmath47 + v\text { , } \label{ai2}\]]where we have introduced the auxiliary function@xmath48the solution for @xmath1 being necessarily one of those stated in ( [ betha ] ) .
the point to be raised in that , whether we consider the very fundamental constraint@xmath49the energy density ( [ ai2 ] ) can be rewritten in the form@xmath50 ^{2}+\varepsilon _ { bps}\text { , }
\label{ai3}\]]where we have defined@xmath51 \text{.}\ ] ] equation ( [ ai3 ] ) shows that the corresponding energy exhibits a well - defined lower bound ( a property inherent to such a first - order construction ) , this bound being saturated when the profile functions @xmath52 and @xmath27 obey@xmath53@xmath54which are the effective bps equations the model ( [ ai1 ] ) engenders . in this sense ,
when ( [ ai4 ] ) and ( [ ai5 ] ) are satisfied , the total energy of the resulting bps configurations can be calculated explicitly , reading ( the upper ( lower ) sign holding for negative ( positive ) values of @xmath55)@xmath56being quantized in terms of @xmath55 itself . here
, we have supposed that @xmath57 vanishes for @xmath58 ( indeed , @xmath59 stands for a finite - energy condition ) , with @xmath60 finite .
another quantity commonly referred when studying first - order vortices is the magnetic flux @xmath61 they exhibit . in the present case ,
the resulting flux reads@xmath62where @xmath63 must be chosen in order to fulfill the finite - energy requirement @xmath64 . here , we have used @xmath65 , prime denoting derivative with respect to the radial coordinate @xmath66 . in the next section ,
we use the first - order expressions we have introduced above to generate regular bps solutions with finite - energy .
it is important to say that the self - dual equations ( [ ai4 ] ) and ( [ ai5 ] ) , together with the fundamental constraint ( [ ai0 ] ) , solve the second - order euler - lagrange ones coming from ( [ ai1 ] ) , therefore providing genuine solutions of the model .
now , we investigate the solutions the first - order framework we have developed provides . here , we proceed as follows : firstly , we choose one particular solution for @xmath1 coming from ( [ betha ] ) . in the sequel
, we use such a solution to solve the differential constraint ( [ ai0 ] ) , from which we obtain the corresponding self - dual potential @xmath67 related to that particular case .
a posteriori , we implement these both solutions ( i.e. for @xmath1 and @xmath67 ) into the general expression for the energy density ( [ ai2 ] ) , this way getting the asymptotic boundary conditions the profile functions @xmath26 and @xmath27 must satisfy in order to generate finite - energy structures .
then , using these conditions and the ones in ( [ ai00 ] ) , we solve the resulting first - order equations ( ai4 ) and ( [ ai5 ] ) numerically , obtaining the corresponding bps solutions for @xmath26 and @xmath27 .
finally , we depict these solutions and the physical profiles they engender ( energy density and magnetic field ) , also calculating their total energies and magnetic fluxes explicitly , from which we compare the final scenarios .
then , let us consider the cases @xmath68 and @xmath69 separately .
this case was partially considered in @xcite , the respective author suggesting that such construction was possible . here , we go a little bit further in such a investigation , showing that the general bps framework we have introduced recovers the expressions proposed in that work ( additionally reinforcing the coherence of our construction ) .
we start by choosing@xmath71from which one gets that @xmath72 , the constraint ( [ ai0 ] ) reducing to@xmath73whose solution is ( we have used @xmath74 for the integration constant)@xmath75i.e . the self - dual potential related to the @xmath76 case
, the resulting bps equations ( [ ai4 ] ) and ( [ ai5 ] ) standing for@xmath77@xmath78 now , in order to solve the equations ( [ v12 ] ) and ( [ v13 ] ) , beyond the behavior in ( [ ai00 ] ) , we need to specify the conditions the profile functions @xmath26 and @xmath27 obey in the asymptotic limit ( these conditions ensuring that the total energy of the final configurations is finite ) . in this sense , we implement ( [ v10 ] ) and ( [ v1 ] ) into the basic expression ( [ ai2 ] ) , from which we get@xmath79 + \frac{g^{2}h^{2}% } { 2}\cos ^{2}\alpha \text{,}\]]with@xmath80the corresponding energy being finite for @xmath81 , @xmath26 and @xmath82 being then constrained to satisfy@xmath83with @xmath84 . at this time
, one can easily calculate the total energy @xmath85 ( [ ai6 ] ) and the magnetic flux @xmath61 ( [ ai7 ] ) of the resulting bps solitons , i.e.@xmath86both ones being quantized in terms of the winding number @xmath55 , as expected .
it is also interesting to investigate the way the fields @xmath87 and @xmath27 approximate the boundary values in ( ai00 ) and ( [ v14 ] ) . in order to perform such calculation , in what follows , we consider @xmath88 only ( i.e. the lower signs in the self - dual expressions ) , for simplicity .
then , following the usual algorithm , we linearize the bps equations ( [ v12 ] ) and ( [ v13 ] ) in order to get the approximate solutions near the origin ( with @xmath89)@xmath90and in the asymptotic regime@xmath91@xmath92 being the masses of the corresponding bosons ( the relation @xmath93 typically defining the bogomolnyi limit ) , @xmath94 and @xmath95 standing for positive constants to be fixed by requiring the correct behavior near the origin and asymptotically .
we summarize the scenario as follows : given the profile functions @xmath52 and @xmath27 satisfying the equations ( [ v12 ] ) and ( [ v13 ] ) , and obeying the boundary conditions ( [ ai00 ] ) and ( v14 ) , the resulting radially symmetric bps vortices exhibit quantized energy and magnetic flux given by @xmath96 and @xmath97 , respectively , whilst approaching the boundaries according the approximate solutions in ( [ v15 ] ) and ( [ v16 ] ) .
it is interesting to note that , whether we implement @xmath89 , the potential in ( [ v1 ] ) can be written as @xmath98 , mimicking exactly the eq .
( 15 ) in @xcite .
moreover , in that same work , the author suggested that the choice @xmath99 does not support self - dual solutions .
nevertheless , as we will demonstrate below , once we choose the self - dual potential conveniently , the case with @xmath100 indeed admits coherent bps solitons .
now , we go further in our investigation by choosing@xmath102whilst giving rise to @xmath103 , the corresponding constraint being@xmath104engendering the solution@xmath105which stands for the self - dual potential related to @xmath106 . here
, we have used @xmath107 for the integration constant . in this case , the bps equations read@xmath108@xmath109 in order to determine the conditions the fields @xmath26 and @xmath27 satisfy in the asymptotic limit , we proceed as before , i.e. given ( [ carai ] ) and ( [ v17 ] ) , the expression ( [ ai2 ] ) for the energy density reduces to@xmath110 + \frac{g^{2}h^{2}% } { 32}\cos ^{2}\left ( 2\alpha \right ) \text{,}\]]where@xmath111 ( solid black line for @xmath112 and dashed red line for @xmath113 ) and @xmath27 ( dot - dashed blue line for @xmath112 and long - dashed orange line for @xmath113 ) obtained via the bps equations ( [ v12 ] ) and ( [ v13 ] ) in the presence of the boundary conditions ( [ ai00 ] ) and ( [ v14 ] ) , for @xmath114.,width=321 ] then , in order to fulfill the finite - energy requirement @xmath64 , @xmath26 and @xmath115 must behave as@xmath116from which we can also calculate the energy @xmath85 and the magnetic flux @xmath117 inherent to the resulting structures , i.e.@xmath118which are again quantized in terms of @xmath55 .
furthermore , we look for the way @xmath26 and @xmath82 behave near the boundaries by linearizing the first - order equations ( [ 2a ] ) and ( [ 2b ] ) around the contours values in ( [ ai00 ] ) and ( [ xx1 ] ) ( again , for simplicity , we use @xmath88 ) , from which one gets the approximate profiles near the origin@xmath119and in the limit @xmath58 , i.e.@xmath120with @xmath121 standing for the masses of the related particles ( @xmath122 still holding ) , @xmath123 and @xmath124 being positive constants to be fixed via the same manner as before .
in such a scenario , given the conditions ( [ ai00 ] ) and ( [ xx1 ] ) , the solutions @xmath26 and @xmath27 that the equations ( [ 2a ] ) and ( [ 2b ] ) provide generate self - dual vortices possessing total energy @xmath125 and magnetic flux @xmath126 , behaving as the approximate profiles ( [ xx2 ] ) and ( xx3 ) in the appropriate limit .
it is interesting to note that the potential in ( [ v1 ] ) can be written as the one in ( [ v17 ] ) whether we implement the redefinitions @xmath127 , @xmath128 and @xmath129 . notwithstanding , also the bps equations ( [ v12 ] ) and ( [ v13 ] ) reduce to those in ( [ 2a ] ) and ( [ 2b ] ) following the very same way ( the corresponding energies behaving in a similar manner , the magnetic fluxes being automatically the same ) .
we highlight that such redefinitions are completely compatible also at level of the second order euler - lagrange equations .
in this sense , we argue that the @xmath100 case mimics those results obtained via @xmath68 , both scenarios effectively describing the same phenomenology , at least concerning the bps solitons at the classical level . for @xmath130 ( solid black line ) , @xmath131 ( dot - dashed blue line ) and @xmath132 ( dashed red line ) .
the profiles are lumps centered at the origin.,width=321 ] we end this section by presenting the numerical solutions we have found for @xmath133 , @xmath27 , @xmath134 and @xmath135 via the bps equations ( [ v12 ] ) and ( [ v13 ] ) in the presence of the boundary conditions ( [ ai00 ] ) and ( v14 ) . we have used @xmath114 , for simplicity
. , conventions being the same as in the previous figure 2 ( with the long - dashed orange line for @xmath113 ) : for @xmath112 , the solution is a lump ; for @xmath136 , the profiles are rings.,width=321 ] in the figure 1 , we depict the numerical solutions to the profile functions @xmath133 ( solid black line for @xmath112 and dashed red line for @xmath113 ) and @xmath27 ( dot - dashed blue line for @xmath112 and long - dashed orange line for @xmath113 ) . in general , these profiles reach the boundary values in a monotonic manner , behaving according the approximate solutions we have calculated previously via the linearization of the bps equations
. we also point out the way the gauge field coherently fulfills the condition @xmath137 , as expected .
we plot the solutions to the magnetic field @xmath134 in the fig .
2 , for @xmath112 ( solid black line ) , @xmath131 ( dot - dashed blue line ) and @xmath132 ( dashed red line ) , all these profiles being lumps centered at the origin ( the absolute value does not depending on the winding number @xmath55 , being equal to the unity ) . also , as the vorticity increases , the solutions spread over greater distances , the corresponding bosons mediating large - range interactions . finally , the fig .
3 shows the profiles to the bps energy density @xmath135 , the conventions being the same ones used in the figure 2 . here , for @xmath112 , the resulting solution is a lump centered at the origin . on the other hand , for @xmath136 ,
the corresponding profiles are rings , their absolute values ( amplitudes ) lying on a finite distance @xmath138 from the origin ( this way defining the `` radius '' of the ring ) . in particular , these amplitudes ( radii ) decrease ( increase ) as the vorticity increases .
in this work , we have considered the gauged @xmath0 model proposed in loginov , whilst obtaining in a coherent way the self - dual bps vortices such model supports .
firstly , we have detailed the planar lagrange density defining the overall model , from which we have verified that such a system supports static solutions presenting no electric field ( @xmath16 satisfying the corresponding gauss law identically ) . in the sequel ,
focusing our attention on those configurations presenting radial symmetry , we have implemented the well - known ansatz ( [ li0 ] ) and ( [ li1 ] ) together with convenient choices for the charges and winding numbers inherent to such maps .
we have verified the constant solutions the profile function @xmath1 supports , this way splitting our investigation in a priori two different branches .
in addition , whilst using a very fundamental constraint , we have rewritten the radially symmetric expression for the effective energy density in such a way highlighting the existence of a lower bound for the corresponding total energy .
we have checked that the aforementioned bound is saturated when the functions @xmath26 and @xmath139 satisfy a particular set of coupled first - order bps equations , also calculating the symbolic values for the total energy and the magnetic flux the final structures exhibit .
we have divided our study according the two solutions @xmath140 admits .
then , whilst considering these two cases separately , we have used the fundamental constraint introduced before to obtain the particular self - dual potentials such scenarios engender , also getting the corresponding bps equations and the appropriate boundary conditions . in view of such conditions ,
we have calculated explicitly the values for the energy and the magnetic flux inherent to the resulting bps structures .
moreover , we have concluded that the two solutions @xmath1 supports in fact engender the very same phenomenology , being physically equivalent .
finally , we have plotted the numerical solutions we found for the relevant profiles , from which we have commented the general aspects they present . in this sense ,
rather natural ideas regarding new works include the search for the nontopological bps vortices the theoretical model we have studied here possibly engenders .
furthermore , it is also interesting to consider the gauged @xmath0 theory when endowed by the chern - simons term ( instead of the maxwell one ) .
these two possibilities are now under investigation , and we hope positive results for a future contribution .
d. bazeia , e. da hora , c. dos santos and r. menezes , phys .
d * 81 * , 125014 ( 2010 ) ; eur .
j. c * 71 * , 1833 ( 2011 ) .
d. bazeia , r. casana , m. m. ferreira jr . and e. da hora , europhys . lett .
* 109 * , 21001 ( 2015 ) .
r. casana , e. da hora , d. rubiera - garcia and c. dos santos , eur .
j. c * 75 * , 380 ( 2015 ) .
r. casana , m. m. ferreira jr . ,
e. da hora and c. miller , phys .
b * 718 * , 620 ( 2012 ) .
r. casana , m. m. ferreira jr .
, e. da hora and a. b. f. neves , eur .
j. c * 74 * , 3064 ( 2014 ) .
r. casana and g. lazar , phys .
d * 90 * , 065007 ( 2014 ) .
r. casana , c. f. farias and m. m. ferreira jr .
d * 92 * , 125024 ( 2015 ) .
r. casana , c. f. farias , m. m. ferreira jr . and g. lazar , phys .
d * 94 * , 065036 ( 2016 ) . c. armendariz - picon ,
t. damour and v. mukhanov , phys .
b * 458 * , 209 ( 1999 ) . v. mukhanov and a. vikman , j. cosmol .
* 02 * , 004 ( 2005 ) .
a. sen , j. high energy phys .
* 07 * , 065 ( 2002 ) .
c. armendariz - picon and e. a. lim , j. cosmol .
. phys . * 08 * , 007 ( 2005 ) .
j. garriga and v. mukhanov , phys .
b * 458 * , 219 ( 1999 ) . r. j. scherrer , phys .
lett . * 93 * , 011301 ( 2004 ) .
a. d. rendall , class .
quantum grav .
* 23 * , 1557 ( 2006 ) .
m. a. mehta , j. a. davis and i. j. r. aitchison , phys .
b * 281 * , 86 ( 1992 ) .
b. m. a. g. piette , d. h. tchrakian and w. j. zakrzewski , phys .
b * 339 * , 95 ( 1994 ) .
d. h. tchrakian and k. arthur , phys .
b * 352 * , 327 ( 1995 ) .
l. sourrouille , a. caso and g. s. lozano , mod .
a * 26 * , 637 ( 2011 ) .
l. sourrouille , mod .
a * 26 * , 2523 ( 2011 ) ; mod .
a * 27 * , 1250094 ( 2012 ) .
a. dadda , m. luscher and p. d. vecchia , nucl .
b * 146 * , 63 ( 1978 ) .
e. witten , nucl .
b * 149 * , 285 ( 1979 ) .
a. m. polyakov , phys .
b * 59 * , 79 ( 1975 ) . m. shifman and a. yung , rev .
phys . * 79 * , 1139 ( 2007 ) . | we study time - independent radially symmetric bps solitons in a @xmath0 model interacting with an abelian gauge field whose dynamics is controlled by the usual maxwell term . in this sense , we develop a consistent self - dual framework verifying the existence of a well - defined lower bound for the corresponding energy .
we saturate such a lower bound by focusing on those solutions satisfying a particular set of coupled first - order differential equations , the bps ones .
we solve the bps equations numerically using appropriate boundary conditions giving rise to regular structures possessing finite - energy .
we also comment the main features these configurations exhibit .
moreover , we highlight that , despite the different solutions we consider for an auxiliary function @xmath1 labeling the model ( therefore splitting our investigation in two a priori distinct branches ) , all resulting scenarios engender the very same phenomenology , being physically equivalent . |
the tremendous progress in experimental flavour physics facilitated by the @xmath0 factories , the tevatron and other experiments shows that the ckm picture of flavour and cp violation describes the data well , both for tree and loop - induced processes . since the standard model ( sm ) is only an effective theory valid up to some energy scale @xmath4 , which is yet to be determined but is expected to lie in the tev region from naturalness considerations , this success is a challenge for theories of new physics ( np ) . assuming generic flavour and cp violating couplings , the bounds on @xmath4 from @xmath5 processes exceed 10000 tev @xcite . on the other hand ,
given that a theoretical understanding of the hierarchical structure of quark masses and ckm mixings is still lacking , this paradox might simply indicate that the flavour structure of np is related to the sm one .
if in fact the yukawa couplings , already present in the sm , are the _ only _ sources of breaking of the flavour symmetry , theories with this property of minimal flavour violation ( mfv , @xcite ) can have tev - scale dynamics without being in conflict with observations @xcite .
then again , mfv might well be too restrictive an assumption .
concrete models aiming to explain the origin of flavour often bring about a certain amount of non - minimal flavour violation ( see _ e.g. _ @xcite ) .
furthermore , the mfv principle does not preclude the presence of flavour - blind cp violating phases @xcite .
therefore , testing whether indeed the yukawa couplings are the only source of flavour breaking and testing whether indeed the ckm phase is the only source of cp breaking are among the major goals of flavour physics in this decade .
the experimental prospects for progress in this direction are excellent , in view of planned dedicated flavour physics experiments like the super @xmath0 factories and searches for electric dipole moments ( edms ) or rare @xmath6 decays ; moreover , in the near future , important results on @xmath7 transitions are expected from the lhc .
[ sec:1][sec:3 ] will therefore discuss three particularly promising probes of np at the lhc : the rare leptonic decays @xmath1 , cp violation in @xmath3 mixing , and angular observables in the @xmath2 decay .
[ sec:4 ] contains a brief summary of a model leading to interesting signatures in precision flavour experiments : supersymmetry with hierarchical squark masses , effective mfv and flavour - blind phases .
.sm predictions and 95% c.l .
experimental upper bounds on the @xmath8 branching ratios @xcite . [ cols="<,<,<",options="header " , ] the decays @xmath1 are strongly helicity suppressed in the sm and have not been experimentally observed yet ; the current upper bounds still lie one / two orders of magnitude above the sm predictions ( see table [ tab : bmumu ] ) but will be improved in the near future by the tevatron and lhc experiments . concerning np effects in @xmath9 , in models where only the sm wilson coefficient @xmath10 receives np contributions , an order - of - magnitude enhancement of the branching ratio is disfavoured due to constraints from inclusive and exclusive @xmath11 transitions on @xmath10 .
in that case , br@xmath12 @xcite .
much larger enhancements are possible in principle in models with contributions to the scalar and/or pseudoscalar wilson coefficients @xmath13 . in two - higgs - doublet models
, a neutral higgs penguin contributes to the branching fraction with an enhancement factor of @xmath14 , where @xmath15 is the ratio of higgs vevs . in the mssm , this dependence is even @xmath16 .
consequently , an upper bound @xmath17 would already constrain numerous well - motivated np scenarios , such as susy guts with a unification of yukawa couplings @xcite . would remain a valid possibility if the trilinear couplings are small , such as in gauge mediation scenarios @xcite . ] an important test of mfv is represented by the measurement of the ratio of the two @xmath8 branching ratios . in mfv ( as defined in @xcite ) , it is simply given by @xmath18 while a simultaneous enhancement over the sm satisfying ( [ eq : sd ] ) would be a strong indication in favour of mfv , a non - sm effect incompatible with ( [ eq : sd ] ) would immediately rule out mfv . and @xmath19 in mfv , the sm4 and four susy flavour models .
the gray area is ruled out experimentally .
the sm point is marked by a star .
taken from @xcite.,scaledwidth=75.0% ] figure [ fig : bqmumu ] ( for details see @xcite ) shows possible values of this ratio attained in several non - mfv models : four susy flavour models studied in @xcite and the sm with 4 generations @xcite .
this highlights the power of the correlation between @xmath9 and @xmath19 to discriminate between different np models .
in the sm , cp violation in @xmath3 mixing is a small effect since the relevant combination of ckm elements has an accidentally small phase , @xmath20 recently however , two experimental hints for a possibly large non - sm contribution to @xmath21 have emerged .
one concerns the mixing - induced cp asymmetry @xmath22 extracted from the time - dependent cp asymmetry in @xmath23 decays , @xmath24 where @xmath25 .
the other concerns the charge asymmetry @xmath26 in dimuon events at d0 , which can be related to the semileptonic cp asymmetries in flavour - specific @xmath27 and @xmath3 decays , @xmath28 , as @xcite @xmath29 with @xmath30 uncertainties on the coefficients on the right - hand side of ( [ eq : asl ] ) .
while 2009 results on @xmath22 showed a discrepancy with the sm somewhere in the ballpark of 3 standard deviations @xcite , 2010 updates seem to be in agreement with the sm at the @xmath31 level @xcite , although no combination has been performed yet .
the d0 result on @xmath26 deviates by @xmath32 from the sm @xcite , interestingly pointing in the same direction as the possible effect in @xmath22 from the lhcb experiment were presented @xcite also showing a preference for an effect in the same direction , although still with a small significance . ] .
if these hints turn out to be genuine signals of np in @xmath3 mixing , it would have far - fetching consequences , in particular for theories with mfv .
if a theory satisfying mfv does not have any source of cp violation beyond the ckm phase , a sizable @xmath3 mixing phase can not be generated . even if the mfv theory has flavour - blind phases , this is nontrivial .
while a two - higgs doublet model with mfv and flavour - blind phases can generate a sizable @xmath21 @xcite , this possibility is precluded in the mssm by the impact of constraints like @xmath33 and @xmath9 @xcite .
a confirmation of @xmath21 deviating significantly from the value in ( [ eq : phis ] ) would thus immediately rule out many well - motivated theories , including the mfv mssm . and the mixing - induced cp asymmetry @xmath22 in the sm4 , the two - higgs doublet model with flavour blind phases and three susy flavour models .
the sm point is marked by a star.,scaledwidth=72.0% ] another interesting tool to discriminate between np models is the correlation between @xmath22 and the branching ratio of @xmath9 . in many models ,
sizable deviations from the sm prediction for @xmath21 are tied to the presence of scalar currents , which can also affect @xmath9 .
this is the case _
e.g. _ for the two - higgs doublet model with mfv and flavour - blind phases ( 2hdm@xmath34 ) or for the susy flavour model of agashe and carone ( ac , @xcite ) . as shown in fig .
[ fig : spsiphi ] , sizable @xmath22 implies a sizable enhancement of br(@xmath9 ) in these models , while the converse is obviously not true . in the sm4 , on the other hand
, even the converse statement is true : if @xmath9 is found with a rate significantly enhanced with respect to the sm , this model unambiguously predicts a sizable deviation also in @xmath22 . in
yet other models , like in the @xmath35 model of @xcite , both observables can be enhanced independently of each other , but a simultaneous enhancement is unlikely .
the exclusive decay @xmath36 gives access to many observables potentially sensitive to np @xcite ; since it is a `` self - tagging '' decay , it also allows a straightforward measurement of cp asymmetries .
the decay poses several theoretical challenges .
in addition to calculating the 7 @xmath37 form factors , one has to estimate non - factorizable strong interaction effects . at intermediate values of the dilepton invariant mass
squared @xmath38 , resonant charmonium production leads to a breakdown of quark - hadron duality .
the most studied region is the low @xmath38 region , where qcd factorization can be used to calculate non - factorizable corrections @xcite and light - cone sum rules to calculate the form factors @xcite .
the high-@xmath38 region above the charmonium resonances has recently attracted increasing attention @xcite . while qcd factorization and lcsr methods are not applicable in this kinematical domain , a local operator product expansion in powers of @xmath39
allows a systematic calculation of the observables @xcite . in @xcite
, it has been argued that , in contrast to the low-@xmath38 region , non - perturbative corrections _ not _ accounted for by the form factors are small ( see also @xcite ) .
a complete set of observables accessible in the angular distribution of the decay and its cp - conjugate is given by the 9 cp - averaged angular coefficients @xmath40 and the 9 cp asymmetries @xmath41 , defined in terms of the angular coefficients @xmath42 as @xcite @xmath43 not all of the @xmath40 and @xmath41 are both theoretically interesting and experimentally promising .
in addition , at @xmath0 factories , at the tevatron and at the early lhc , statistics is quite limited so a full angular analysis is not possible and it might be easier to consider one- or two - dimensional angular distributions depending on a limited set of observables .
in addition to the observables observables . ] which have already been studied by babar @xcite , belle @xcite and cdf @xcite the differential branching ratio , the forward - backward asymmetry and the @xmath44 longitudinal polarization fraction two observables sensitive to np that could be extracted from a one - dimensional angular distribution are the cp - averaged @xmath45 and the cp - asymmetry @xmath46 . while they are both negligibly small in the sm , they could be nonzero in np models with right - handed currents @xcite .
two additional observables could be extracted from a two - dimensional angular distribution and might be accessible even during the early lhc running : the cp - averaged observable @xmath47 and the cp - asymmetry @xmath48 . in @xcite
, it has been shown that even with an integrated luminosity of only @xmath49 , lhcb can measure @xmath47 with a precision that already allows to probe certain np scenarios .
since @xmath48 is accessible from the same angular distribution , the naive expectation is that the sensitivity should be comparable . just as @xmath46 , @xmath48 is a t - odd cp asymmetry , meaning that it is not suppressed by small strong phases @xcite .
sizable effects in @xmath48 of up to 20% are expected in well - motivated np scenarios like the one discussed in the next section .
as mentioned in the introduction , the mfv principle provides a symmetry argument to explain the absence of any evidence for flavour violation beyond the sm , but does not address the absence of signals of cp violation beyond the ckm phase , _
e.g. _ in edms . in supersymmetry , a different way to address
the flavour problem is to assume a strong hierarchy between the squarks of the first two versus the ones of the third generation : the heaviness of the former helps evading the strong bounds from @xmath6 physics , while the lightness of the latter preserves the susy solution to the gauge hierarchy problem . for a generic flavour structure of the soft susy breaking terms
, this hypothesis is however by far insufficient to ensure the absence of excessive flavour violation and needs to be extended by some amount of flavour alignment ( see @xcite and references therein ) . in ref .
@xcite , the hierarchical sfermion idea was combined with a specific flavour symmetry breaking pattern inspired by mfv .
the starting point is a global flavour symmetry in the quark sector of the form @xmath50 where @xmath51 acts like baryon number , but only for the right - handed up - type and the left - handed ( s)quark fields .
@xmath52 is subsequently broken down to baryon number by a single spurion , the down - type yukawa coupling @xmath53 . as a consequence ,
it was shown in @xcite that the theory is essentially mfv - like at low energies ( therefore the name `` effective mfv '' ) , with the most significant constraint coming from cp violation in neutral kaon mixing .
interestingly , hierarchical sfermions also ameliorate the susy cp problem compared to mfv since the one - loop contributions to the experimentally accessible edms involve the superpartners of first generation fermions .
therefore , in ref .
@xcite the effective mfv ( emfv ) framework was analyzed allowing all the cp violating phases not forbidden by the flavour symmetry . with hierarchical squark masses , the usual argument that flavour blind phases
need to be tiny to meet the edm bounds does not apply in emfv .
the main results of @xcite can be summarized as follows , * the one - loop contributions to the edms are under control if the first generation up - squark and sneutrino masses fulfill the following bounds , @xmath54 * two - loop contributions from barr - zee type diagrams not suppressed by first or second generation sfermion masses give contributions to the electron edm that are in the ballpark of the experimental bound in the case of a sizable phase in the @xmath55 term , and are usually below it if the stop trilinear coupling ( but not @xmath55 ) is complex . *
even if the edms are under control , potentially visible effects can arise in cp asymmetries in @xmath0 physics , induced by one - loop contributions involving third generation sfermions to the magnetic and chromomagnetic @xmath7 dipole operators . and
the angular cp asymmetry @xmath56 in @xmath2 in two scenarios with a complex @xmath55 term ( left ) or complex @xmath57 term ( right ) .
the gray points are allowed by all constraints except @xmath58 , while the blue points are compatible with all constraints .
the orange points in the left - hand plot have @xmath59.,title="fig:",scaledwidth=48.0% ] and the angular cp asymmetry @xmath56 in @xmath2 in two scenarios with a complex @xmath55 term ( left ) or complex @xmath57 term ( right ) .
the gray points are allowed by all constraints except @xmath58 , while the blue points are compatible with all constraints .
the orange points in the left - hand plot have @xmath59.,title="fig:",scaledwidth=46.5% ] as an example for the possible effects in @xmath0 physics , fig . [
fig : svsa ] shows the correlation between the mixing - induced cp asymmetry in @xmath60 and the angular cp asymmetry @xmath48 ( integrated in the low @xmath38 range ) in @xmath2 discussed in sect .
[ sec:3 ] for two scenarios : first , assuming an arbitrary phase of the @xmath55 term , second , assuming an arbitrary phase for the stop trilinear coupling .
in particular in the second scenario , sizable effects in the cp asymmetries can be generated without violating the edm bounds term and a real @xmath55 term @xcite .
of course , the two setups are easily distinguishable on the basis of their different spectrum . ] .
this study provides an example of a model with interesting signatures in flavour physics observables and also highlights the importance of dedicated flavour physics experiments , like edm searches or super @xmath0 factories , as complements to the lhc .
the results are applicable to all mssm scenarios with hierarchical sfermions ( see _ e.g. _ @xcite ) , with possible additional effects in the presence of non - minimal flavour violation .
the lhc era is in full swing , also in flavour physics .
imminent results on decays like @xmath9 or @xmath2 and on cp violation in @xmath3 mixing discussed in this talk will help shed light on the question whether there is new physics at the tev scale , and subsequently whether the yukawa couplings are the only source of breaking of the flavour and cp symmetries , as in the sm .
supersymmetry with hierarchical squark masses , combined with effective mfv @xcite , can solve the susy flavour _ and _ cp problems and leads to potentially visible effects in cp asymmetries in @xmath0 physics , without violating edm bounds . | the origin of flavour and cp violation is among the most important open questions in particle physics .
imminent results from the lhc as well as planned dedicated flavour physics experiments might help to shed light on this puzzle .
this talk concentrates on the np sensitivity of the rare @xmath0 decays @xmath1 and @xmath2 and of the cp violating phase in @xmath3 mixing .
a brief summary of a supersymmetric model with interesting signatures in the flavour sector is presented . |
twisted calabi - yau algebras , that is , algebras @xmath0 that satisfy a twisted poincar duality in hochschild ( co)homology , have been the object of intense recent study due to their natural prevalence in various flavours of noncommutative geometry , see e.g. @xcite . more precisely , such algebras have isomorphic hochschild homology and cohomology up to a twist in the coefficients given by some algebra automorphism @xmath6 , @xmath7 this generalises ginzburg s notion of a calabi - yau algebra @xcite which cover the case where @xmath2 .
see the main text for notation and precise definitions . by analogy with the situation for poisson manifolds @xcite or in tomita - takesaki theory
@xcite it is natural to expect that the smash product @xmath8 is an ( untwisted ) calabi - yau algebra .
some partial results in that direction have been obtained in , or can be derived from @xcite . however , to the best of our knowledge no general proof of this fact has been obtained yet . specifically , we make no assumptions about the existence of a grading on @xmath0 , or about properties of the automorphism @xmath9 , e.g. finiteness of order or more generally , semisimplicity .
hence the aim of the present paper is to prove : [ maintheorem ] let @xmath0 be a @xmath3-twisted calabi - yau algebra of dimension @xmath10
. then @xmath4 and @xmath5 are calabi - yau algebras of dimension @xmath11 .
the structure of the paper is as follows : in the first section , we provide some background on the hochschild ( co)homology of algebras and the cup and cap products . in section 2
, we recall a theorem of van den bergh on poincar type duality in hochschild ( co)homology and define twisted calabi - yau algebras .
the third section describes twisted cyclic homology and the underlying notion of a paracyclic module of which the hochschild complex @xmath12 of an algebra @xmath0 with coefficients twisted by an algebra automorphism @xmath3 is an example .
we also show , as a consequence of the basic paracyclic theory , that the hochschild homology @xmath13 is invariant under the natural action of @xmath3 for any algebra @xmath0 and @xmath1 which implies the invariance of the cohomology @xmath14 of a twisted calabi - yau algebra under the action of the modular automorphism . in section 4
, we introduce the smash products @xmath4 and @xmath5 and recall a result of farinati @xcite which shows that the calabi - yau property of @xmath5 is implied by that of @xmath4 .
the section is then concluded by the proof of the main theorem .
the second author acknowledges support by the epsrc grant `` hopf algebroids and operads '' and the newton institute cambridge which provided a great environment for finishing this paper .
fix a field @xmath15 of characteristic zero .
all algebras under consideration will be unital and associative @xmath15-algebras .
unadorned @xmath16 and @xmath17 shall denote @xmath18 and @xmath19 respectively . if @xmath0 is an algebra , we shall denote by @xmath20 and @xmath21 its opposite and enveloping algebras . the antipodal map @xmath22 satisfying @xmath23 is an involution and induces inverse functors @xmath24 identifying the categories of left and right @xmath25-modules .
there are further equivalences between the categories of left and right @xmath25-modules and the category of @xmath0-bimodules with a symmetric action of @xmath15 . if @xmath26 and @xmath27 are left ( resp .
right ) @xmath25-modules , we shall denote by @xmath28 and @xmath29 ( resp .
@xmath30 and @xmath31 ) the induced left and right actions of @xmath0 .
let @xmath26 be an @xmath0-bimodule .
for each pair of automorphisms @xmath32 , we define the _ twisted bimodule _
@xmath33 to be the @xmath15-space @xmath26 together with the @xmath0-bimodule structure : @xmath34 if either @xmath35 or @xmath3 is the identity , we shall supress it from the notation , writing for example @xmath36 instead of @xmath37 .
the following lemma is standard .
let @xmath38 be automorphisms .
then * the map @xmath39 is an isomorphism of bimodules .
in particular , @xmath40 . * the bimodules @xmath0 and @xmath41 are isomorphic if and only if @xmath3 is an inner automorphism .
see , for example , @xcite . the _ hochschild homology _ of an algebra @xmath0 with coefficients in a right @xmath25-module @xmath26
is given by @xmath42 and is realised as the homology of the _ hochschild ( chain ) complex _
@xmath43 where @xmath44 and the boundary map @xmath45 satisfies @xmath46 the _ hochschild cohomology _ of an algebra @xmath0 with coefficients in a left @xmath25-module @xmath27 is @xmath47 and is realised as the cohomology of the _ hochschild ( cochain ) complex _
@xmath48 where @xmath49 and the coboundary map @xmath50 satisfies @xmath51 the hochschild cohomology @xmath52 , _ a priori _ just a @xmath15-space , may also be considered as a graded commutative algebra whose multiplication @xmath53 called the _ cup product _ , is induced by the associative product @xmath54 on the hochschild complex @xmath55 where @xmath56 is the map satisfying : @xmath57 the _ cap product _ is the map @xmath58 defined on the level of hochschild ( co)chains by @xmath59 the tensor product @xmath60 is formed using the right and left actions @xmath61 and @xmath62 respectively and is considered as a right @xmath25-module using the remaining actions @xmath63 and @xmath64 . taking the cohomology coefficients in the above to be @xmath65 , the cap product endows the hochschild homology @xmath66 with the structure of a ( graded ) right @xmath52-module via @xmath67
the most general framework for poincar type duality in hochschild ( co)homology is provided by a well known theorem of van den bergh which we state below . first , we need the following definition .
an algebra @xmath0 is said to be _ homologically smooth _ if there exists a finite length resolution of the @xmath25-module @xmath0 by finitely generated projective @xmath25-modules .
( @xcite ) let @xmath0 be a homologically smooth algebra .
assume that there exists an integer @xmath68 such that @xmath69 for all @xmath70 and further assume that @xmath71 is an invertible @xmath25-module .
then , for all left @xmath25-modules @xmath26 , there are natural isomorphisms @xmath72 when the conditions of the theorem are satisfied , the algebra @xmath0 is said to have _ van den bergh duality of dimension @xmath10 _ and the ( right ) @xmath25-module @xmath71 is called the _ dualising bimodule _ of @xmath0 . in this case
, we necessarily have that @xmath10 is equal to the dimension @xmath73 of @xmath0 which is , by definition , the projective dimension of @xmath0 as an @xmath25-module ( see @xcite ) . for an @xmath74-dimensional closed orientable manifold @xmath26 , the poincar duality isomorphism @xmath75 in singular cohomology
is given by taking the cap product with a fundamental class @xmath76 \in h_n(m)$ ] . in @xcite and
@xcite it is shown that the dualising isomorphism in the ( co)homology of an algebra with van den bergh duality may be realised analogously .
first , we must define the fundamental class . given any algebra @xmath0 and any integer @xmath68 , abbreviate @xmath77 as in the statement of theorem 2.2 ( however , for the moment we do not assume any of the further conditions to be satisfied ) .
then , the cap product @xmath78 provides a @xmath15-linear map @xmath79 in @xcite , it is proven that if @xmath0 has van den bergh duality of dimension @xmath10 , then @xmath80 is an isomorphism .
one then defines the _ fundamental class _ of @xmath0 to be the unique element @xmath81 such that @xmath82 . using this terminology
, we then have : if @xmath0 is an algebra with van den bergh duality of dimension @xmath10 and @xmath26 is any left @xmath25-module , then @xmath83 is an isomorphism .
see @xcite or @xcite
. taking @xmath84 , the proposition says that @xmath85 is freely generated as a @xmath52-module by the fundamental class @xmath86 . in @xcite
, ginzburg introduced the study of calabi - yau algebras , a noncommutative generalisation of the coordinate rings of calabi - yau varieties .
specifically , a calabi - yau algebra is an algebra @xmath0 with van den bergh duality whose dualising bimodule @xmath87 is isomorphic to @xmath0 as right @xmath25-modules .
more generally , we have : an algebra @xmath0 is said to be _
@xmath3-twisted calabi - yau _ if it has van den bergh duality and the dualising bimodule @xmath87 is isomorphic to @xmath41 for some @xmath1 .
the automorphism @xmath3 is variously called the _ modular _ or _ _ nakayama automorphism _ _ of @xmath0 .
observe that the modular automorphism is only unique up to inner automorphisms . if @xmath3 is the identity ( or more generally an inner automorphism ) , then the algebra @xmath0 is a calabi - yau algebra in the sense of ginzburg .
if @xmath0 is a @xmath3-twisted calabi - yau algebra , any choice of isomorphism @xmath88 identifies the fundamental class @xmath81 with an element of @xmath89 which we shall refer to as _ a _ fundamental class for @xmath0 and by abuse of notation , shall denote by @xmath86 . the twisted calabi - yau condition appears to have first been explicitly defined in @xcite where the authors used the term _
rigid gorenstein_. they showed that a wide class of noetherian hopf algebras , including for example , the quantised function algebras @xmath90 of connected complex semisimple algebraic groups @xmath91 , are what we now refer to as twisted calabi - yau .
other examples of twisted calabi - yau algebras include quantum homogeneous spaces such as podle quantum 2-sphere @xcite , the deformation quantization algebra of a calabi - yau poisson variety @xcite , koszul algebras whose koszul dual is frobenius @xcite , group algebras of poincar duality groups ( e.g. the group algebras of fundamental groups of closed aspherical manifolds ) @xcite and as - regular algebras @xcite .
the cyclic cohomology of a noncommutative algebra was introduced in the early 1980s independently by connes @xcite and tsygan @xcite . shortly afterwards
, the corresponding theory of cyclic homology was formulated by loday and quillen @xcite .
the cyclic homology of an algebra @xmath0 is based on the _ cyclic operator _
@xmath92 on the hochschild complex whose @xmath74th degree component is given by the signed permutation @xmath93 this endows the simplicial module @xmath94 with the additional structure of a _
cyclic module_. associated to this structure is _ connes - tsygan boundary map _
@xmath95 which anticommutes with the hochschild boundary @xmath96 and leads to connes mixed @xmath97-bicomplex whose total homology is the cyclic homology @xmath98 of @xmath0 .
we refer the reader to @xcite for more detail and precise definitions . in @xcite , in order to develop an appropriate analogue of the trace of a @xmath99-algebra for the compact quantum groups of woronowicz
, the authors defined the twisted hochschild and cyclic cohomologies of an algebra @xmath0 relative to an automorphism @xmath1 .
these extend the usual hochschild and cyclic cohomology of @xmath0 , which may then be viewed as the corresponding twisted theories relative to the automorphism @xmath2 .
the dual theories of twisted hochschild and cyclic homology were then observed to be often less degenerate than the untwisted theories and in particular to avoid the so called ` dimension drop ' in hochschild homology that occurs in examples of quantum deformations .
more precisely , for @xmath100 the only automorphism for which the twisted hochschild homology of @xmath101 does not vanish in degree 3 is by the hochschild - kostant - rosenberg theorem the identity .
however , for @xmath102 not a root of unity this happens precisely for the positive powers of the modular automorphism of @xmath101 , see @xcite .
this phenomenon has also been observed in similar examples , see e.g. @xcite .
we now define the notion of a paracyclic module @xcite ( introduced independently under the name ` duplicial module ' in @xcite ) which generalises cyclic modules and underlies twisted cyclic homology .
first , we recall the following : a _ simplicial @xmath15-module _ is a graded @xmath15-module @xmath103 together with maps @xmath104 for all @xmath105 and @xmath106 such that @xmath107 @xmath108 a
_ paracyclic @xmath15-module _ is a simplicial @xmath15-module @xmath109 together with maps @xmath110 for all @xmath111 satisfying @xmath112 a _
cyclic @xmath15-module _ is a paracyclic @xmath15-module @xmath109 such that @xmath113 . in particular , for every paracyclic @xmath15-module @xmath109 , we define the _ associated cyclic module _ @xmath114 .
the motivating example of a paracyclic module is the hochschild complex @xmath12 associated to an algebra @xmath0 together with an automorphism @xmath1 . indeed , one easily checks the paracyclic relations to prove : the hochschild complex @xmath12 together with the maps @xmath115 satisfying @xmath116 @xmath117 @xmath118 is a paracyclic module . for any paracyclic module @xmath109
we define the following maps : * the _ simplicial _ and _ acyclic boundaries _
@xmath119 * the _ norm operator _ @xmath120 * the _ extra degeneracy _ @xmath121 * the _ connes - tsygan boundary map _ @xmath122
unlike as is the case for cyclic modules , for a paracyclic module @xmath123 does in general neither anticommute with the simplicial boundary @xmath96 , nor does it square to zero .
however , we do have the following : for any paracyclic module @xmath109 , the simplicial boundary @xmath96 and the connes - tsygan boundary map @xmath123 satisfy the relations @xmath124 one establishes the formula directly using the subsiduary relations @xmath125 which are easily verified by computation .
the proposition shows that for any paracyclic module @xmath109 , the images of @xmath96 and @xmath123 on the associated cyclic complex @xmath126 anticommute .
the cyclic homology of @xmath109 is then defined to be the total homology of connes mixed @xmath97-bicomplex @xmath127_{\bb_\bullet}&\ar[d]_{\bb_\bullet } & \ar[d]_{\bb_\bullet } & \ar[d]_{\bb_\bullet } \\
c^\mathrm{cyc}_3 \ar[d]_{\bb_\bullet}&\ar[l]_{\bb_\bullet } c^\mathrm{cyc}_2 \ar[d]_{\bb_\bullet } & \ar[l]_{\bb_\bullet}c^\mathrm{cyc}_1 \ar[d]_{\bb_\bullet}&\ar[l]_{\bb_\bullet } c^\mathrm{cyc}_0 \\ c^\mathrm{cyc}_2 \ar[d]_{\bb_\bullet}&\ar[l]_{\bb_\bullet } c^\mathrm{cyc}_1 \ar[d]_{\bb_\bullet } & \ar[l]_{\bb_\bullet } c^\mathrm{cyc}_0 & \\
c^\mathrm{cyc}_1 \ar[d]_{\bb_\bullet } & \ar[l]_{\bb_\bullet } c^\mathrm{cyc}_0 & & \\ c^\mathrm{cyc}_0 & & & } \ ] ] and the hochschild homology is defined to be the homologies of the columns . in case @xmath128
, this is denoted by @xmath129 . in general
, the hochschild homology differs from the simplicial homology of @xmath109 ( which is for @xmath128 the hochschild homology @xmath13 ) but they are isomorphic if @xmath109 is _ quasicyclic _ ( see definition 3.6 ) .
one can also interpret proposition 3.4 as the fact that @xmath123 is a chain homotopy @xmath130 for the associated complex computing the simplicial homology underlying the paracyclic structure .
returning to the example of the hochschild complex @xmath12 , the above proposition implies the following invariance property of the fundamental class of a twisted calabi - yau algebra which will be key to the proof of theorem [ maintheorem ] : [ fixed ] if @xmath0 is a @xmath3-twisted calabi - yau algebra of dimension @xmath10 , then the fundamental class @xmath131 is invariant under the action of @xmath132 , that is , if @xmath133 is any cycle representing @xmath86 , then we have @xmath134=\tt_d([z])=\tt_d(\omega_a)=\omega_a.\ ] ] in general , it might not be possible to choose a representative of the fundamental class that is invariant under @xmath135 on the chain level
. however , if @xmath136 is a _ quasi - cyclic module _ in accordance with the following definition , then it is possible to choose an invariant representative for @xmath86 .
a paracyclic @xmath15-module @xmath109 is said to be _ quasi - cyclic _
if @xmath137 indeed , for a quasicyclic module @xmath109 , it follows that the projection @xmath138 of simplicial complexes is a quasi - isomorphism , so the homologies of the two complexes are isomorphic to each other and by construction also to the homology of the @xmath135-invariants ( see @xcite*proposition 2.1 ) .
the motivating example is the paracyclic module @xmath12 associated to any algebra @xmath0 and automorphism @xmath3 that acts semisimply on @xmath0 . indeed ,
if @xmath3 is semisimple then so is @xmath135 and the required splitting comes from the eigenspace decomposition of @xmath12 .
let @xmath0 be an algebra and let @xmath1 .
then @xmath3 generates an action of the group algebra @xmath139 $ ] given by @xmath140 where we identify @xmath139 $ ] with the laurent polynomials @xmath141 $ ] . the _ smash product _ or _ skew group algebra _
@xmath5 is the @xmath15-vector space @xmath142 $ ] with multiplication given by the rule @xmath143 the smash product or _
skew semigroup _
algebra @xmath4 is the subalgebra of @xmath5 consisting of all sums of pure tensors of the form @xmath144 where @xmath145 . for all smash products @xmath4
, there is an isomorphism @xmath146 , \quad a \otimes x^i \mapsto ax^i\ ] ] where the _ skew polynomial algebra _ @xmath147 $ ] is formed by adjoining the variable @xmath148 to @xmath0 subject to the commutation relation @xmath149 . similarly ,
the smash products @xmath5 are isomorphic to the _ skew laurent algebras _
@xmath150 $ ] . whenever it causes no confusion , we shall tacitly make use of these identifications writing , for example @xmath148 to denote @xmath151 .
the smash products @xmath5 are examples of the general notion of the smash product @xmath152 of a hopf algebra @xmath153 with an @xmath153-module algebra @xmath0 .
we refer to @xcite*chapter 4 . for a full definition .
this construction still makes sense if @xmath153 is only a bialgebra which justifies our use of terminology in calling @xmath4 a smash product .
the following is a short excursus , recalling a general notion of noncommutative localisation proposed by farinati in @xcite , of which the algebra extension @xmath154 is an example .
it is also proven in @xcite that van den bergh duality is stable under such localisation and furthermore the dualising module of a localisation is explicitly described . in particular
, it follows that the calabi - yau property of @xmath5 is implied by that of @xmath155 .
let @xmath0 be an algebra with van den bergh duality of dimension @xmath10 and dualising bimodule @xmath87 and let @xmath156 be a localisation such that @xmath162 as @xmath25-modules .
then , @xmath159 has van den bergh duality of dimension @xmath10 with dualising bimodule @xmath163 .
[ wigan ] let @xmath0 be a @xmath3-twisted calabi - yau algebra of dimension @xmath10 .
then the smash product @xmath4 is homologically smooth and @xmath165 , & \bullet = d+1 \\ 0 , & \bullet \neq d+1\end{cases}\ ] ] as @xmath166-modules where the actions of @xmath4 on @xmath167 $ ] are given by @xmath168 \otimes x^k ) & = \sigma^{-k}(a)\cdot [ f ] \otimes x^k \\ x \rla ( [ f ] \otimes x^k ) & = [ f ] \otimes x^{k+1 } \\ ( [ f ] \otimes x^k ) \rra a & = [ f ] \cdot a \otimes x^k \\ ( [ f ] \otimes x^k ) \rra x & = [ ( \sigma^{-1})^{\otimes 2 } \circ f \circ \sigma^{\otimes d } ] \otimes x^{k+1}.\end{aligned}\ ] ] here the actions of @xmath0 on @xmath169 are induced by those of the right @xmath25-module structure on the coefficients @xmath170 .
it follows from this theorem that @xmath4 is twisted calabi - yau and that the modular automorphism extends the identity automorphism of @xmath0 .
such automorphisms of @xmath4 are parameterised by central units @xmath171 of @xmath0 where the automorphism associated to @xmath171 satisfies @xmath172 ( cf .
@xcite*proposition . 22 . ) .
it therefore remains to show that the modular automorphism is the one corresponding to @xmath173 .
fix an isomorphism @xmath88 .
recall that the choice of such an isomorphism fixes a fundamental class @xmath174 . the cap product with @xmath86
thus yields a @xmath15-linear isomorphism @xmath175 \ar[rr]^-{(\omega_a \cap - ) \otimes \operatorname{id } } & & h_0(a,{{_{\sigma^{-1}}}{a } _ { } } \otimes { { _ { } } { a}_{\sigma^{-1 } } } ) \otimes \k[x ] \ar[rr]^-{([a \otimes b]\mapsto ba)\otimes \operatorname{id } } & & a \otimes \k[x].}\ ] ] the @xmath176-module structure on @xmath167 $ ] from theorem [ wigan ] then induces a right action of @xmath177 on @xmath178 $ ] by transport of structure : it is clear that for all @xmath179 $ ] and @xmath180 , this action is given by @xmath181 together with the right action of @xmath148 on @xmath178 $ ] that we shall now determine . pick a cycle @xmath182 such that @xmath183 $ ] . then , we have the commutative diagram of ( co)chain complexes : @xmath184 \ar[d]_-{(\sigma^{-1})^{\otimes 2}\circ - \circ \sigma^{\otimes d}\otimes \operatorname{id}}\ar[rrr]^-{(z \cap - ) \otimes \operatorname{id } } & & & c_{d-\bullet}(a,{{_{\sigma^{-1}}}{a } _ { } } \otimes a_{\sigma^{-1 } } ) \otimes \k[x ] \ar[d]^-{(\sigma^{-1})^{\otimes d-\bullet+2}\otimes\operatorname{id } } & \\
c^\bullet(a , a\otimes a_{\sigma^{-1}})\otimes \k[x ] \ar[rrr]^-{(\tt_d(z ) \cap - ) \otimes \operatorname{id } } & & & c_{d-\bullet}(a,{{_{\sigma^{-1}}}{a } _ { } } \otimes a_{\sigma^{-1}})\otimes \k[x]}\ ] ] where @xmath185 $ ] is viewed as a ( co)chain complex concentrated in degree @xmath186 and @xmath187 comes from the paracyclic structure on @xmath136 described in proposition 4.3 .
now corollary [ fixed ] gives @xmath134=\omega_a,\ ] ] so the induced diagram on ( co)homology takes in degree @xmath10 the form @xmath188 \ar[d]_-{\rra x } \ar[rr]^-{(\omega_a \cap
-)\otimes \operatorname{id } } & & h_0(a,{{_{\sigma^{-1}}}{a } _ { } } \otimes a_{\sigma^{-1}})\otimes \k[x]\ar[rr]^-{([a \otimes b]\mapsto ba)\otimes \operatorname{id } } & & a \otimes \k[x ] \ar[d]^-{\sigma^{-1}\otimes \operatorname{id}}\\ \ar[rr]^-{(\omega_a \cap -)\otimes \operatorname{id } } h^d(a , a\otimes a_{\sigma^{-1 } } ) \otimes \k[x ] & & h_0(a,{{_{\sigma^{-1}}}{a } _ { } } \otimes a_{\sigma^{-1 } } ) \otimes \k[x ] \ar[rr]^-{([a \otimes b]\mapsto ba)\otimes \operatorname{id } } & & a \otimes \k[x]}.\ ] ] consequently , the right action of @xmath148 on @xmath178 $ ] is given by @xmath189 finally , the bijection @xmath190 { \stackrel{}{\longrightarrow } } a \rtimes_\sigma \n , \qquad a\otimes x^k \mapsto \sigma^{k}(a ) \otimes x^k\ ] ] yields an isomorphism @xmath191 as right @xmath166-modules thus completing the proof . in @xcite , using a spectral sequence due to stefan @xcite , it is shown that if @xmath153 is a calabi - yau hopf algebra and @xmath0 is an @xmath153-module algebra with van den bergh duality , then the smash product @xmath152 also has van den bergh duality although the argument used there does not completely determine the dualising bimodule @xmath192 .
considering the special case where @xmath193 $ ] and @xmath0 is a @xmath3-twisted calabi - yau algebra whose @xmath153-module structure is given by the modular automorphism @xmath3 , a straightforward adaptation of the main argument in the proof above shows that the terms @xmath194 of the spectral sequence are invariant under @xmath3 , which can be used to fully determine the dualising module and show that it is indeed isomorphic to @xmath5 .
the same result does not directly apply to smash products with the calabi - yau bialgebra @xmath195 $ ] since in particular it does not satisfy the hypotheses of stefan s spectral sequence .
however , one can still use the arguments of @xcite to show that a similar spectral sequence for smash products @xmath4 exists and then the proof in @xcite proceeds in the same way .
however , referring to theorem [ wigan ] allowed us to formulate the proof in a shorter and more direct way . the results in @xcite
show more generally that any ore extension @xmath196 $ ] of a twisted calabi - yau algebra @xmath0 is also twisted calabi - yau .
the authors prove this result by means of a bicomplex computing @xmath196 $ ] , based on one from @xcite .
similarly as with @xcite , the proof does not explicitly determine the modular automorphism .
in the case where @xmath197 so that the ore extension @xmath196 $ ] is the skew polynomial ring @xmath147\iso a \rtimes_\sigma \n$ ] , this bicomplex is an explicit construction of the spectral sequence for @xmath4 along the lines of stefan s one as mentioned above .
specialising further to the case where @xmath3 is the modular automorphism of @xmath0 , one obtains theorem [ wigan ] as stated here .
one interesting feature of the actions of @xmath4 on @xmath178 $ ] is that they would appear to be more naturally stated as a left rather than right @xmath166-module structure .
although this distinction may seem unimportant , it becomes more pronounced when one views hochschild ( co)homology as an instance of the ( co)homology of hopf algebroids as in @xcite .
the duality isomorphism takes right cohomology modules and produces left modules in homology .
we refer the interested reader to @xcite and the references contained therein for further details . | twisted calabi - yau algebras are a generalisation of ginzburg s notion of calabi - yau algebras .
such algebras @xmath0 come equipped with a _ modular automorphism _
@xmath1 , the case @xmath2 being precisely the original class of calabi - yau algebras . here
we prove that every twisted calabi - yau algebra may be extended to a calabi - yau algebra .
more precisely , we show that if @xmath0 is a twisted calabi - yau algebra with modular automorphism @xmath3 , then the smash product algebras @xmath4 and @xmath5 are calabi - yau . |
an important discovery at relativistic heavy ion collider ( rhic ) in recent years is the realization of strongly interacting quark gluon plasma ( sqgp ) @xcite . for the collective effects from rhic experiments , known as radial and elliptic flow
, the qgp could be well described by ideal hydrodynamics .
it implies that the qgp at rhic is the most perfect fluid @xcite .
for this reason , shuryak has pointed out there should be lots of bound states @xcite especially for light and heavy qq bound states at @xmath0 , where @xmath1 is the transition temperature . for the physical quark mass
the qcd transition is a non - singular crossover at finite temperature . in recent lattice results , at high temperatures and small densities ,
it is a crossover from hadronic phase to qgp phase @xcite . from recent theoretical studies
, we have realized the rich phases of qcd theory , such as color superconducting phase @xcite , pion condensation @xcite , color glass condensate and quarkyonic phase @xcite .
now the rhic results tell us that after deconfinement , the quarks are not free immediately , as we usually thought for a weakly coupled qgp ( wqgp ) , but still in strong coupled state which leads to a new state of nuclear matter as the sqgp . how to understand the formation mechanism of sqgp is an important problem @xcite . as we know
, there are confine potential and color coulomb potential between quarks in vacuum , know as @xmath2 . with temperature @xmath3
increased to some critical temperature , the confine potential disappears , while the coulomb potential remains , which could still be very strong @xcite .
thus in literatures , the effective coulomb potential has been often used to describe the strong interaction of sqgp @xcite .
but from the theoretical point of view , it is lack of a more fundamental theoretical model to understand these bound states of quarks . in this paper
we wish to use non - topological fl model to study this problem .
fl model has been widely discussed in past decays @xcite .
it has been very successful in describing phenomenologically the static properties of hadrons and their behaviors at low energy .
the model consists of quark fields interacting with a phenomenological scalar field @xmath4 .
the @xmath4 field is introduced to describe the complicated nonperturbative features of qcd vacuum .
it naturally gives a color confinement mechanism in qcd theory .
the model has been also extended to finite temperatures and densities to study deconfinement phase transition @xcite .
however it seems that the deep meaning of the soliton solutions in deconfinement phase transition has not been revealed in the past studies .
the main purpose of this paper is to study in detail the properties of the solitons in fl model before and after deconfinement , and provide a natural explanation and description of sqgp through this effective theoretical model .
the organization of this paper is as follows : in section 2 we give a brief introduction of fl model .
the field equations and the effective potential are derived . in section 3 , the soliton equation of fl model
is solved for different boundary conditions and the physical meanings of these solitons in deconfinement are thoroughly discussed . in section 4 , a phase diagram of deconfinement phase transition is given .
the last section is the summary .
we start from the lagrangian of the fl model , = |(i_^-g)+12(_)(^)-u ( ) , whereu()=12!a^2 + 13!b^3 + 14!c^4+b .
@xmath5 represents the quark field , and @xmath4 denotes the phenomenological scalar field . @xmath6 and @xmath7 are the constants which are generally fitted in with producing the properties of hadrons appropriately . for the thermal equilibrium system , the @xmath4 field will be replaced by @xmath8 , where @xmath9 is the gibbs thermal average of the @xmath4 field , and @xmath10 is the fluctuation . at mean field approximation , one can obtain the field equations as follows , & & ( i_^-g|)=0 , + & & _ ^|=- , [ mean ] where @xmath11 is one loop effective potential at finite temperature and density .
notice that at mean field approximation the quantum corrections have been neglected .
the @xmath11 could be derived through finite temperature field theory @xcite , v_eff = u(|)+1(1-e^-e_)- , [ eff ] in which @xmath12 is the inverse temperature , @xmath13 is the chemical potential , and @xmath14 is a degenerate factor , @xmath15 . besides @xmath16 , @xmath17 , where @xmath18 and @xmath19 are the effective masses of the quark and @xmath4 field respectively . from equation ( [ mean ] )
, one could see that the properties of the soliton field @xmath9 depend completely on the effective potential @xmath11 .
as known , by the effective potential at finite temperature and density , one can study the deconfinement phase transition in fl model .
thus the properties of solitons in deconfinement , especially the relations between solitons and deconfinement , could be well studied through solving the equation ( [ mean ] ) at finite temperature and density . in our calculation , the parameters are chosen to be @xmath20 .
the effective mass of @xmath4 field is fixed at @xmath21 @xcite . at zero temperature and density
, the @xmath11 is just the @xmath22 and could be plotted in fig.[f1 ] .
there are two minima : one corresponds to the perturbative vacuum at @xmath23 , another corresponds to the physical vacuum at @xmath24 .
for the static and spherically symmetric soliton field , the equation ( [ mean ] ) becomes + 2r=. [ soliton ] we will solve this equation numerically . a standard numerical package colsys will be used for the numerical calculation @xcite . for solving the soliton equation ( [ soliton ] ) , one should first determine the boundary condition .
there are two different cases of the boundary conditions for solving the equation @xcite . in the following
the two cases will be studied separately . in this case
, we may consider the mechanical analog in which there is a point particle at positon " @xmath4 and a time
r " , moving in a potential " @xmath25 .
the discussions about this mechanical analog could be found in ref .
equation ( [ soliton ] ) is then the movement equation of the particle . because the energy conservation will be applied in this case , the second term at l.h.s of equation ( [ soliton ] ) will be omitted .
thus the equation becomes = .
[ soliton1 ] in order to determine the boundary condition , suppose at time " @xmath26 , the position " of the particle is at @xmath23 . from fig.[f1 ] , if the particle is pushed very gently to start moving , the particle will move to another point just at @xmath27 where @xmath28 , noticing that the particle is moving in the potential " @xmath25 .
because of energy conservation , the particle will then move back and return to the point @xmath23 .
thus one could obtain the following boundary condition , |(r=-)=0 , |(r=+)=0 . from fig.[f1 ] , the physical vacuum @xmath24 is stable , and @xmath29 is the bag constant .
the quarks are confined in a soliton bag . at zero temperature and density
, the soliton solution represents a hadron . , @xmath30 , @xmath31 , @xmath32 , @xmath33 and @xmath34 .
] at finite temperature , the @xmath11 could be plotted in fig.[f2 ] .
there are two critical temperatures @xmath35 and @xmath36 . at @xmath35 ,
the two minima degenerate and the bag constant @xmath37 . at @xmath36 ,
the minimum @xmath24 just vanishes . for different configurations of @xmath11 , the equation ( [ soliton1 ] )
could be numerical solved by the numerical package colsys .
however the @xmath11 is highly non - linear in @xmath9 as shown in equation ( [ eff ] ) .
it could not be directly used in colsys evaluation . at finite temperature , as the @xmath11 could be plotted as shown in fig.[f2 ] , we can fit the curve in terms of @xmath38 , @xmath39 and @xmath40 with the effective parameters @xmath41 , @xmath42 and @xmath43 .
v_eff=12!a(t)^2 + 13!b(t)^3 + 14!c(t)^4+b(t ) . in this form
, the @xmath11 could be applied in colsys for the numerical calculation . in the following
we will solve equation ( [ soliton1 ] ) and discuss the soliton solutions for different configurations of @xmath11 .
, @xmath44 , @xmath30 and @xmath31 .
( b ) @xmath32 and @xmath33.,title="fig : " ] , @xmath44 , @xmath30 and @xmath31 .
( b ) @xmath32 and @xmath33.,title="fig : " ] at @xmath45 , the physical vacuum at @xmath24 is stable .
the bag constant @xmath46 and its value decreases with temperature increasing .
the corresponding quarks are confined .
the equation ( [ soliton1 ] ) could be numerically solved and the soliton solutions are plotted in fig.[f3]a .
one could see that the solitons at temperature @xmath45 do not change so much when compared to that at zero temperature except the peaks of the solitons become higher when temperature increases .
the quarks keep confined in a hadronic state until the critical temperature @xmath47 the two vacuums degenerate . at this time , still required by the energy conservation the boundary condition is |(r=-)=0 , |(r=+)=_v .
the soliton becomes topological as shown in fig.[f3]a .
as the bag constant @xmath37 at this time , the hadrons are destructed and the quarks are deconfined .
if we keep increasing temperature to @xmath48 , from fig.[f2 ] , the perturbative vacuum @xmath23 is stable .
the equation ( [ soliton1 ] ) will be solved under the boundary condition fig.[f3]b .
the soliton solutions still exist but become very different from those at @xmath45 .
the peaks of the solitons become downward .
when temperature increasing , they become flat . these solitons do not represent hadrons any more but the bound states of quarks .
though the quarks are deconfined , they are still in strong coupled states .
that is to say the system is in a sqgp phase . until @xmath49 ,
the physical vacuum @xmath24 vanishes and only the perturbative vacuum @xmath23 exists , has the soliton solution disappeared . at this time
the quarks become free .
the system goes into a quasi - free gas phase of qgp . , @xmath50 , @xmath51 , @xmath52 , @xmath53 and @xmath54 .
] at finite temperature and density , we could also make the similar discussions . in fig.[f4 ] , we show the @xmath11 at @xmath55 and different chemical potentials .
there are also two critical chemicals @xmath56 and @xmath57 . at @xmath56
the two vacuums degenerate ; at @xmath57 the vacuum of @xmath58 just vanishes .
the corresponding soliton solutions are plotted in fig.[f5]a and fig.[f5]b .
it is clear that at @xmath59 , the system is in a hadronic phase ; at @xmath60 , the system is in a sqgp phase ; at @xmath61 , the system becomes quasi - free gas of qgp . , @xmath50 and @xmath51 .
( b ) @xmath52 and @xmath53.,title="fig : " ] , @xmath50 and @xmath51 .
( b ) @xmath52 and @xmath53.,title="fig : " ] , @xmath44 , @xmath30 and @xmath31 .
( b ) @xmath32 and @xmath33.,title="fig : " ] , @xmath44 , @xmath30 and @xmath31 .
( b ) @xmath32 and @xmath33.,title="fig : " ] in this case , at @xmath62 , the boundary condition is taken as .|_r=0=0 , .||_r=0 . the equation ( [ soliton ] )
could be numerically solved at finite temperature and density for different configurations of @xmath11 . from fig.[f2 ] , at @xmath45 , the soliton solutions are plotted in fig.[f6]a .
the physical vacuum @xmath24 is stable and the bag constant @xmath46 .
the solitons still represent that the quarks are confined and the system is in a hadronic phase . until @xmath47
the system are deconfined .
the corresponding soliton is also plotted in fig.[f6]a . at @xmath48 ,
the boundary condition is .|_r=0=0 ,
.||_r=_v . the soliton solutions are solved and plotted in fig.[f6]b .
these solitons are quite different from those at @xmath45 .
as the system is already deconfined , these solitons represent the quarks are in a bound state .
the system is in a sqgp phase . at @xmath63 , there are no soliton solution any more , the system goes to a quasi - free gas of qgp . for different chemical potentials at fixed temperature @xmath55 , according to the @xmath11 as shown in fig.[f4 ] , the corresponding soliton solutions could be solved and plotted in fig.[f7]a and fig.[f7]b .
it is clear that at @xmath64 it is a hadronic phase ; at @xmath60 , it is in a sqgp phase ; at @xmath65 it is a quasi - free gas of qgp . ,
@xmath50 and @xmath51 .
( b ) @xmath52 and @xmath53.,title="fig : " ] , @xmath50 and @xmath51 .
( b ) @xmath52 and @xmath53.,title="fig : " ] from above discussions , it is clear that for the two different boundary conditions , the physical meaning of the solitons are the same : before deconfinement the solitons represent hadrons while after deconfinment the solitons represent bound states of quarks .
for the two cases , the phase structure of deconfinement remains unchanged and unique .
now we are in a position to plot the phase diagram of deconfinement phase transition in fl model . when the chemical potential is fixed , one could obtain two critical temperatures . increasing the chemical potential to another fixed value
, the other two corresponding critical temperatures could be obtained , and so on .
the @xmath66 phase diagram could be plotted as shown in fig.[f8 ] .
the full line is the dividing line between hadronic phase and deconfined quark phase .
the dashed line further divides the deconfined quark phase into the sqgp phase and the wqgp phase .
in this paper the soliton solutions in fl model at finite temperatures and densities have been thoroughly discussed for two kinds of different boundary conditions .
the physical interpretations of the solitons in deconfinement phase transition have been given .
as in fl model the system is deconfined at the time that the bag constant becomes zero , we indicate that the solitons before deconfinement are hardrons , while the solitons after deconfinement represent the strong coupled bound states of quarks , which leads to the sqgp phase .
the qcd phase structure still needs lots of thorough investigations , especially for the finite density areas , where the results are often model dependent .
here we only present a possible phase diagram based on fl model .
the phase structure of deconfinement phase transition here is qualitatively consistent with that of shuryak @xcite .
the whole phase diagram is divided into three phases : the hadronic phase , the sqgp phase and the wqgp phase .
shuryak and i. zahed , phys.rev .
c49 ( 2004 ) 021901 .
brown , c - h .
lee , m. rho and e.v .
shuryak , nucl.phys .
a740 ( 2004 ) 171 . m. gyulassy and l. mclerran , nucl .
a750 , ( 2005 ) 30 .
d. teaney , j. lauret and e.v .
shuryak , phys .
86 ( 2001 ) 4783 .
kolb , p. huovinen , u. heinz and h. heiselberg , phys .
b500 ( 2001 ) 232 .
d. teaney , phys.rev .
c68 ( 2003 ) 034913 .
z. fodor and s.d .
katz , j. high energy phys .
050 ( 2004 ) 0404 .
f. karsch , poscpod07 ( 2007 ) 026 ; poslat2007 ( 2007 ) 015 .
m. alford , k. rajagopal and f. wilczek , phys .
lett . b422
( 1998 ) 247 ; k. rajagopal , prog .
131 ( 1998 ) 619 ; r. rapp , t. schafer , e.v .
shuryak and m. velkovsky , phys .
81 ( 1998 ) 53 .
d.t . son and m.a .
stephanov , phys .
86 ( 2001 ) 592 ; k. splittorff , d.t . son , and m.a .
stephanov , phys .
d64 ( 2001 ) 016003 ; j.b .
kogut and d. toublan , phys .
d64 ( 2001 ) 034007 .
mclerran and r. venugopalan , phys .
d49 ( 1994 ) 2233 ; e. iancu and l. mclerran , phys .
b510 ( 2001 ) 145 ; l. mclerran and r. pisarski , nucl .
a796 ( 2007 ) 83 .
lee , nucl .
a750 ( 2005 ) 1 . k. yagi , t. hatsuda and y. miake , quark - gluon plasma , cambridge univ . press ( 2005 ) .
d. zwanziger , phys .
( 2003 ) 102001 ; phys . rev .
d 70 ( 2004 ) 094034 .
r. friedberg and t.d .
lee , phys .
d15 , ( 1977 ) 1694 ; d16 , ( 1977 ) 1096 ; d18 , ( 1978 ) 2623 .
r. goldflam and l. wilets , phys .
d25 ( 1982 ) 1951 .
birse , prog .
phys . 25 ( 1990 ) 1 .
h. reinhardt , b.v .
dang and h. schulz , phys .
b159 ( 1985 ) 161 .
m. li , m.c .
birse and l. wilets , j.phys .
g13 ( 1987 ) 1 .
wang , j.r .
li and l.s .
liu , phys .
d41 ( 1990 ) 2288 ; s. gao , e.k .
wang and j.r .
li , phys .
d46 ( 1992 ) 3211 ; s.h . deng and j.r .
li , phys.lett .
b302 ( 1993 ) 279 . h. mao , r.k . su and w.q .
zhao , phys .
c74 ( 2006 ) 055204 ; h. mao , m.j . yao and w.q .
zhao , phys .
c77 ( 2008 ) 065205 .
u. ascher , j. christiansen and r.d .
russell , acm trans .
software 7 ( 1981 ) 209 .
lee , particle physics and introduction to field theory , harwood academic , new york ( 1981 ) . | the non - topological fl model is studied at finite temperature and density .
the soliton solutions of the fl model in deconfinement phase transition are solved and thoroughly discussed for different boundary conditions .
we indicate that the solitons before and after the deconfinement have different physical meanings : the soliton before deconfinement represents hadron , while the soliton after the deconfinement represents the bound state of quarks which leads to a sqgp phase . the corresponding phase diagram is given .
# 1#2
|
the most complete kinematical information obtainable for a distant stellar system is the distribution of line - of - sight velocities at every point in the image .
velocity distributions are crucial for understanding the dynamical states of slowly - rotating stellar systems like elliptical galaxies and globular clusters , since velocity dispersions alone place almost no constraints on the form of the potential unless one is willing to make ad hoc assumptions about the shape of the velocity ellipsoid ( @xcite ) .
velocity distributions are also useful when searching for kinematically distinct subcomponents ( e.g. @xcite ; @xcite ) .
the velocity distribution at point @xmath2 in the image of a stellar system , @xmath3 , can be related to the data in different ways depending on the nature of the observations . in a system like a globular cluster , for which the data usually consist of individual stellar velocities ,
the velocity distribution is just the frequency function of stellar velocities defined by those stars with apparent positions near to @xmath2 .
since measured velocities are always in error , the observed and true @xmath0 s are related via a convolution integral . in a distant , unresolved galaxy ,
one typically measures the integrated spectrum of many stars along a line of sight .
the observed spectrum is then a convolution of the velocity distribution of these stars with the broadening function of the spectrograph , and the spectrum of a typical star . with both sorts of data ,
the goal is to find a function @xmath0 , at some set of points @xmath2 , such that @xmath4 the log likelihood of observing the data @xmath5 given @xmath6 is large . maximizing this quantity over the space of all possible functions
@xmath0 is unlikely to yield useful results , however , since any @xmath0 that maximizes the likelihood ( assuming it exists , which it often will not ) is almost certain to be extremely noisy .
this is obviously true if the data are related to the model via a convolution , since the process of deconvolution will amplify the errors in the data .
but it is equally true if @xmath0 is simply the frequency function of observed velocities , since the most likely distribution corresponding to an observed set of @xmath7 s is just a sum of delta functions at each of the measured velocities .
one is therefore forced to place smoothness constraints on the solution . but
smoothing always introduces a bias , i.e. a systematic deviation of the solution from the true @xmath0 .
the nature of the bias is obvious when the smoothing is carried out by imposing a rigid functional form on @xmath0 , since the true function will almost certainly be different from this assumed form .
but even nonparametric smoothing generates a bias since it effectively averages the data over some region .
furthermore , because the required degree of smoothing increases with the amplitude of the noise in the data , the error from the bias goes up as the quality of the data falls .
an ideal algorithm for estimating @xmath0 would therefore be one in which the bias introduced by the smoothing was effectively minimized , so that the derived @xmath0 was close to the true function even when the data were so poor that a great deal of smoothing was required .
one way to accomplish this is to make use of prior knowledge about the likely form of @xmath0 .
many studies of stellar and galactic systems have shown that @xmath0 is often close to a gaussian .
this fact suggests that we infer @xmath0 by maximizing a quantity like @xmath8 the `` penalized log likelihood , '' where the penalty functional @xmath9 is large for any @xmath0 that is noisy and zero for any @xmath0 that is gaussian .
a natural choice for such a penalty functional has been suggested by @xcite : @xmath10 ^ 2 dv.\ ] ] this functional assigns zero penalty to any @xmath11 $ ] , i.e. any gaussian velocity distribution , and a large penalty to any @xmath0 that is rapidly varying .
the limiting estimate as the smoothing parameter @xmath12 tends to infinity is the normal distribution that best corresponds , in a maximum - likelihood sense , to the data .
thus varying @xmath12 takes one from an estimate of @xmath0 that is very noisy but that reproduces the data well , to the `` infinitely smooth '' maximum likelihood gaussian fit to @xmath0 .
when the data are copious and accurate , the degree of smoothing required , i.e. the value of @xmath12 , will be small and the inferred @xmath0 will be close to the true function .
when the data are poor , @xmath12 must be increased to deal with the noise ; however even a very large value of @xmath12 will yield an @xmath0 that is nearly gaussian and that is therefore likely to be close to the true velocity distribution .
regardless of the quality of the data , there will always be a formally optimal choice of @xmath12 that yields a solution that is closest in some sense to the true @xmath0 neither too noisy nor too biased .
here we apply penalized likelihood methods to the recovery of velocity distributions from kinematical data of two sorts : doppler - broadened spectra , and discrete velocities .
there are of course a large number of excellent algorithms already in use for extracting velocity distributions from absorption - line spectra ( @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) .
is another approach really needed ?
many of the existing algorithms are essentially nonparametric and hence yield unbiased estimates of @xmath0 when the signal - to - noise ratio of the data is large .
their behavior given data with small s / n is more variable , however , due to the different ways in which they carry out the smoothing .
for instance , in algorithms that represent the unknown @xmath0 via a basis - set expansion , @xmath13 the smoothing is accomplished by truncating the expansion after a finite number of terms .
adding more terms decreases the bias by giving the algorithm more freedom to match the true @xmath0 ; however the level of fluctuations in the solution increases as the bias falls since the high - order terms will always try to reproduce the noise in the data .
the optimal estimate for data of a given quality will therefore contain only a limited number of terms , and if the data are poor , this number will be small ; hence the solution will be biased toward the functional form selected for @xmath14 .
of course one can choose @xmath15 to be the normal distribution ( e.g. @xcite ) but its mean and variance must be specified , and the optimal choices for both parameters will depend in some complex way on the number of terms retained in the expansion and on the level of noise in the data .
worse , one does not have complete freedom in this approach to adjust the smoothing to its optimal level since the basis set is fixed and its terms are discrete .
most of the spectral deconvolution schemes currently in use have continuously - adjustable smoothing and so do not suffer from this latter defect .
however few of these schemes incorporate any prior knowledge about the likely form of @xmath0 and so the solutions they return given noisy data tend to be unphysical , with functional forms that are determined primarily by the mechanics of the smoothing .
( in fact it is a common practice to fall back on parametric methods when the data are poor . ) for instance , both the wiener - filtered algorithm of rix & white ( 1992 ) and the kernel - based algorithm of kuijken & merrifield ( 1993 ) yield @xmath16 in the limit of infinite smoothing . since real velocity distributions
are unlikely to be well approximated by constant functions , the estimates produced by these schemes can be substantially biased .
the approach advocated here is neither better nor worse than the existing ones when the data are of high quality .
however it has an advantage when the data are poor , since even a large degree of smoothing is likely to bias the solution only slightly .
furthermore the amount of smoothing can be adjusted with infinite precision by varying @xmath12 . in practice
it may be difficult to find the optimal choice of @xmath12 given noisy data , but there are bootstrap techniques for choosing @xmath12 that work well in most cases ; and one can always elect to be guided by physical intuition when deciding how smooth @xmath0 should be .
most important , even a gross overestimate of @xmath12 will yield no worse than a gaussian fit to the velocity distribution , which is perhaps the best that can be hoped for when the signal - to - noise ratio or the number of observed velocities is small .
( much of the uncertainty in @xmath0 s derived from absorption line spectra is due to systematic errors such as incorrect continuum subtraction , template mismatch , limited resolution , etc .
these problems afflict every spectral deconvolution scheme and have been treated at length by other authors .
we have nothing new to say about these systematic effects and will simply ignore them our focus is on the uncertainty in @xmath0 that is generated by noise in the data rather than by systematic errors . )
below we evaluate the performance of penalized - likelihood algorithms for estimating @xmath0 from spectral data ( 2 ) and from discrete velocities ( 3 ) .
the former sort of data are routinely obtained for distant galaxies and the latter for globular clusters , clusters of galaxies and systems of emission - line objects around galaxies .
the same formalism works equally well in both cases ; essentially all that needs to be changed is the form of the matrix that relates the observable quantities to @xmath0 .
we also show how the cross - validation score can be used to estimate the optimal degree of smoothing from the data .
we then ( 4 ) apply the formalism to the recovery of the velocity distribution near the center of the globular cluster @xmath1 centauri using a new sample of stellar radial velocities from an imaging spectrophotometer .
the optical spectrum at any point in the image of a galaxy is made up of the sum of the spectra of its component stars along the line of sight , doppler shifted according to their line - of - sight velocities @xmath7 . if the galaxy spectrum is @xmath17 and the spectrum of a single star
after convolution with the instrumental response is @xmath18 , then @xmath19 the convolution of @xmath0 with the template .
we accordingly seek a function @xmath20 such that @xmath21 ^ 2 + \alpha p_g(n ) \label{penlik}\ ] ] is minimized , where @xmath22 is the observed spectral intensity at wavelength @xmath23 .
we first define a discrete grid in velocity space , @xmath24 , where @xmath25 and @xmath26 .
typically @xmath27 and @xmath28 , where @xmath29 is the mean velocity of the system and @xmath30 is the expected velocity dispersion .
the number of grid points @xmath31 should be large enough that the numerical solution to the minimization problem is essentially independent of @xmath31 , yet small enough that the minimization which requires a computational time that scales as @xmath32 does not take inordinately long .
we mostly used @xmath33 velocity grid points in what follows . a discrete representation of the right hand side of eq .
( [ penlik ] ) on this grid is @xmath34 ^ 2 + \alpha\delta^{-5 } \sum_{j=2}^{m-2}\left [ -\log n_{j-1}+3\log
n_j-3\log n_{j+1}+\log n_{j+2}\right]^2 .
\label{sildisc}\ ] ] in the first term , @xmath35 is computed by fitting an interpolating spline to the template spectrum and assuming a linear dependence of the ( unknown ) function @xmath6 on @xmath7 between the grid points .
the convolution can then be represented as a matrix multiplication , i.e. @xmath36 @xmath37 @xmath38 @xmath39 one then varies the @xmath31 parameters @xmath40 until a minimum is obtained ; the use of @xmath41 in the penalty functional guarantees that the solution will remain everywhere positive without the need to impose additional positivity constraints .
minimization was carried out using the nag routine e04jaf and required about one minute on a dec alpha 3000/700 machine .
if the true @xmath0 is a gaussian , the optimal estimates will be obtained from this algorithm by choosing @xmath12 to be very large
the algorithm becomes essentially parametric in this limit , returning the normal distribution that best approximates the data .
but we would like to show that the algorithm works well even in cases where @xmath0 deviates strongly from a gaussian .
the crucial problem , of course both for this algorithm and any other lies in choosing the correct degree of smoothing .
figure 1 shows three estimates of @xmath0 based on pseudo - data generated from the very non - gaussian velocity distribution @xmath42 + 0.3\exp[{-(v - v_2)^2/2\sigma^2}]\right ) , \label{mkbf}\ ] ] with @xmath43 km s@xmath44 , @xmath45 km s@xmath44 and @xmath46 km s@xmath44 .
this @xmath0 was found by @xcite to provide a good description of the stars in the disk of the sab galaxy ngc 7217 .
merrifield & kuijken s broadening function was convolved with a template spectrum from a k0iii star , kindly provided by t. williams ; to this broadened spectrum was added gaussian random noise with amplitude @xmath47 such that @xmath48 .
the resultant spectrum was then sampled at 800 points in the wavelength range @xmath49 and these `` data '' were given to the minimization routine .
figure 1 illustrates the famous `` bias - variance '' tradeoff of solutions to ill - conditioned problems .
when @xmath12 is small , the estimated @xmath0 matches the true function in an average sense , but fluctuates strongly from grid point to grid point . for larger @xmath12 ,
the estimate is nearly correct , with fluctuations of only a few percent around the true @xmath0 . when @xmath12 is increased still more , the solution is driven toward a single gaussian with roughly the same mean and dispersion as that of the true @xmath0 .
all of these estimates for @xmath0 reproduce the input spectrum with acceptable accuracy , but @xmath0 itself is correctly recovered only when @xmath12 is chosen appropriately a characteristic feature of ill - conditioned problems .
nevertheless an appropriate choice of @xmath12 allows one to recover even this very non - gaussian @xmath0 with good accuracy .
figure 1 also shows three attempts to recover the broadening function @xmath50 \label{peak}\ ] ] with @xmath51 km @xmath52 , @xmath53 km s@xmath44 .
this @xmath0 is more peaked than a gaussian , and is designed to represent the velocity distribution in the halo of a spherical galaxy dominated by radial orbits ( @xcite ) .
once again the nature of the solution depends on @xmath12 , with large values of @xmath12 generating a gaussian approximation to the true broadening function .
but the optimal @xmath12 produces an estimate of @xmath0 that is again quite close to the true function , and the velocity dispersion derived from the estimated @xmath0 remains close to the true dispersion even when @xmath12 is far from its optimal value .
the examples just discussed were based on data with a high signal - to - noise ratio .
a more stringent test is the recovery of a non - gaussian @xmath0 from noisy data .
since a larger degree of smoothing will be required , the penalty functional will tend to drive the solution away from the true @xmath0 as the level of the noise increases .
figure 2 shows how the average error in the estimated @xmath0 varies with the signal - to - noise ratio of the data . for each value of s
/ n , 300 random realizations of the same spectrum were generated from the merrifield - kuijken broadening function and the value of @xmath12 that minimized the average , integrated square deviation between the estimates @xmath54 and the true function @xmath6 was found .
the integrated square error was defined as @xmath55 ^ 2 dv\over\int n^2(v ) dv } ; \label{ise}\ ] ] the normalizing function in the denominator was added so that a value ise @xmath56 corresponds to an rms deviation of roughly 100% between @xmath6 and @xmath54 .
the mise was then defined as the mean value of the ise over the 300 realizations using this `` optimal '' value of @xmath12 .
figure 2 shows that the mise falls roughly as a power law with s / n , with a logarithmic slope of approximately @xmath57 .
the same information is presented in more detail in figure 3 , which shows the average estimate and its 95% variability bands for each value of s / n .
this figure suggests that the existence of two peaks in @xmath0 is recoverable for s / n @xmath58 . in these examples , the optimal choice of @xmath12 and its associated mise
could be computed since the correct form of @xmath0 was known . in general one
does not know @xmath0 , of course , and the choice of @xmath12 must be based somehow on information contained within the data itself .
one way of doing this is via the `` cross - validation score '' ( @xcite ; @xcite ) .
one seeks the value of @xmath12 that minimizes @xmath59}\circ t)_i\right]^2,\ ] ] where @xmath60}$ ] is the minimizer of eq .
( [ sildisc ] ) with the @xmath61th data point left out . in effect , the cv measures the degree to which the spectral intensities predicted by @xmath0 are consistent with the observed intensities .
this is not quite the same as asking how close the estimated @xmath0 is to the true @xmath0 , but it is probably the best that one can do in practice ( @xcite ) .
figure 4 shows the result of two attempts to recover the optimal @xmath12 from fake spectra generated using the merrifield - kuijken broadening function with s / n@xmath62 and @xmath63 .
the values of @xmath12 that minimize the cv are tolerably close to the values that actually minimize the ise .
more to the point , the ise of the @xmath0 s generated using the cv estimates of @xmath12 differ only negligibly from those obtained using the optimal @xmath12 s .
these examples suggest that one can indeed hope to recover a useful estimate of the optimal smoothing parameter from the data alone . at the very least ,
such an estimate would provide a starting point when searching for the @xmath12 the produces the physically most appealing estimate of @xmath0 .
in many stellar systems , information about @xmath64 ) is most naturally obtained in the form of discrete velocities .
if the velocities are measured with negligible error , @xmath0 is simply their frequency function , which can be defined as the function that maximizes the penalized log - likelihood @xmath65 subject to the constraints @xmath66 ( @xcite ) .
the penalty functional is needed since , in its absence , the optimal estimate would be a set of delta - functions at the measured velocities .
but the uncertainty in the measured velocities is sometimes comparable to the width of @xmath0 .
for instance , radial velocities of faint stars in globular clusters may have measurement errors of a few km s@xmath44 compared to intrinsic velocity dispersions of @xmath67 km s@xmath44 .
the observable function is then not @xmath0 but rather its convolution with the error distribution , @xmath68 . assuming that the errors have a normal distribution , with dispersion @xmath47 , we have @xmath69\ dv'\ ] ] and one accordingly seeks the @xmath0 that maximizes @xmath70 subject to the same constraints on @xmath6 . following silverman ( 1982 ) , we find the solution to this constrained optimization problem as the unconstrained maximizer of the functional @xmath71 this problem is formally very similar to the deconvolution problem solved above , and a discrete representation of ( [ sildisc2 ] ) on a grid in velocity space is @xmath72 ^ 2 - n\sum_{j=1}^m \epsilon_jn_j,\ ] ] with @xmath73 , @xmath74 .
the convolution of @xmath6 with @xmath75 is again represented as a matrix , eq .
( [ matrix ] ) , with @xmath76 dv,\ ] ] @xmath77 dv,\ ] ] which can be expressed in terms of the error function .
noise in the data now comes from two sources : measurement errors , as described by @xmath47 ; and finite - sample fluctuations due to the limited number @xmath78 of measured velocities .
figure 5 shows how the mise depends on @xmath78 for data sets generated from the velocity distribution @xmath79 + \exp[{-(v - v_2)^2/2\sigma^2}]\right ) , \label{flat}\ ] ] with @xmath80 km s@xmath44 , @xmath81 km s@xmath44 and @xmath82 km s@xmath44 .
this flat - topped velocity distribution was designed to mimic @xmath0 near the projected center of a globular cluster containing an abundance of nearly - radial orbits .
( examples of such models and their velocity distributions may be found in @xcite ) .
the two sets of points in figure 5 correspond to pseudo - data generated with zero velocity errors ( circles ) and with @xmath83 km s@xmath44 ( squares ) ; the latter value is roughly one - third of the intrinsic velocity dispersion .
the mise falls off roughly as @xmath84 for both types of data , almost as steep as the @xmath85 dependence of parametric estimators .
of course the mean error is larger , at a given @xmath12 , for the sample with nonzero @xmath47 ; however an increase in sample size from @xmath86 to @xmath87 produces the same decrease in the mise as a reduction in the measurement uncertainties from @xmath88 to @xmath89 km s@xmath44 .
thus even relatively large measurement errors can be overcome by a modest increase in sample size ( assuming , of course , that the distribution of errors is well understood ) .
figure 6 shows the average estimates obtained with the optimal smoothing parameters and their 95% variability bands .
the non - gaussian nature of @xmath0 is surprisingly well reproduced even for @xmath86 , but the two peaks only begin to be clearly resolved for @xmath90 . various techniques , including a version of the cross - validation score described above , can be used to estimate optimal smoothing parameters for data like these .
the `` unbiased '' or `` least - squares '' cross - validation score ( scott 1992 , p. 166 ; silverman 1986 , p. 48 ) is defined as @xmath91}\circ p)_i \label{ucv}\ ] ] where @xmath92}$ ] is an estimate of @xmath6 obtained by omitting the @xmath61th velocity .
the value of @xmath12 that minimizes the ucv is an estimate of the value that minimizes the ise of @xmath93 .
figure 7 shows the dependence of the ucv on @xmath12 for two data sets , with @xmath86 and @xmath94 , generated from the velocity distribution of eq .
( [ flat ] ) with @xmath95 .
the minimum in both curves occurs at a value of @xmath12 close to the value that actually minimizes the ise .
the results just presented suggest that increasing the number of measured velocities in globular clusters may be a greater priority than reducing measurement errors if the goal is to determine @xmath0 , since existing techniques can already extract stellar radial velocities with greater precision than the uncertainty @xmath83 km s@xmath44 adopted here . because one would like to estimate @xmath0 at several different points in an image , the total number of velocities required for a single globular cluster would be in the thousands at least .
fortunately , data sets of this size are now becoming available for a number of stellar systems . here
we analyze radial velocities of a new sample of 4200 stars near the center of the globular cluster @xmath1 centauri .
the velocities were measured using the rutgers fabry - perot interferometer on the ctio 1.5 m telescope and were kindly made available by c. pryor , who also carried out an analysis of the measurement errors .
figure 8 shows the spatial distribution of the observed stars .
the effective field of view of the fabry - perot is about 2.75 arc minutes in radius and is offset by about 1.1 arc minutes e / se from the cluster center as determined by meylan et al .
the core radius of @xmath1 centauri is around 2.5 arc minutes ( @xcite ) so the observed velocities lie mostly within the projected core .
the core of a globular cluster is perhaps not a very auspicious place to look for non - gaussian velocity distributions .
however @xmath1 centauri is fairly young in a collisional - relaxation sense , with an estimated central relaxation time of a few billion years ( @xcite ) .
thus its velocity distribution might still be non - maxwellian .
furthermore there is evidence for two , chemically distinct populations in @xmath1 centauri ( norris et al . 1996 ) which may have formed at different epochs and hence may have different kinematics .
finally , we note that the theoretical expectation of maxwellian velocity distributions in globular clusters has rarely been tested by direct determination of line - of - sight velocity distributions . for this reason alone
, it seems worthwhile to estimate @xmath0 in @xmath1 centauri .
about 20 stars were removed from the original sample because their fabry - perot line profiles showed contamination by h@xmath12 emission .
velocity uncertainties of the remaining stars were estimated using standard procedures ( e.g. @xcite ) ; a typical estimated error was 3 - 5 km s@xmath44 .
although the estimated errors were found to correlate with stellar magnitude , this fact was ignored in the analysis and @xmath47 was simply set equal to the rms estimated uncertainty of all the stars in each subsample , about 4 km s@xmath44 in each case . for comparison , the central velocity dispersion of @xmath1 centauri is 15 - 20 km s@xmath44 ( meylan et al .
1995 ) .
one would like to estimate @xmath0 independently at many different points in this observed field of view .
however figure 6 suggests that sample sizes less than a few hundred are not very useful for detecting departures from normality .
the compromise , as illustrated in figure 8 , was to identify five , partially overlapping fields of one arc minute radius containing about one - half of the observed stars .
two fields lie along the estimated rotation axis of the cluster , at a position angle of @xmath96 east from north with their centers displaced one arc minute from the cluster center .
the three additional fields were situated along a perpendicular line .
the orientation of the rotation axis was estimated from this sample , but is consistent with a determination based on @xmath97 500 stars with velocities measured by coravel ( @xcite ) .
the offset between the fabry - perot field of view and the cluster center was used to advantage by centering the fifth field at a distance of 2.5 arc minutes from the cluster center along the direction of maximal rotation .
the number of stars in each of the subsamples , along with their mean velocity and velocity dispersion ( with estimated errors removed ) in km s@xmath44 , are given in table 1 .
the average number of stars per field was 653 for the four inner fields , with 456 stars in field 5 .
the velocity distributions in fields 1 and 5 exhibit the greatest apparent departures from normality .
figure 9 shows the dependence of the inferred @xmath0 in these two fields on the value of the smoothing parameter @xmath12 .
figure 10 displays estimates of @xmath0 in all five fields , for one choice of @xmath12 , and their estimated 95% confidence bands .
the confidence bands were computed in the usual way via the bootstrap ( scott 1992 , p. 259 ) ; the choice of @xmath12 is justified below . also shown
are the normal distributions with the same mean and variance as the inferred @xmath0 s .
none of the recovered velocity distributions is strikingly non - gaussian , although the confidence bands in fields 1 and 5 do barely exclude the normal distribution at one or more points .
the inferred @xmath0 for field 5 is more centrally peaked than a gaussian and has what might be described as a tail or bump at large positive velocities .
it is tempting to interpret this curve as resulting from the superposition of two normal distributions with different means and variances , though such an interpretation seems extravagant given the relatively small amplitude of the deviations .
the choice of the optimal @xmath12 for these estimates presented certain difficulties .
when dealing with random samples drawn from the normal distribution , the value of @xmath12 that minimizes the error in an estimate of @xmath0 would clearly be very large , since an infinite @xmath12 always returns a normal distribution . because the velocity distributions in @xmath1 centauri do appear to be quite gaussian
, we might expect the optimal @xmath12 s to be `` nearly infinite '' and hence difficult to estimate from the data alone .
in fact the ucv score ( eq . [ ucv ] ) was found not to have a minimum at any value of @xmath12 in any of these five sub - samples ; instead the function @xmath98 was always found to asymptote to a constant value as @xmath12 was increased .
a similar result was obtained using the `` likelihood cross - validation '' score ( silverman 1986 , p. 53 ) .
these results certainly do not imply that the distribution of velocities in @xmath1 centauri must be exactly gaussian , since cross - validation is not a precise prescription for determining @xmath12 and often fails to give an extremum .
but it appears that these velocity distributions are close enough to gaussian that the cross - validation technique can not find a significant difference between the estimates made with finite and infinite @xmath12 .
the following alternative scheme was adopted for selecting the optimal @xmath12 .
a crude estimate of @xmath68 can be obtained by replacing each of the measured velocities by a kernel function of fixed width .
if @xmath68 is exactly gaussian , and if the kernel is also gaussian , the optimal window width ( i.e. dispersion ) for the kernel may be shown to be @xmath99 ( silverman 1986 , p. 45 ) .
fixed - kernel estimates of @xmath68 in each of the five fields were generated from the data using this optimal @xmath100 , and compared to estimates of @xmath68 using the penalized - likelihood algorithm with various values of @xmath12 .
these comparisons could not be made precise since the fixed - kernel estimates have a larger bias and do not compensate for the velocity measurement errors .
nevertheless it is reasonable to assume that the degree of `` roughness '' in the optimal kernel - based estimates ought to be similar to that in the penalized - likelihood estimates made with the optimal value of @xmath12 .
the plots of @xmath0 in fig .
10 were made using these estimates of the optimal @xmath12 s .
we conclude from this analysis that the evidence for non - gaussian velocity distributions in our five fields near the center of @xmath1 centauri is marginal at best .
we note that the strongest deviations appear in the field that is farthest from the center .
perhaps a sample of stars even farther from the core where the relaxation time exceeds the age of the universe would show even larger departures from normality .
the @xmath1 centauri radial velocities analyzed here were obtained by k. gebhardt , j. hesser , c. pryor and t. williams using the rutgers fabry - perot interferometer . c. pryor devoted considerable time to reducing the observations and estimating the measurement uncertainties in this sample .
conversations with c. joseph , m. merrifield , h .- w .
rix , p. saha , r. van der marel and t. williams were very helpful for understanding the ins and outs of spectral deconvolution .
c. pryor read parts of the manuscript and made useful suggestions for improvements .
this work was supported by nsf grant ast 90 - 16515 and nasa grant nag 5 - 2803 .
bender , r. 1990 , , 229 , 441 dejonghe , h. 1987 , , 224 , 13 dejonghe , h. & merritt , d. 1992 , , 391 , 531 franx , m. & illingworth , g. 1988 , , 327 , l55 gebhardt , k. , pryor , c. , williams , t. & hesser , j. e. 1994 , , 107 , 2067 green , p. j. & silverman , b. w. 1994 , nonparametric regression and generalized linear models , london : chapman & hall , ch . 8 kuijken , k. & merrifield , m. r. 1993 , , 264 , 712 merrifield , m. r. & kuijken , k. 1994 , , 432 , 575 merritt , d. 1989 , in dynamics of dense stellar systems , edited by d. merritt ( cambridge university press , cambridge ) , p. 75
meylan , g. 1996 , private communication meylan , g. , mayor , m. , duquennoy , a. & dubath , p. 1995
, , 303 , 761 norris , j. e. , da costa , g. s. , freeman , k. c. & mighell , k. j. 1996 , asp conference series vol .
92 , formation of the galactic halo ... inside and out , edited by h. morrison & a. sarajedini , p. 375
peterson , c. j. & king , i. r. 1975 , , 80 , 427 rix , h .- w . ,
franx , m. fisher , d. & illingworth , g. 1992 , , 400 , l5 rix , h .- w .
& white , s. d. m. 1992 , , 254 , 389 saha , p. & williams , t. 1994 , , 107 , 1295 scott , d. w. 1992 , multivariate density estimation , new york : john wiley & sons silverman , b. w. 1982 , ann .
10 , 795 silverman , b. w. 1986 , density estimation for statistics and data analysis , london : chapman & hall thompson , j. r. & tapia , r. a. 1990 , nonparametric function estimation , modeling , and simulation ( siam , philadelphia ) van der marel , r. & franx , m. 1993 , , 407 , 525 wahba , g. 1980 , tech .
595 , dept . of statistics ,
univ . of wisconsin ,
madison , wi wahba , g. 1990 , spline models for observational data , philadelphia : siam bias - variance tradeoff in estimates of non - gaussian @xmath0 s from absorption line spectra .
input spectra were generated using two different broadening functions ( thin curves ) with added noise of amplitude s / n=20 .
( a ) - ( c ) : merrifield - kuijken broadening function , eq .
( [ mkbf ] ) ; ( d ) - ( f ) : broadening function of eq . ( [ peak ] ) .
@xmath0 was then estimated ( thick curves ) using three different values of @xmath12 .
( a ) , ( d ) : undersmoothed ; ( b ) , ( e ) : optimally smoothed ; ( c ) , ( f ) : oversmoothed . in the limit of large @xmath12 ,
i.e. infinite smoothing , the estimates tend toward a gaussian with approximately the same mean and dispersion as the true @xmath0 .
dependence of the mean integrated square error of the recovered broadening function on the signal - to - noise ratio of the spectrum , for spectra generated from the merrifield - kuijken broadening function ( eq .
( [ mkbf ] ) . for each value of s / n , 300 noise realizations of the same spectrum were generated and the value of @xmath12 that minimized the average square deviation between the true and estimated @xmath0 s was found .
the ordinate is the mise of the estimates using this optimal @xmath12 .
average estimates of @xmath0 , and their 95% variance bands , based on 300 noise realizations of spectra generated from the merrifield - kuijken broadening function ( [ mkbf ] ) . as the signal - to - noise ratio increases
, the average estimate tends toward the true broadening function and the variance of the estimates decreases .
dependence of the cross - validation score ( cv ) on the smoothing parameter @xmath12 for two spectra generated using the merrifield - kuijken broadening function ( [ mkbf ] ) , with @xmath101 and @xmath102 .
the value of @xmath12 at the minimum of the cv curve is an estimate of the value of the smoothing parameter that minimizes the integrated square error of the estimated @xmath0 .
arrows indicate the values of @xmath12 that actually minimize the ise of the broadening functions inferred from these two spectra .
dependence of the mean integrated square error of the recovered @xmath0 on the number of velocities @xmath78 , for pseudo - data generated from the flat - topped velocity distribution of eq .
( [ flat ] ) .
squares : @xmath83 km s@xmath44 ; circles : @xmath95 .
dependence of the unbiased cross validation score ( ucv ) on the smoothing parameter @xmath12 for data generated from the velocity distribution of eq .
( [ flat ] ) , with @xmath95 .
the value of @xmath12 at the minimum of the ucv curve is an estimate of the value that minimizes the integrated square error of @xmath103 .
arrows indicate the values of @xmath12 that actually minimize the ise of the velocity distributions inferred from these two data sets .
map of stars with observed radial velocities in the globular cluster @xmath1 centauri .
the origin of the coordinate system is the center of the cluster as defined by meylan & mayor ( 1986 ) ; major tick marks are separated by one arc minute .
the fabry - perot field of view is offset by approximately 1.1 arc minute from the origin .
the dashed line is an estimate of the projected rotation axis of the cluster .
velocity distributions were computed for stars in the five circular fields shown .
penalized - likelihood estimates of @xmath0 in two fields in @xmath1 centauri .
the value of the smoothing parameter @xmath12 increases from ( a ) to ( c ) .
thin lines are normal @xmath0 s with the same mean velocity and velocity dispersion as the estimated @xmath0 s .
penalized - likelihood estimates of @xmath0 in five fields in @xmath1 centauri .
heavy solid lines are the estimates ; dashed lines are 95% bootstrap confidence bands ; thin solid lines are the normal distributions with the same mean velocity and velocity dispersion as the estimated @xmath0 s .
arrows indicate mean velocities . | line - of - sight velocity distributions are crucial for unravelling the dynamics of hot stellar systems
. we present a new formalism based on penalized likelihood for deriving such distributions from kinematical data , and evaluate the performance of two algorithms that extract @xmath0 from absorption - line spectra and from sets of individual velocities .
both algorithms are superior to existing ones in that the solutions are nearly unbiased even when the data are so poor that a great deal of smoothing is required . in addition , the discrete - velocity algorithm is able to remove a known distribution of measurement errors from the estimate of @xmath0 .
the formalism is used to recover the velocity distribution of stars in five fields near the center of the globular cluster @xmath1 centauri .
rutgers astrophysics preprint series no
. 191 |
conversion gain , @xmath3 ( @xmath0 ) , is a fundamental parameter in detector characterization that is used to measure many detector properties , including read noise , dark current , and quantum efficiency ( qe ) .
measuring these parameters to a higher precision is becoming more important as the demand for low - signal observations increases and the scientific requirements evolve in complexity . the upcoming joint dark energy mission ( jdem ) , for example , requires the precise measure of the dark energy equation equation of state , @xmath4 and @xmath5 .
this measurement is derived from the luminosity distance versus redshift relationship of type ia supernovae out to high redshifts . to obtain the necessary accuracy ,
the mission requires detector characterization measurements to within 2% @xcite .
a traditional method for measuring conversion gain is the photon transfer technique @xcite .
this method , however , is limited in that it does not account for charge coupling between pixels .
charge coupling attenuates the photon shot noise of individual pixels , resulting in an underestimated variance and an overestimated gain .
figure [ f1 ] illustrates the coupling of the central pixel with the four nearest neighbors , both for ( a ) symmetric and ( b ) partially - symmetric cases .
the degree of coupling is given by coupling coefficients , @xmath1 and @xmath6 .
interpixel capacitance ( ipc ) , one of the more dominant charge coupling mechanisms , is caused by small amounts of stray capacitance .
early simulations first predicted the effects of ipc @xcite .
later studies measured the degree of coupling using various methods , including cosmic ray hits @xcite , externally calibrated capacitors @xcite , and hot pixels @xcite .
@xcite provide a detailed mathematical discussion .
these works all indicate that ignoring ipc effect within infrared detectors can result in an overestimate of gain by up to 20% .
@xcite show these effects to be as a high as a factor of 2 within the fully depleted silicon - based hybrid cmos arrays , such as the rockwell hyvisi h2rg arrays . while the conversion gain is ultimately measured via standard stars once
the instrument has been launched and commissioned ( e.g. * ? ? ?
* ) , detectors must meet specifications and science missions must be planned much earlier .
an overestimate of the gain can significantly distort the assessment of the detector and mission potential .
ipc becomes more noticeable in near - infrared detectors as pixel size decreases .
better methods of measuring @xmath3 therefore have broad applicability to astronomical detector characterization .
for example , one popular method relates the strength of the inter - pixel coupling and , subsequently , the `` true '' variance to the autocorrelation function of the detector @xcite . we discuss this method further in section [ sec : validation ] . a technique involving x - rays provides an alternative , straightforward method to measure the conversion gain .
this technique quantifies the necessary energy to liberate a single electron - hole pair in the semiconductor , known as the pair creation energy .
if the x - ray energy is known , the number of electrons liberated by each x - ray photon follows .
the conversion gain is then derived from the observed instrumental counts ( adu ) produced by a single x - ray .
already , this is a popular technique within the ccd community , where each k@xmath1 x - ray photon is known to liberate @xmath71620 electrons in silicon @xcite .
this absolute calibration offers additional benefits that aid detector design and performance studies .
for example , the calibrated response provides a fundamental standard for measuring drifts and/or changes in a detector s system gain over time .
additionally , if the conversion gain is given by a reliable alternative method , such as @xcite s autocorrelation technique , the x - ray response serves as a direct probe of the internal quantum efficiency ( qe ) , also known as the charge collection efficiency . in other words , this technique measures the detector response to individual x - rays so that the qe does not include any contribution from external factors such as the anti - reflective coating or long - wavelength photons that are not absorbed . while not commonly used by the astronomical community , the internal qe is an important measurement to detector manufacturers in their quest to build more efficient detectors .
a semi - empirical trend exists between the pair creation energy and semiconductor bandgap @xcite : @xmath8 where @xmath9 is the pair - production energy ( in ev ) and @xmath10 is the band - gap energy ( in ev ) . an exact pair - production energy , however
, is not known for the commonly used mercury - cadmium - telluride ( hgcdte ) infrared detector arrays .
the limiting factor has been the cadmium - zinc - tellruide ( cdznte ) substrate found on many hgcdte detectors , which absorbs the x - rays before they reach the depletion region of the detector .
the infrared detectors for the hst wide - field camera 3 ( wfc3 ) are not limited in this respect because they are all substrate - removed . in this paper , we provide calibrated measurements of the x - ray response of 1.7 hgcdte detectors .
section [ sec : methods ] presents our data . to convert the x - ray intensity from adu to electrons , we establish the `` true '' variance and gain of a pixel using classical propagation of errors .
section [ sec : math ] discusses the mathematical framework for this method .
monte carlo simulations validate the fidelity of this technique .
section [ sec : results ] presents the results .
we find that in response to the x - rays , these 1.7 hgcdte detectors yield 2273 @xmath2 137 electrons , which corresponds to a pair - production energy 2.61 @xmath2 0.16 ev .
we describe in detail the various systematic errors inherent within this measurement .
we have collected x - ray data from eight teledyne imaging system s ( tis ) 1.7 h1r flight grade wfc3 detectors , which are molecular beam epitaxy ( mbe ) grown and cdznte substrate - removed .
these detectors have a read noise of @xmath715 @xmath11 rms per read ( @xmath721 @xmath11 cds ) and a dark current of @xmath70.02 @xmath12 at 145 k. all data were collected at the nasa goddard space flight center ( gsfc ) detector characterization laboratory ( dcl ) .
the x - ray source was placed at a normal incidence 1.5 cm from the center of the [email protected] cm fpa .
the data were all collected in 2008 and have since been archived . for each detector
, we obtained 10 exposures , each consisting of 6 correlated double samples ( cds ) taken up - the - ramp , where each cds pair is composed of the @xmath14 and @xmath15 frames up the ramp , where @xmath16 is a number between 1 and 6 .
we performed a standard dark subtraction with median frames created from the 10 exposures .
we include plots for only fpa 152 , which is a typical representation of the other detectors we tested .
we discuss any differences observed in the other detectors .
figure [ f2a ] shows a histogram of individual pixel intensities ( in adu ) from the 10 exposures of fpa 152 .
key features include the dominant k@xmath1 peak , the less dominant k@xmath6 peak , and a low - intensity plateau ( that extends down to zero signal ) .
the two peaks show broadening likely due to a combination of several effects .
off - centered x - ray hits and charge diffusion , which occurs on the 0.1-pixel level , cause broadening to the left .
a changing band - gap energy with penetration depth causes broadening to the right .
read noise , pixel - to - pixel gain variations , and the intrinsic fano factor cause overall broadening .
we fit the peak of each profile using idl s gaussian procedure , which performs a combined gaussian and polynomial fit .
the low - intensity plateau results mostly from charge diffusion and non - centered hits .
the single pixel histogram in figure [ f2a ] , however , ignores charge coupling effects that cause some of the signal from the x - ray to appear in neighboring pixels , such as in figure [ f1 ] .
the histogram in figure [ f3 ] shows the fraction of the total charge in fpa 152 that couples to neighboring pixels ( where the total charge is defined as the intensity in the central pixel and four nearest neighbors ) .
the plot reveals a 1 - 2% coupling in all primary directions .
while there is a distinct symmetry in both the horizontal and vertical directions , the horizontal coupling tends to be larger than the vertical coupling .
this partial symmetry suggests a non - symmetric multiplexer associated with the horizontal readout mechanism .
overall , all nearest neighbors ( solid line ) show an average coupling of @xmath71.5% while a distribution of random pixels ( dotted line ) shows , as expected , no coupling to the x - rays . adding the nearest neighbor flux
presents the most straightforward method by which to restore the lost charge .
we do not perform aperture photometry .
while coupling is observed in other pixels , such as diagonal and second - neighbor pixels , monte carlo simulations reveal this effect to be negligible on the 1% level .
the additional read noise from these pixels is greater than the actual signal .
we conclude that the additional noise from these pixels would be more detrimental to the total signal than the small amount of charge lost when adding only nearest neighbors .
we consider only hits that are isolated ( defined within a 5@xmath135-pixel area ) and well - centered ( defined as being associated with the peak intensity in figure [ f2a ] , as opposed to the low - intensity plateau ) .
this threshold is indicated by the vertical line in figure [ f2a ] . in this case , `` well - centered '' is a somewhat arbitrary definition , and we find that shifting the threshold can alter the results by tens of adu .
we discuss the implications of this effect in our discussion of systematic errors in section [ sec : syserr ] .
figure [ f2b ] plots a histogram of only the selected pixel intensities with the nearest neighbor charge restored .
expectedly , the peak of this distribution shifts to a higher intensity , consistent with the fractional coupling observed in figure [ f3 ] .
the additional read noise from the neighboring pixels blends the k@xmath1 and k@xmath6 peaks .
we again take the peak by fitting the distribution with idl s gaussian routine .
the peak in this case represents the detector s response ( in adu ) to a single x - ray , for which the energy now includes a weighted contribution from the k@xmath6 line .
given a k@xmath1-to - k@xmath6 peak ratio of 7:1@xcite , we approximate the x - ray energy as a weighted average : 5.9375 kev .
the pair - production energy of the detectors ultimately requires the total intensity of the x - rays to be in electrons .
the conversion of the detector response in figure [ f2b ] from adu into electrons , therefore , must be calculated with an independent measure of the conversion gain .
the photon transfer method is insufficient because charge coupling attenuates the photon shot noise and decreases the measured variance . here
, we explore a technique to establish the `` true '' variance via the classical propagation of errors .
the technique has the advantage of allowing for a direct measurement of the covariance matrix from the data , which can serve as a diagnostic of other correlations that may be present in addition to ipc .
we find this method works well at all signal levels and is consistent with the results of @xcite .
we therefore implement this method throughout our analysis .
we begin by considering the simplest possible example of small symmetric crosstalk to the four nearest neighbors .
( a partially symmetric crosstalk consistent with figure [ f3 ] is more complicated and considered in section [ sec : partsym ] . ) as we discuss in section [ sec : methods ] , we ignore diagonal and second - neighbor pixels because coupling effects are negligible on the 1% level for our detectors . for larger ipc , other neighbors must be considered .
the detector is uniformly illuminated and only poisson noise is present ( i.e. no read noise ) .
this model is true for high signal levels for which shot noise in the signal dominates .
crosstalk is assumed to be symmetric and to only affect the nearest neighbors .
a @xmath17 pixel subarray is considered : @xmath18 the interpixel charge coupling can be represented by a detector point spread function ( psf ) that , when convolved with the subarray , `` blurs '' the charge between pixels .
this kernel is represented as : @xmath19 where @xmath1 is the coupling coefficient . in other words
, @xmath1 is the fractional crosstalk between the central pixel and its nearest neighbors .
the `` true '' signal , @xmath20 , in a pixel is therefore a linear combination of the measured signal in the central pixel of interest , @xmath21 , and its neighbors , @xmath22 where @xmath23 is some fraction of the signal in pixel @xmath24 .
in fact , equation [ eqn : lincomb ] represents the convolution of a kernel that undoes the blurring caused by the psf . in other words , @xmath25 must be the inverse psf , @xmath26 where @xmath27 is the convolution operator .
for the psf of equation [ eqn : psf ] , the inverse psf is , @xmath28 by the associative property of convolution , @xmath29 equation [ eqn : lincomb ] can be written as , @xmath30 the `` true '' variance of the pixel can be shown using the classical propagation of errors , @xmath31 where @xmath32 is the covariance between pixels @xmath33 and @xmath34 .
the covariance matrix is given as @xmath35 where @xmath36 is the number of pixels .
@xcite show that to bring a one percent correlation signal up to the noise level , the number of samples must satisfy the equation @xmath37 a coupling coefficient of @xmath71% requires @xmath38 samples .
the covariance matrix can be found experimentally ( see section [ sec : fullsym1 ] ) with @xmath38 independent integrations , but our archival data set is limited to @xmath39 integrations for each fpa . to obtain the necessary number of samples in a realistic period of time requires a unique detector readout mode capable of sampling only small windows of pixels .
while such readout modes are currently being implemented for projects such as the james webb space telescope ( jwst ) , these readout modes are unavailable for hst because they are not required by the planned science .
alternatively , the covariance matrix may be derived from theory under the assumption that noise in neighboring pixels is correlated solely with shot noise in the pixel of interest .
the theoretical covariance matrix is given as : @xmath40 equation [ eqn : covarmat ] requires knowledge of only the coupling coefficient , @xmath1 , and the measured noise , @xmath41 .
we use equation [ eqn : covarmat ] throughout our analysis .
substituting equations [ eqn : truesig ] and [ eqn : covarmat ] into equation [ eqn : properr ] gives the `` true '' variance in the pixel of interest , @xmath42 where @xmath43 is the measured variance .
@xcite derived the same results to first order in @xmath1 using fourier methods .
monte carlo simulations validate the propagation of errors technique given by equation [ eqn : newvar ] .
each simulation models charge coupling in a uniformly illuminated @xmath44 detector array .
illumination levels range in brightness from @xmath45 , where @xmath46 is the mean integrated flux per pixel .
the shot noise is distributed with a poisson distribution .
charge coupling is simulated via convolution of a psf given by equation [ eqn : psf ] , with a coupling coefficient @xmath47 .
a gauss - normal distribution of read noise , @xmath48 , is added last .
we choose these parameters to be consistent with the actual data observed in section [ sec : results ] .
the `` true '' variance of the data when no charge coupling effects are present is given as : @xmath49 the simulated charge coupling attenuates the variance .
we implement three different techniques to recover the `` true '' variance : ( 1 ) a 1 + 8@xmath1 approximation , ( 2 ) the autocorrelation technique given by @xcite , and ( 3 ) the propagation of errors technique described above in section [ sec : fullsym ] . the first method attempts to restore the variance by multiplying by a @xmath50 coefficient .
this coefficient accounts for the @xmath51 variance loss given by a small @xmath1 approximation in equation 28 of @xcite .
the technique is commonly used within the near - infrared community because of its straightforward implementation .
the coupling coefficient , @xmath1 , is typically measured during a dark exposure by utilizing the single pixel reset feature of the h2rg multiplexer to set a single pixel to a voltage different from its neighbors @xcite .
the crosstalk induced by ipc on neighboring pixels is observed directly .
the autocorrelation procedure described by @xcite relates the `` true '' variance to the autocorrelation matrix : @xmath52 + 2\sum_{i , j}d[i , j]d[i+1,j ] + \\ \left . \left . \left . \left . \left . \left .
2\sum_{i , j}d[i , j]d[i , j+1]),\right.\right.\right.\right.\right.\right.\right.\right .
\label{eqn : moore34}\end{gathered}\ ] ] where @xmath53 is the difference of two equally illuminated flat fields and @xmath36 is the number of pixels . equation [ eqn : moore34 ] makes a small @xmath1 approximation and considers all contributions from pixels beyond the four nearest neighbors to be negligible . in addition , @xcite assume ipc to be the dominant coupling mechanism . other coupling mechanisms , however , typically are present , such as correlations in the the photon arrival ( e.g. bose = einstein or hanbury - brown - twiss ) and carrier diffusion ( e.g. carrier - carrier interaction ) . despite these assumptions ,
this variance estimator is typically favored by the near - infrared community @xcite .
finally , method three implements the propagation of errors technique described in section [ sec : fullsym ] .
the `` true '' variance is given by equation [ eqn : newvar ] .
the covariance matrix in equation [ eqn : covarmat ] directly measures contributions from all additional coupling mechanisms that may be present .
additional data could potentially resolve the different coupling mechanism contributions .
nonetheless , like @xcite , we assume ipc to be the only coupling mechanism for the purposes of this paper . we simulate our real data by creating only 10 integrations .
since the propagation of errors technique requires @xmath54 independent integrations to calculate the covariance matrix experimentally , we instead use the theoretical form given by equation [ eqn : covarmat ] , where the coupling coefficient , @xmath1 , and the noise , @xmath41 , are measured directly .
section [ sec : alpha ] describes the measurement of the coupling coefficient in more detail below .
figure [ f4 ] plots the ratio of the measured to the true variance versus signal for each of the above methods . as we predict , no correction results in an underestimated variance .
the rise at small signals is due to the dominating read noise component . at high signal levels ( @xmath55 ) , the @xmath50 approximation effectively estimates the variance , although there is a disagreement of @xmath71% due to the small @xmath1 approximation . at low signal levels , however , this technique is ineffective because the dominanting read noise does not couple between pixels .
the autocorrelation technique restores the `` true '' variance to within 1% at all signal levels .
this method , however , does not fail at low signal levels because read noise does not contribute to the autocorrelation terms .
the propagation of errors technique is most effective at restoring the `` true '' variance at all signal levels .
again , there is no read noise contribution to the autocorrelation coefficients .
this technique is not as straightforward as the autocorrelation technique because it requires a direct measurement of the coupling coefficient , but it allows us to restore the variance most effectively .
for this reason , we implement the propagation of errors technique throughout the rest of this paper .
the propagation of errors technique ( equation [ eqn : newvar ] ) requires knowledge of the coupling coefficient , @xmath1 . unlike the simulations in section [ sec : validation ] , where the coefficient is a known input
, the actual coupling coefficient must be measured directly from the data .
@xcite show that @xmath1 can be extracted directly from the autocorrelation matrix of uniformly illuminated exposures .
the ratio , @xmath56 , of the nearest neighbor and unshifted autocorrelation coefficients is given as : @xmath57 table [ tab1 ] lists the autocorrelation matrix of fpa 152 .
the other detectors yield similar matrices .
the autocorrelation matrix is computed from the differenced image of two identically illuminated exposures . to account for @xmath58-noise variations , we averaged differenced images computed from pairs of flat - field exposures with varying delay times between exposures . we assume symmetry in both the horizontal and vertical directions so that the autocorrelation of each pixel with itself ( i.e. unshifted ) is given by the upper left ( [ 0,0 ] ) matrix value .
all values are normalized to the unshifted autocorrelation coefficient .
the nearest neighbors ( [ 0,1 ] and [ 1,0 ] ) have correlations that are typically between 1 - 5% .
even diagonal and second neighbor pixels show some correlation although the degree is much less than the nearest neighbors . solving for @xmath1 , equation
[ eqn : ratio ] indicates the coupling coefficients are typically between 1 - 2% .
figure [ f5 ] plots the corrected conversion gain versus signal and compares it to the uncorrected values for fpa 152 . as predicted by the monte carlo simulations , the uncorrected gain is typically overestimated by @xmath59% for coupling coefficients of @xmath71.5% .
the relevant conversion gain , @xmath3 , should correspond to signal levels equivalent to the x - ray intensity , which is highlighted by the vertical line in figure [ f5 ] ( @xmath7800 adu ) .
the figure reveals , however , that the signal levels of the gain data are typically higher . while an obvious solution would be to retake the necessary data , our analysis is limited to archival data .
instead , we attempt to estimate the conversion gain at low signal levels and minimize the associated systematic error .
figure [ f5 ] shows a correlation between the conversion gain and signal level .
this trend arises from the relationship between the capacitance of the diode and bias voltage , which is well understood for hawaii devices ( e.g. * ? ? ?
. the capacitance increases with signal level because the voltage , and hence the depletion width of the pn junction , across the diode decreases . for low signal levels ( i.e. small capacitances )
an electron generates a larger voltage swing and the gain ( in @xmath11/adu ) is smaller . to better understand the gain behavior of these devices at low signal levels , we performed alternate gain measurement techniques , including an analysis of the photon transfer of warm pixels in dark ramps and solving for the gain in dark ramps via equation 1 of @xcite .
we also compared our results to ultra - low light level data collected for similar fpas undergoing testing for the jwst near - infrared spectrograph project ( rauscher , priv . comm . ) .
these results all confirm the gain decreases with signal level and that the decrease is smooth and continuous .
we therefore choose to fit a curve to the high signal gain data and extrapolate to lower signal levels .
the choice of curve to fit is unclear because the true behavior of the gain is not well described at these low signal levels , where the data tend to be noisier .
figure [ f5a ] suggests that an exponential fit is sufficient .
figure [ f5b ] , however , indicates that a polynomial provides a better fit .
we find the likely gain is bound by the extremes of these two curves and take the conversion gain to be an average of the two .
table [ tab2 ] lists the `` true '' gain for all the detectors .
we discuss the systematic error associated with this method in section [ sec : syserr ] .
we calculate the pair - production energy by dividing the x - ray response ( in @xmath11 ) into the energy of a single x - ray , which includes a weighted contribution from the blended k@xmath6 peak .
the x - ray response ( in @xmath11 ) is given by the product of the measured peak adu intensity ( measured in section [ sec : methods ] ) with the `` true '' conversion gain ( measured in section [ sec : math ] ) .
we must first divide the gain by the detector internal qe to account for charge that may have recombined before being collected in the pixels depletion regions .
figure [ f6 ] plots the measured qe of each detector .
the relevant qe measurement from figure [ f6 ] depends on the wavelength of the photons with similar penetration depths as x - rays .
the x - ray penetration in hgcdte follows an exponentially falling probability distribution .
the attenuation length is defined as the depth , @xmath60 , into the material , measured along the surface normal , where the intensity of x - rays falls to 1/@xmath61 of its value at the surface .
our model , @xmath62 is based on data from @xcite and subsequent modeling , where @xmath60 is given in microns and @xmath63 is the x - ray energy in ev .
the x - ray penetration depth is towards the very back of these detectors and has an associated cutoff wavelength of @xmath71.68 @xcite .
the qe of a detector , however , typically combines the effects of several external factors , such as the number of photons that pass through the detector and the number of photons reflected by the detector s anti - reflective ( ar ) coating , and the internal qe , which includes only losses in the detector crystal .
since our tests measure the detector response to individual x - rays , which are either absorbed or not , we are concerned only with the internal qe . for the plot in figure [ f6 ] , we subtract the ar coating contribution with a model ar curve .
we also assume that all photons are absorbed as long as they are short - ward of the roll - off at long wavelengths .
therefore , instead of measuring the qe at precisely the associated cutoff wavelength calculated above ( @xmath71.68 ) , we measure the qe just before the sharp roll - off , as indicated by the vertical line .
we discuss the systematic error associated with this method in section [ sec : syserr ] .
table [ tab2 ] lists the resulting pair - production energy for each detector .
figure [ f7 ] plots the pair - production energy versus cutoff wavelength and bandgap energy .
fpas 129 and 153 stand out from the rest .
interestingly , these 2 detectors exhibit traits and data sets uncharacteristic of the typical fpas that we tested . for example
, fpa 129 has a large error bar traceable to the unique data available for this fpa . unlike the other fpas , the conversion gain data for fpa 129 consist of only 5 data points , all of which have signal levels @xmath645000 adu .
figure [ f5b ] indicates that the slope of the conversion gain only begins to change at signal levels @xmath655000 adu .
considering the intensity is @xmath7800 adu , we approximate the lower signal gain from an analysis of the photon transfer of warm pixels in dark ramps , which has a large associated error .
the resulting extrapolation of the fpa 129 data set is significantly less constrained and , subsequently , has a much larger systematic error than the other fpas . given the limited data set , we do not believe that the error associated with fpa 129 represents the typical systematic error associated with our procedure and therefore exclude this data set from our final average . the data point for fpa 153 in figure [ f7 ]
is clearly isolated from the other fpas .
fpa 153 has a significantly lower qe than the other fpas , especially at the cutoff wavelength .
in fact , we find that the qe for this fpa tends to swing @xmath66 between different data sets .
furthermore , fpa 153 shows the most rapid decline in conversion gain at low signal levels .
we do not completely understand the odd behavior fpa 153 exhibits , but we recognize that this is not standard behavior for these fpas .
we therefore do not include the measurement in our final average .
averaging over the remaining fpas , the pair - production energy of 1.7 hgcdte is 2.61 @xmath2 0.16 ev , which corresponds to 2273 @xmath2 137 electrons produced by each x - ray . for comparison purposes , figure [ f7 ] plots the pair - production energy of several other common semiconductor materials , as well as the semi - empirical trend given by @xcite .
the average pair - production energy for 1.7 hgcdte falls within a few percent of the value predicted by @xcite , which is within the given error .
we did not necessarily predict this agreement given that the trend - line is only semi - empirical .
in fact , the measured pair - production energy for germanium does not fall even close to the trend - line .
nonetheless , figure 1 of @xcite reveals that a majority of materials do fall reasonably close to the line .
the overall consistency of the 1.7 results indicates that the results may be an easy , straightforward sanity check for a given fpa .
for example , the inconsistent results for fpas 129 and 153 indicated a bad data set and a misbehaving fpa , respectively .
in other words , if the calculated pair - production energy for a given fpa falls far from the predicted value , the testers may wish to review the other fpa characteristics .
we identify four possible statistical errors in our measurements : the ( 1 ) detector response to the x - rays in adu , ( 2 ) the qe , ( 3 ) the conversion gain , and ( 4 ) the coupling coefficient , @xmath1 . given the large number of samples in all cases , the associated statistical errors are negligible .
specifically , we determine the detector response from the peak in figure [ f2 ] . in this case
, the large number of pixels greatly reduces the error of the mean to @xmath67% .
similarly , the large number of pixels reduces the error bars of the measured qe and conversion gain to @xmath67% .
the curves fit to the gain data subsequently have negligible error bars .
finally , equation [ eqn : covariance ] indicates that a coupling coefficient of @xmath71% requires @xmath38 samples to obtain a signal - to - noise @xmath68 . in terms of our autocorrelation matrices , individual exposures @xmath38 pixel samples ( 1024@xmath131024 array ) easily satisfy this requirement .
the systematic errors tend to be larger than the statistical errors described above .
specifically , we identify three possible systematic errors in our methods : measurements of the ( 1 ) detector response to the x - rays in adu , ( 2 ) the qe , and ( 3 ) the conversion gain .
we measure the detector response peak by fitting the histogram in figure [ f2b ] with a combined gaussian and polynomial using idl s gaussian routine .
the peak of the distribution varies , however , depending on the signal threshold chosen to indicate a well - centered hit in figure [ f2a ] .
we set the threshold at the `` kink '' where the low - intensity plateau and gaussian peak meet .
this point is rather arbitrary and may not be optimal . varying the threshold by a reasonable amount ( e.g. @xmath6510 adu ) in either direction shifts the gaussian peak in figure [ f2b ] by a similar amount . this shift , which typically results in only @xmath71% error , presents only a minor contribution to the overall systematic error . measuring
the qe introduces an additional systematic error . as discussed in section [ sec : ppe ] , the x - rays have an associated cut - off wavelength of @xmath71.68 . at these long wavelengths , however , the qe is a combination of both internal losses and unabsorbed photons .
concerned only about electrons generated by absorbed x - rays , we desire a qe that measures internal losses only .
we assume that all photons are absorbed short - ward of the qe roll - off at long wavelengths in figure [ f6 ] . instead of measuring the qe at 1.68 , we estimate a shorter wavelength at which the qe represents internal losses only , as indicated by the vertical line in figure [ f6 ] .
we assign an error of 0.05 to this estimated wavelength as it is not entirely clear where the roll - off begins .
this error corresponds to a systematic error for the qe , which depends on the degree of slope in the qe curve . for all fpas ,
the resulting qe systematic error is @xmath655% .
the measurement of the conversion gain introduces the largest systematic error to our results .
figure [ f5 ] shows the measured gain as a function of signal level .
the desired gain for our purposes must be measured at signal levels comparable to the x - ray intensities , which are typically @xmath7800 adu .
our archival gain data , however , is limited to higher signal levels than necessary . as discussed in section [ sec : gain ] , we therefore choose to fit a curve to the high signal gain data and extrapolate to lower signal levels . given the contrasting nature of the trends in figures [ f5a ] and [ f5b ] , we find the likely conversion gain at low signal levels is bound by the extremes of both an exponential and polynomial fit .
we therefore take the conversion gain to be an average of these two fits .
the associated systematic error is given by the difference between the average and the values of the two curves at the desired intensity ( in this case , @xmath7800 adu ) .
we find a systematic error of @xmath6510% in all cases .
the systematic errors are listed in table [ tab2 ] and included on the plot in figure [ f7 ] .
while these systematic errors contribute to no more than @xmath711% in quadrature , we believe these errors can be significantly reduced with more appropriate data .
for example , we measure the gain by extrapolating our data sets down to low signal levels .
future tests should be sure to measure the gain directly at these signal levels
. such foresight would reduce the systematic errors by at least 50% , if not more .
in this paper , we have calibrated the x - ray response of teledyne 1.7 hgcdte flight grade detectors for the hst wfc3 .
our conversion of the x - ray intensities from adu into electrons implements a technique that restores the `` true '' gain of the detectors using the classical propagation of errors .
the advantage to this technique is that it allows for a direct measure of the covariance matrix from the data .
monte carlo simulations have validated this technique as being most effective at restoring variance that is lost due to charge coupling effects .
the sample of eight 1.7 h1r detectors yields 2273 @xmath2 137 electrons per x - ray , which corresponds to a pair - production energy for 1.7 hgcdte of 2.61 @xmath2 0.16 ev .
this result includes both statistical and systematic errors .
better designed tests can significantly reduce these errors in the future .
nonetheless , these results now make the gain measurement technique of potential interest for near - infrared arrays , as it has been in the ccd community for many years .
the x - ray technique provides a valid measurement of the conversion gain at signal levels comparable to the x - ray intensity . this low signal level ( @xmath7800 adu )
is consistent with the signal levels at which the gain is typically reported for wfc3 detectors .
this technique provides a straightforward alternative to the photon transfer technique , which can overestimate the conversion gain within hgcdte detectors by up to @xmath720% ( and si - pin arrays by more than a factor of 2 ) when not accounting for charge coupling effects , such as ipc . the known detector response to
also will aid future detector design and performance studies , such as measuring drifts in the system gain or the internal qe .
although these measurements were made using teledyne hawaii detectors , we believe that they are likely to be applicable to hgcdte detectors from other vendors , so long as differences in cutoff wavelength are correctly accounted for . users of teledyne s older substrate - on should not use this method as the substrate absorbs the x - rays .
already , @xcite have performed similar measurements of the x - ray response for 2.5 hgcdte h2rg detectors .
they report results consistent with the overall trend in figure [ f7 ] , but these results fall significantly above the predicted line compared to the 1.7 results .
the authors use different calibration techniques and do not report a detailed review of their systematic errors . in the future , we plan to implement our method to check the 2.5 results , as well as test hgcdte with other cutoff wavelengths ( e.g. 5 ) .
this research was supported by nasa as part of the hst s wide field camera 3 program and as part of the james webb space telescope project .
ori fox wishes to thank nasa s graduate student researcher program for a grant to uva .
[ sec : appdx ]
in most hst detector arrays , the detector psf is symmetric in reflection about the @xmath60 and @xmath69 axes to within a few a few percent . for the most accurate measurements of conversion gain , this asymmetry needs to be accounted for when it is detectable . in this case
, the detector psf can be written as follows , @xmath70 and the deconvolution kernel is , @xmath71 following the same methods as were used in section [ sec : fullsym ] , we arrive at an expression for @xmath72 that is accurate to second order in @xmath1 and @xmath6 , @xmath73 because the mathematics are complex , we do not show the steps here , other than to note that all computations were done using the _ mathematica _ symbolic mathematics program . for our calibration of x - rays
, we believe that the method described in this paper ( with extensions to remove the need to assume a functional form for the detector psf as a prior ) is likely to have the advantage that the covariance matrix can be explicitly computed from the data to serve as a diagnostic of other correlations that may be present in the data in addition to ipc . unfortunately , doing this requires specialized detector readout modes that have not been implemented for hst because they are not required by the planned science .
if we do not assume a theoretical form for the covariance matrix , equation [ eqn : newvar ] can be written as follows , @xmath74 the individual @xmath32 can then be measured using the relation , @xmath75 in equation [ eqn : covar ] , @xmath16 refers to the exposure sequence number , which needs to be large for accurate measurement .
, b. j. , fox , o. , ferruit , p. , hill , r. j. , waczynski , a. , wen , y. , xia - serafino , w. , mott , b. , alexander , d. , brambora , c. k. , derro , r. , engler , c. , garrison , m. b. , johnson , t. , manthripragada , s. s. , marsh , j. m. , marshall , c. , martineau , r. j. , shakoorzadeh , k. b. , wilson , d. , roher , w. d. , smith , m. , cabelli , c. , garnett , j. , loose , m. , wong - anglin , s. , zandian , m. , cheng , e. , ellis , t. , howe , b. , jurado , m. , lee , g. , nieznanski , j. , wallis , p. , york , j. , regan , m. w. , hall , d. n. b. , hodapp , k. w. , bker , t. , de marchi , g. , jakobsen , p. , & strada ,
, , 119 , 768 schubnell , m. , barron , n. , bebek , c. , borysow , m. , brown , m. g. , cole , d. , figer , d. , lorenzon , w. , bower , c. , mostek , n. , mufson , s. , seshadri , s. , smith , r. , & tarl , g. 2006 , proc .
spie , 6276 , 62760q c c c c c c c fpa 129 & 13.74 & 3.25 & 2.98/1.50 & 2.55 & 21.6 & 2.52 ( 0.012 @xmath2 0.31 ) + fpa 148 & 14.59 & 2.64 & 1.93/1.17 & 2.34 & 11.6 & 2.64 ( 0.002 @xmath2 0.07 ) + fpa 150 & 19.80 & 2.73 & 2.07/1.29 & 2.38 & 12.7 & 2.65 ( 0.002 @xmath2 0.10 ) + fpa 152 & 19.04 & 2.72 & 1.93/1.12 & 2.40 & 11.9 & 2.47 ( 0.005 @xmath2 0.03 ) + fpa 153 & 18.02 & 2.67 & 2.02/1.11 & 2.35 & 11.9 & 1.98 ( 0.005 @xmath2 0.09 ) + fpa 154 & 18.38 & 2.67 & 1.97/1.25 & 2.23 & 16.6 & 2.60 ( 0.003 @xmath2 0.05 ) + fpa 160 & 15.68 & 2.67 & 2.04/1.28 & 2.30 & 14.0 & 2.64 ( 0.002 @xmath2 0.04 ) + fpa 165 & 13.37 & 2.62 & 2.22/1.46 & 2.18 & 17.0 & 2.68 ( 0.007 @xmath2 0.06 ) | a technique involving x - rays provides a straightforward method to measure the response of a detector .
the detector s response can lead directly to a calculation of the conversion gain ( @xmath0 ) , as well as aid detector design and performance studies .
we calibrate the x - ray energy response and pair production energy of hgcdte using 8 hst wfc3 1.7 flight grade detectors .
the results show that each k@xmath1 x - ray generates 2273 @xmath2 137 electrons , which corresponds to a pair - production energy of 2.61 @xmath2 0.16 ev .
the uncertainties are dominated by our knowledge of the conversion gain . in future studies
, we plan to eliminate this uncertainty by directly measuring conversion gain at very low light levels . |
recent work on the phenomenology of large @xmath2 gauge theories has revealed that theories based on ` moose diagrams ' ( aka ` quiver theories ' ) generate extra dimensions in the large quiver limit .
explicitly , the lagrangian takes the form of a lattice field theory plus extra irrelevant matter that completes the theory in the uv .
this so - called ` dimensional deconstruction'@xcite ( see also @xcite,@xcite ) has an elegant realization in string theory , where the quiver theories arise as the worldvolume theories of d - branes on geometric orbifolds .
some rather clever moose phonomenology has led to novel presentations of several interesting systems , including the mysterious interacting @xmath3 @xmath1 and little string theories@xcite , hinting that there is more to be learned from this subtle regulation scheme .
of course , there already exist methods for generating higher - dimensional branes , in particular matrix theory constructions@xcite which lead to non - commutative worldvolumes .
what is the relation between these two approaches ? can one deconstruct noncommutative dimensions ? in this note we present evidence that the worldvolume theories of d - branes probing @xmath4 orbifolds with discrete torsion provide , in the @xmath5 large - moose limit , precisely such a construction .
the new non - commutative directions appear exactly as in the original case ; their non - commutativity derives directly from the extra phases specifying the discrete torsion .
this can be motivated as follows .
first , as we will explain in detail below , the partition function for closed strings on the torus @xmath6 with discrete torsion @xmath7 , @xmath8 , is identical to that on the same torus but with a constant background @xmath9-field such that @xmath10 .
thus the theory of a d@xmath11-brane on a torus with discrete torsion @xmath12 is dual to the theory of a d@xmath13-brane wrapping the t - dual torus with a ( rescaled ) constant background @xmath14-field @xmath15 - which is exactly noncommutative sym on the same torus . now consider the orbifold @xmath16 with discrete torsion @xmath12 .
this is a cone over base @xmath17 ; for large @xmath2 this is a very sharp cone .
consider a thin region of this cone far from the orbifold fixed point , where the geometry is approximately @xmath18 ; in the @xmath19 limit , the local physics precisely reproduces the toroidal orbifold with discrete torsion .
this implies that the quiver theory of a d@xmath20-brane on @xmath16 with discrete torsion reproduces , in the large @xmath2 limit , the noncommutative theory of a d@xmath21-brane wrapping a torus with a constant background @xmath9-field specified by the discrete torsion . in the following
we explicitly verify this conjecture by demonstrating that this quiver theory is equivalent to sym on an @xmath22-point fuzzy torus with a noncommutativity parameter specified by the discrete torsion @xmath23 . ) , ( [ openvsclosed ] ) . ]
( in this relationship we keep the closed string volume fixed . ) in the @xmath19 limit , this precisely reproduces sym on a smooth torus with constant @xmath9-field correctly specified by the discrete torsion . in the language of @xcite
, this shows that the orbifold @xmath16 with discrete torsion can be used to `` deconstruct '' a wide class of noncommutative theories .
in particular , one can use the @xmath16 orbifolds with discrete torsion to deconstruct noncommutative d@xmath20-brane theories for @xmath24 . for @xmath25
, one can take a further strong coupling limit to obtain a deconstruction of ncos theory@xcite . for @xmath26 ,
the d - brane sym theories are not uv complete ; it is a remarkable fact that the deconstructed theories appear to contain precisely the degrees of freedom required to complete the d - brane theories to @xmath1 @xcite or little string theories @xcite , respectively , as demonstrated in @xcite .
we begin by reviewing the original ` deconstruction ' phenomenon@xcite .
we then recall the quiver theories on branes probing orbifolds with discrete torsion and demonstrate how they deconstruct non - commutative theories .
we compare the deconstruction of noncommutative d - branes to their matrix theory constructions , finding agreement in the large moose limit , and argue that finite moose provide stringy realizations of gauge theories on fuzzy tori .
we close with further speculations and open questions .
( for related earlier work , see eg @xcite . )
consider the worldvolume theory of a single d0-brane on a supersymmetric @xmath27 orbifold .
( a d3-brane probe of this orbifold was used in @xcite to deconstruct the six - dimensional ( 2,0 ) theory . )
the orbifold can be thought of as a local model for an @xmath28 singularity in a k3 manifold , preserving half of the original supersymmetry .
a general technique for constructing worldvolume theories of d - branes on orbifolds was described in a remarkable paper by douglas and moore @xcite ; we follow their procedure . parameterizing the target space with five real scalars @xmath29 and two complex scalars @xmath30 and @xmath31 , the geometric action of the @xmath32 generator is @xmath33 the massless worldvolume fields of the parent @xmath34 @xmath35 gauge theory are gauge fields @xmath36 , scalars @xmath37 , and majorana - weyl spinors in the * 16 * of @xmath38 , @xmath39 . as in @xcite
, we construct @xmath40 , a weyl spinor of @xmath41 , out of the components of @xmath42 having @xmath43 or @xmath44 .
similarly the weyl spinor @xmath45 will contain the components of @xmath42 with either @xmath46 or @xmath47 .
when acting on d - branes , the orbifold group may have an additional action on the chan - paton indices .
the action of the generator @xmath48 of the @xmath32 orbifold group can thus be written @xmath49 where the @xmath50-matrices belong to a faithful representation of the orbifold group . in the case at hand
, we can express @xmath51 in a convenient basis as @xmath52 the fields surviving the orbifold projection ( [ fullorbifoldaction ] ) are thus @xmath53 this spectrum can be conveniently represented by so called moose or quiver diagrams ( see fig .
figure 1 : the @xmath54 moose . on
the left is the moose for @xmath55 , @xmath56 , and @xmath57 . on the right
is the moose for @xmath58 , @xmath59 , and @xmath60 . the classical potential , descending from the potential of the parent theory by restricting to fields which survive the orbifold projection , includes the following term for every @xmath61 : @xmath62 there are also terms quartic in @xmath63 and @xmath64 forcing @xmath65 and @xmath66 to be independent of @xmath67 .
the moduli space has a coulomb branch , where all @xmath68 and @xmath69 vanish and where @xmath70 can be independently varied .
it also has a higgs branch with @xmath71 we have added a factor of @xmath72 to ( [ potentialforx ] ) so that the higgs expectation values ( [ higgsz1z2 ] ) correspond to moving a d0-brane ( consisting of @xmath2 fractional d0-branes ) to a distance @xmath73 from the orbifold point .
the trick is now to study low energy fluctuations around a particular point on the higgs branch given by some fixed values of @xmath74 and @xmath75 .
the leading order potential for @xmath76 , for example , will be @xmath77 for large @xmath2 and @xmath78 slowly varying with @xmath67 , this looks very much like a lattice discretization of @xmath79 with @xmath80 and effective lattice spacing @xmath81 .
it is a stimulating and life - affirming exercise to check that the rest of the lagrangian takes the correct form to reproduce , in the continuum limit , the worldvolume theory of a d1-brane .
there is a simple geometric reason that this works . for large @xmath2
the orbifold is a sharp cone over base @xmath82 . for large @xmath83 ,
far from the fixed locus , the local geometry seen by the d0-brane is approximately a cylinder of radius @xmath84 , with the compact coordinate being in the @xmath85 direction @xcite .
t - duality along this coordinate ( which is valid far from the fixed locus ) produces a d1-brane wrapping an @xmath86 of radius @xmath87 and string coupling @xmath88 with these values of @xmath89 and @xmath90 , the factor in front of the integral in ( [ potentialforcontinuousx ] ) becomes precisely the d1-brane tension @xmath91 it is simple and remarkable to check that the interaction terms in the d1-brane world - volume theory thus obtained are correctly normalized .
we can of course t - dualize without approximating the geometry by a cylinder @xcite .
the t - dual of the full orbifold geometry involves @xmath2 ns5-branes evenly spaced along the t - dual circle , while the fractional d0-branes at the orbifold point become d - strings stretched between the ns5-branes .
it is easy to find similar quiver theories which mock up higher dimensional d - branes at energies lower than @xmath92 ( alternatively , whose ir physics in the limit of @xmath93 reproduces a higher - dimensional d - brane worldvolume theory ) . in supersymmetric cases ,
the number of supercharges is increased in this limit , because an orbifold generally breaks some supersymmetry , whereas a torus with periodic boundary conditions does not . in our example , the number of supercharges is doubled , giving 16 real supercharges in the end .
the scalings of the various couplings also deserve attention .
consider for concreteness the deconstruction of @xmath94 @xmath95 theory with small but finite coupling @xmath96 .
this proceeds as above by studying light fluctuations off the higgs branch of the cyclic moose , similar to @xcite .
let @xmath97 be the gauge coupling in the original @xmath98 quiver theory .
out along the higgs branch , the effective coupling of the surviving @xmath95 gauge theory is @xmath99 the relation between the @xmath94 and @xmath100 couplings is as usual for kaluza - klein reduction , @xmath101 expressed in terms of the coupling in the original quiver theory , the @xmath94 coupling is thus @xmath102 so holding the @xmath94 coupling and ( emergent ) radius @xmath89 fixed and finite while taking @xmath19 gives finite @xmath103 but requires taking the original coupling of the higgsed moose large , @xmath104 the upshot is that , while the gauge coupling of the original quiver theory is getting large , the gauge couplings of both the @xmath105 d@xmath20-branes far form the orbifold fixed point and of the deconstructed d@xmath21-branes can be kept arbitrarily weak .
more generally , consider the deconstruction of a @xmath106 dimensional theory with small but finite t hooft coupling @xmath107 . in terms of the effective lattice spacing @xmath108 and quiver coupling @xmath97 ,
@xmath109 the ( very strongly higgsed ! )
original quiver theory is thus strongly coupled in the continuum limit @xmath110 . on the other hand , for @xmath2 large but fixed energy scale comparable to @xmath111 in the deconstructed theory
, these configurations have extremely large higgs expectation values @xmath112 , with the result that the low - energy excitations are not localized in the moose but spread over many gauge groups .
this makes the effective interactions small , so the dynamics can be studied perturbatively , which is the statement that the effective @xmath113-dimensional t hooft coupling @xmath114 can be held weak in the continuum limit .
second , the matrix hamiltonian , which gives for example ( [ potentialforx ] ) , is usually said to be valid only for small separations between the d - branes .
naively , one might worry that ( [ potentialforx ] ) becomes inapplicable when @xmath83 approaches the string length @xmath115 .
however , as was discussed in e.g. @xcite , the true limitation is to energies lower than the string scale , i.e. to strings shorter than @xmath115 .
this requires @xmath116 so that the model will be rich enough to accurately describe the physics .
this bound ( [ stinglengthinequality ] ) translates to requiring the lattice spacing @xmath108 to be _ larger _ than the string length , @xmath117 happily , this is not an obstacle , as we are interested in the decoupling limit , i.e. in the physics at energies much lower than @xmath118 .
finally , let s review some of the salient properties of the @xmath119 orbifold _ without _ discrete torsion .
( this orbifold was used to deconstruct ( 1,1 ) little string theory in @xcite . )
choosing three real transverse coordinates @xmath120 and three complex coordinates @xmath121 the geometric action of the two orbifold group generators is @xmath122 preserving one quarter of the original supersymmetry .
the field content corresponding to @xmath105 transverse type ii d@xmath20-branes ( @xmath123 ) descends from a configuration of @xmath124 d@xmath20-branes in the parent theory ; a judicious choice of basis gives the action on gauge indices as @xmath125 where @xmath126 and @xmath127 .
the surviving spectrum is summarized by the quiver diagram on fig .
2 . figure 2 : @xmath16 moose for bifundamental scalars .
note that it is periodic . in @xmath100 @xmath128 language , @xmath129 bifundamental chiral superfields survive , plus adjoints from vector multiplets for each of the @xmath22 @xmath95 gauge groups .
( of course , if @xmath130 , we have to dimensionally reduce these fields . ) giving identical vevs to all ` horizontal ' and ` vertical ' bifundamentals in the quiver diagram generates two discretized dimensions forming a rectangular torus .
giving a further vev to the ` diagonal ' bifundamentals gives instead a slanted torus . in both cases ,
the resulting low energy theory can be thought of as a latticized version of the worldvolume theory of @xmath105 wrapped d(@xmath131)-branes .
( see @xcite for details ) .
as we have seen , d@xmath20-branes on @xmath132 orbifolds deconstruct d@xmath21-branes .
we will argue that discrete torsion makes the two new world - volume coordinates non - commutative in an appropriate continuum limit .
we begin with a review of quiver theories on d - branes probing orbifolds with discrete torsion , discuss the basic strategy , and proceed with an explicit example .
first let us review the physical meaning of discrete torsion , relate it to a t - dual @xmath9-field , and give an overview of the logical structure of our construction .
orbifolds with discrete torsion @xcite generalize geometric orbifolds by adding to the twisted sectors of the path integral phases which depend on the orbifold group elements defining that sector .
modular invariance forces the discrete torsion to lie in @xmath133 , i.e. the torsion is a two - cocycle of the orbifold group .
a trivial example is the torus , @xmath134 , whose partition function is a sum over winding @xmath135 twisted sectors @xmath136 adding discrete torsion ammounts to adding phases of the form @xmath137 importantly , this is identical to the partition function for the torus with a constant background longitudinal @xmath9-field with @xmath138 , so the torus with discrete torsion is identical to the torus with constant background @xmath9-field .
this fits with the fact that the @xmath9-field takes values in @xmath139 .
let s consider the @xmath16 orbifold with discrete torsion @xmath140 labeled by the integer @xmath141 ( mod @xmath2 ) in @xmath142 for @xmath2 large , this is a sharp cone over base @xmath17 , which in particular contains a @xmath143 factor . by taking @xmath19 ( and @xmath144
fixed ) while moving away from the orbifold fixed point so as to keep the volume of the torus fixed , we recover the background @xmath145 , where the twisted sectors of the orbifold become the winding sectors on the torus , so the discrete torsion of the orbifold becomes discrete torsion on the torus . now probe the @xmath16 torsion orbifold with a d@xmath11-brane far from the fixed point and again consider the constant volume , large @xmath2 limit .
from the above , this limit is identical to the theory of a d@xmath11-brane on the same torus in the presence of a background @xmath9-field .
t - dualizing both legs of the torus gives a d@xmath13-brane wrapping the dual torus with a rescaled background @xmath9-field , whose worldvolume theory is sym on a noncommutative torus with noncommutativity given by @xmath146 .
thus we can realize noncommutative sym as the large-@xmath2 limit of the quiver theory of a d - brane on a @xmath16 orbifold with discrete torsion .
detailed study of this quiver theory reveals that , in this limit , the theory becomes precisely sym on an @xmath22-point fuzzy torus , with noncommutativity given in terms of the volume of the torus and the discrete torsion .
but the large @xmath2 limit of this fuzzy sym is exactly sym on a noncommutative torus .
thus the quiver theory of a d@xmath20-brane probing a @xmath16 with discrete torsion precisely deconstructs the worldvolume theory of a d@xmath21-brane wrapping a noncommutative torus .
this can be expressed in the commuting diagram ( figure 3 ) = 0.8 figure 3 : deconstruction is t - duality .
note that @xmath147 .
the scalings indicated are for fixed volume of the torus seen by closed strings .
this provides a simple and useful physical interpretation of discrete torsion . since the possibility of including discrete torsion depends on the non - vanishing of @xmath148 , having discrete torsion means that the orbifold can be presented as a fibration whose fibres contain 2-cycles .
the interpretation suggested by our analysis is that the discrete torsion should be understood as the @xmath9-field along ( the t - dual of ) each fibre .
the utility of this interpretation is that it applies just as well in the case of orbifolds with fixed points ( at which points the fibration becomes singular ) , where fractional branes can wrap the shrunken cycle , as to orbifolds with freely acting orbifold groups ( non - singular fibrations ) , such as @xmath149 , where there is no shrunken cycle to wrap . in the remainder of this section
we explicitly verify the above story for @xmath16 . as explained in @xcite , chan - paton indices in the worldvolume theories of d - branes probing orbifolds with discrete torsion transform in projective representations of the orbifold group , @xmath150 , where the phase @xmath151 again lies in @xmath133 . in the following we focus on the orbifold group @xmath152 , for which @xmath153 , so the choice of discrete torsion is specified by one number , @xmath154 . for @xmath2 and @xmath154 relatively prime ( which we will assume throughout for simplicity ) , there is a unique irreducible projective representation of the orbifold group , which can be realized as @xmath155 matrices involving the usual clock and shift operators . to get a representation describing @xmath105 d@xmath20-branes on the orbifold , we tensor this irreducible representation with @xmath156 matrices .
we emphasize that this phenomenon works for general @xmath9-fields . to motivate this , consider deconstruction of a general manifold @xmath157 via an orbifold with quantum symmetry @xmath158 such that @xmath159 in the continuum limit .
the discrete torsion is classified by @xmath160 ; in the continuum limit this becomes @xmath161 , which classifies the @xmath9-field along the emergent dimensions .
the physics of d - branes in the supersymmetric @xmath119 orbifold with discrete torsion was described in @xcite .
the two - cocycle classes @xmath162 of @xmath163 will be represented by @xmath164 where @xmath165 for @xmath2 even and @xmath166 for @xmath2 odd , @xmath167 labels the possible choices of discrete torsion , and we restrict for simplicity to @xmath2 and @xmath154 relatively prime ( as in @xcite ) . with these conventions , @xmath168 generates @xmath32 .
the geometric action of the orbifold group is @xmath169 we will be interested in the case of @xmath105 d@xmath170-branes , consisting of @xmath124 fractional d@xmath20-branes , probing the orbifold . in the parent @xmath171 sym theory , we have , in the @xmath172 language , one vector multiplet @xmath173 and three chiral superfields @xmath174 and @xmath175 .
( in general , we will use a hat to denote matrices @xmath176 ) .
the @xmath177-matrices acting on the chan - paton sector can be chosen as @xmath178 we define shift and clock matrices @xmath179 and @xmath180 , satisfying @xmath181 , as in @xcite ( where they were called @xmath182 and @xmath183 ) . for odd @xmath2 ,
@xmath184 for even @xmath2 , @xmath179 is as above and @xmath180 is defined using @xmath185 as @xmath186 the fields left invariant by the full orbifold action are of the form @xmath187 the superpotential of the parent theory may be written @xmath188 where @xmath189 is the d@xmath20-brane tension for string coupling @xmath190 .
the orbifold - projected superpotential is then @xmath191 where we have added a factor of @xmath192 as in section [ ordinaryd - branedeconstruction ] .
the f- and d - terms are @xmath193 @xmath194
^ 2 + [ \phi_2 , \phi_2^\dagger]^2 + [ \phi_3 , \phi_3^\dagger]^2\right ) , \label{d - term}\ ] ] where @xmath195 means @xmath196 . in the continuum limit , @xmath197 corresponds to a stack of coincident d@xmath21-branes ; for simplicity we fix @xmath198 .
@xmath199 note that ( [ backgroundforfuzzytorus ] ) has zero energy and thus lies on the moduli space .
it is the spectrum of light fluctuations off the moduli space that reveals the presence of emergent dimensions .
focussing for clarity on scalars , we rewrite the bosonic parts @xmath200 and @xmath201 of the chiral superfields @xmath202 and @xmath203 as @xmath204 where @xmath205 are hermitian @xmath156 matrices , @xmath206 are @xmath156 unitary matrices , and @xmath207 with these conventions , the background ( [ backgroundforfuzzytorus - vanishingz3 ] ) corresponds to @xmath208 and @xmath209 . substituting these into ( [ f - term ] ) , we get , among other terms , @xmath210 this is precisely the plaquette operator of @xmath95 gauge theory on a fuzzy torus @xcite , with @xmath211 and @xmath212 being the usual link variables ! it is a remarkable fact that the full set of fluctuations flesh out a certain fuzzy torus gauge theory : besides the @xmath179-@xmath180 part ( [ plaquette ] ) of @xmath213 , " we can identify the @xmath20-@xmath179 and @xmath20-@xmath180 parts coming form the kinetic terms for @xmath202 and @xmath203 . together with the original @xmath20-@xmath20 piece , this forms the kinetic term of a gauge field living in @xmath20 continuous spacetime dimensions and two discrete dimensions forming a fuzzy torus .
further , @xmath214 and @xmath215 appear as adjoint scalars on this fuzzy torus , the extra pieces of their kinetic terms appearing in ( [ f - term ] ) and ( [ d - term ] ) . for example , the d - term gives @xmath216 " , while
@xmath217 " originates from the f - term .
there are also terms involving other fields and other interactions which one might not have expected in a simple gauge theory on a fuzzy torus .
while some of them reflect the fact that changing different expectation values effectively deforms the fuzzy torus ( e.g. from a flat to a slanted torus ) , some do not have an immediately obvious interpretation .
this should probably not be too much of a surprise , as the most naive stringy realization of gauge theory on a fuzzy torus , i.e. the matrix theory construction , is unstable .
the procedure for taking the continuum limit of fuzzy geometries is standard and will not be repeated here .
our interest now lies in finding the scalings of various physical quantities in this limit . to identify the lattice spacing
, we will use the fact that the normalization of the term in the lagrangian involving @xmath218 and @xmath219 is @xmath220 comparing ( [ plaquette ] ) and ( [ l - dot - terms ] ) to the normalization in @xcite we get @xmath221 the radii and volume of the torus are @xmath222 @xmath223 the most relevant quantity , of course , is the emergent noncommutativity parameter @xmath224 . for this purpose ,
we formally write @xmath225 @xmath226 = i \theta^{\ \ ! \ ! \tilde 1 \tilde 2}.\ ] ] since we defined @xmath168 , and @xmath227 for @xmath2 even and @xmath166 for @xmath2 odd , the relation @xmath228 implies @xmath229 from the overall normalization and using the results of @xcite , we can also read off the gauge coupling @xmath230 we will continue this discussion in section 3.5 after relating the fuzzy moose to the matrix theory construction of higher - dimensional d - branes .
the standard matrix theory construction of @xmath105 noncompact d@xmath21-branes with @xmath19 d@xmath20-branes@xcite begins with the ordinary matrix lagrangian in @xmath231 and expands around a background of @xmath232 matrices satisfying @xmath233 = i \theta^{\ \ ! \ !
\tilde 1 \tilde 2 } , \label{commutatorx1x2}\ ] ] the other @xmath234 being zero . the matrices @xmath235 and @xmath236 should generate the space of @xmath22 linearly independent matrices , so that any @xmath232 matrix can be expressed as a @xmath237 matrix whose entries are functions of @xmath238 and @xmath239 .
the background ( [ commutatorx1x2 ] ) satisfies the equations of motion , and by studying its fluctuations , one recovers the non - commutative field theory describing the higher - dimensional d - branes .
note that the matrix potential can be obtained by dimensional reduction of the kinetic term and the superpotential of @xmath240 sym theory .
expressed in the @xmath241 variables , the superpotential is @xmath242 which reproduces ( [ orbifoldsuperpotential ] ) in the limit @xmath243 , if we choose @xmath244 . a finite @xmath2 analog of ( [ commutatorx1x2 ] ) can be constructed with the same matrices as in sections [ fuzzydbranesfromznzn ] and [ thecomtinuumlimit ] , i.e. @xmath245 however , this background does not minimize the potential and will evolve with time . for this reason , we should talk only about very large d - branes , for which the decay is slow .
alternatively , one might add other terms to the potential , which would stabilize ( [ finitenmatrixbackground ] ) . to our knowledge
, such a stable construction has not been realized within string theory .
let s compare this to the construction in sections [ fuzzydbranesfromznzn ] and [ thecomtinuumlimit ] .
the backgrounds about which we expand , ( [ backgroundforfuzzytorus - vanishingz3 ] ) and ( [ finitenmatrixbackground ] ) , are formally identical , differing in the @xmath246 factor in the orbifold superpotential ( [ orbifoldsuperpotential ] ) . in general , this is an important distinction .
on the other hand , we are free to take @xmath247 ( @xmath248 ) . in this limit
, the physics of the two approaches should be the same .
indeed , this can be explicitly checked . in both cases we end up with a stack of d@xmath21-branes . since we want to keep @xmath224 in ( [ thetas ] ) fixed , the d@xmath21-branes will be very large , decompactifying in the strict limit .
this is precisely the situation in which the matrix theory configuration becomes stable .
now , we would like to compare the scalings of various parameters in the matrix theory ( for very large @xmath249-branes ) to the quiver theory scalings found in section [ thecomtinuumlimit ] . at first sight , they seem manifestly different ; the matrix background ( [ finitenmatrixbackground ] ) describes a torus with radii @xmath250 and @xmath251 , while the fuzzy moose radii ( [ fuzzytorusradii ] ) are inversely proportional to @xmath74 and @xmath75 !
sober second thoughts reveal that it is incorrect to compare the radii in this way .
the radii @xmath74 and @xmath75 are as measured by the closed string metric , while those in ( [ fuzzytorusradii ] ) should be compared to open string quantities .
more precisely , recall that there is an infinite number of possible descriptions of non - commutative theories @xcite , differing by the choice of the two - form @xmath252 ( not to be confused with the chiral superfields @xmath253 ) appearing in the commutation relations @xmath254 = i \theta^{\ \ ! \ ! \tilde i \tilde j } , \quad [ \partial_{\tilde i } , x^{\tilde j } ] = i \delta_{\tilde i}^{\tilde j } , \quad [ \partial_{\tilde i } , \partial_{\tilde j } ] = -i \phi'_{\tilde
i \tilde j } \quad ( \tilde i , \tilde j = \tilde 1 , \tilde 2)\ ] ] as explained in @xcite , the choice which matrix theory naturally selects is @xmath255 ( this applies also to the fuzzy moose theory . ) for this value of @xmath256 , the relation between open and closed string parameters is @xmath257 @xmath258 and @xmath259 are the open string metric and coupling , respectively , while @xmath260 and @xmath190 denote their closed string counterparts . here
we condense notation , manipulating matrices as if they had only indices @xmath261 and @xmath262 and suppressing other components . using the continuum results of @xcite in a frame where @xmath263 we can express the @xmath9-field along the brane as @xcite @xmath264 where @xmath265 is the volume of the torus as seen by closed strings . the open string metric ( from [ openvsclosed ] ) is @xmath266 giving the volume seen by open strings @xmath267 the corresponding gauge coupling is @xmath268 where we have used @xmath269 , as identified in the previous section .
these are exactly the results given by the fuzzy moose , provided we set @xmath270 in choosing the discrete torsion . at first sight
this is somewhat disconcerting ; why does the moose have this extra parameter that does not appear in the strictly infinite-@xmath2 matrix theory , and what does it mean physically , anyway ? are the fuzzy moose with different @xmath154 really different theories ?
the resolution comes from morita equivalence in the non - commutative theory , which derives from t - duality in the original theory .
as promulgated by seiberg and witten @xcite , morita equivalence relates a noncommutative theory on a flat torus with metric @xmath258 , gauge coupling @xmath271 and rational theta parameter @xmath272 to a commutative theory with parameters @xmath273 as the rescalings of the metric and gauge coupling do not depend on @xmath154 , the fuzzy moose with the different @xmath154 we consider are all morita equivalent to the same commutative theory on the same torus with the same coupling , and thus equivalent to each other .
this equivalence must be read with a bit of care . for finite @xmath2 ,
t - duality on the orbifold is more subtle than on the cylinder ( in particular , since winding is conserved only @xmath274 @xmath2 , the dual momenum is conserved only @xmath274 @xmath2 ) , so the morita equivalence may be only approximate .
this in fact seems necessary , since the orbifold theories with discrete torsion for any finite @xmath2 and different @xmath154 ( again all relatively prime ) appear manifestly different - the surviving @xmath32 quantum symmetry groups are embedded differently in the @xmath132 , and the quiver theory superpotentials contain different phases .
it is only in the strict large @xmath2 limit that this naive t - duality is exact and the theories become truly identical .
happily , it is precisely in this limit that we compare to the strictly infinite-@xmath2 matrix theory construction @xcite , which has no such parameter in the first place .
this remarkable agreement provides further evidence that the fuzzy moose agrees with matrix theory in the large @xmath2 limit . that they disagree slightly at finite @xmath2
again should not be worrying : for finite @xmath2 , the matrix theory background does not solve the f - term constraints and is thus not stable , while the moose theory is .
physically , the two presentations are essentially two different regularizations of the noncommutative theory which need agree only in the deregulated limit - which they do .
we have presented considerable evidence that quiver theories living on d - branes probing orbifolds with discrete torsion deconstruct higher - dimensional non - commutative theories in the large - moose limit .
the lagrangian of the moose theory far along its higgs branch reproduces , for large moose , the lagrangian ( [ f - term ] ) , ( [ plaquette ] ) for the fuzzy torus . in the strict large moose limit
this becomes a gauge theory on a noncomutative torus .
the fuzzy moose agrees with the matrix theory construction in the strict large @xmath2 limit .
it is remarkable that the fuzzy moose is completely well defined for all @xmath2 , providing a novel and consistent regularization of noncommutative theories and an explicit realization in string theory of gauge theory on a fuzzy torus .
this begs the question of how the fuzzy moose encodes the uv / ir correspondence of the continuum non - commutative theory .
for example , non - susy fuzzy moose theories should deconstruct non - susy noncommutative theories , in which we expect ir poles in physical processes arising from uv degrees of freedom @xcite ; how are these divergences regulated in the fuzzy moose theory ?
this should be a fruitful ground for exploration .
one immediate extension of our construction is to note that since the strong coupling limit of @xmath275 sym is ncos theory , taking the strong coupling limit of the fuzzy moose d@xmath276-branes should provide an explicit deconstruction of ncos theory .
it is also tempting to speculate about theories one might ( de)construct from more baroque orbifold geometries . along these lines , it seems that similar arguments might be used in orbifolds of the conifold with discrete torsion , which have been studied extensively in @xcite .
additionally , while the fuzzy moose on d@xmath13- or d@xmath277-branes naively deconstructs noncommutative @xmath278- and @xmath279-brane theories , arguments similar to those in @xcite suggest that they should actually deconstruct some uv completion of these theories , namely some generalization of @xmath1 and little string theories , or perhaps even some more general 5-brane theory with a 3-form generalization of non - commutativity .
we would like to thank n. arkani - hamed , a. cohen , k. dasgupta , l. motl , m. van raamsdonk , e. silverstein , l. susskind and y. zunger for very useful discussions and comments on drafts of this paper .
m. f. would particularly like to thank l. motl for extensive correspondence .
a. a. is supported in part by an nsf graduate fellowship , by doe contract de - ac03 - 76sf00515 , and by viewers like you .
m. f. is supported in part by a stanford graduate fellowship , by nsf grant phy-9870115 , and by the institute of physics , academy of sciences of the czech republic under grant no .
ga - avr a10100711 .
both authors are supported by the beautiful planet earth .
n. arkani - hamed , a. g. cohen and h. georgi , `` ( de)constructing dimensions , '' phys . rev .
lett . * 86 * ( 2001 ) 4757 [ arxiv : hep - th/0104005 ] .
c. t. hill , s. pokorski and j. wang , `` gauge invariant effective lagrangian for kaluza - klein modes , '' phys .
d * 64 * ( 2001 ) 105005 [ arxiv : hep - th/0104035 ] .
n. arkani - hamed , a. g. cohen and h. georgi , `` electroweak symmetry breaking from dimensional deconstruction , '' phys .
b * 513 * ( 2001 ) 232 [ arxiv : hep - ph/0105239 ] . n. arkani - hamed , a. g. cohen and h. georgi , `` accelerated unification , '' [ arxiv : hep - th/0108089 ] .
n. arkani - hamed , a. g. cohen and h. georgi , `` twisted supersymmetry and the topology of theory space , '' [ arxiv : hep - th/0109082 ] .
h. c. cheng , d. e. kaplan , m. schmaltz and w. skiba , `` deconstructing gaugino mediation , '' phys .
b * 515 * ( 2001 ) 395 [ arxiv : hep - ph/0106098 ] . c. csaki , g. d. kribs and j. terning , `` 4d models of scherk - schwarz gut breaking via deconstruction , '' [ arxiv : hep - ph/0107266 ] . k. sfetsos , `` dynamical emergence of extra dimensions and warped geometries , '' nucl .
b * 612 * ( 2001 ) 191 [ arxiv : hep - th/0106126 ] .
h. chankowski , a. falkowski and s. pokorski , `` unification in models with replicated gauge groups , '' [ arxiv : hep - ph/0109272 ] . i. rothstein and w. skiba , `` mother moose : generating extra dimensions from simple groups at large n , '' [ arxiv : hep - th/0109175 ] .
t. banks , n. seiberg and s. h. shenker , `` branes from matrices , '' nucl .
b * 490 * ( 1997 ) 91 [ arxiv : hep - th/9612157 ] .
m. li , `` strings from iib matrices , '' nucl . phys .
b * 499 * ( 1997 ) 149 [ arxiv : hep - th/9612222 ] .
h. aoki , n. ishibashi , s. iso , h. kawai , y. kitazawa and t. tada , `` noncommutative yang - mills in iib matrix model , '' nucl .
b * 565 * ( 2000 ) 176 [ arxiv : hep - th/9908141 ] .
n. ishibashi , `` a relation between commutative and noncommutative descriptions of d - branes , '' [ arxiv : hep - th/9909176 ] .
n. ishibashi , s. iso , h. kawai and y. kitazawa , `` wilson loops in noncommutative yang - mills , '' nucl .
b * 573 * ( 2000 ) 573 [ arxiv : hep - th/9910004 ] .
i. bars and d. minic , `` non - commutative geometry on a discrete periodic lattice and gauge theory , '' phys .
d * 62 * ( 2000 ) 105018 [ arxiv : hep - th/9910091 ] .
j. ambjorn , y. m. makeenko , j. nishimura and r. j. szabo , `` finite n matrix models of noncommutative gauge theory , '' jhep * 9911 * ( 1999 ) 029 [ arxiv : hep - th/9911041 ] .
j. ambjorn , y. m. makeenko , j. nishimura and r. j. szabo , `` nonperturbative dynamics of noncommutative gauge theory , '' phys .
b * 480 * ( 2000 ) 399 [ arxiv : hep - th/0002158 ] .
j. ambjorn , k. n. anagnostopoulos , w. bietenholz , t. hotta and j. nishimura , `` large n dynamics of dimensionally reduced 4d su(n ) super yang - mills theory , '' jhep * 0007 * ( 2000 ) 013 [ arxiv : hep - th/0003208 ] .
j. ambjorn , y. m. makeenko , j. nishimura and r. j. szabo , `` lattice gauge fields and discrete noncommutative yang - mills theory , '' jhep * 0005 * ( 2000 ) 023 [ arxiv : hep - th/0004147 ] .
l. alvarez - gaume and s. r. wadia , `` gauge theory on a quantum phase space , '' phys .
b * 501 * ( 2001 ) 319 [ arxiv : hep - th/0006219 ] .
a. h. fatollahi , `` gauge symmetry as symmetry of matrix coordinates , '' eur .
j. c * 17 * ( 2000 ) 535 [ arxiv : hep - th/0007023 ] .
n. seiberg , `` a note on background independence in noncommutative gauge theories , matrix model and tachyon condensation , '' jhep * 0009 * ( 2000 ) 003 [ arxiv : hep - th/0008013 ] .
r. c. myers , `` dielectric - branes , '' jhep * 9912 * ( 1999 ) 022 [ arxiv : hep - th/9910053 ] .
n. seiberg , l. susskind and n. toumbas , `` strings in background electric field , space / time noncommutativity and a new noncritical string theory , '' jhep * 0006 * , 021 ( 2000 ) [ arxiv : hep - th/0005040 ] .
r. gopakumar , j. maldacena , s. minwalla and a. strominger , `` s - duality and noncommutative gauge theory , '' jhep * 0006 * ( 2000 ) 036 [ arxiv : hep - th/0005048 ] .
m. berkooz , m. rozali and n. seiberg , `` on transverse fivebranes in m(atrix ) theory on @xmath280 , '' phys .
b * 408 * ( 1997 ) 105 [ arxiv : hep - th/9704089 ] .
n. seiberg , `` new theories in six dimensions and matrix description of m - theory on @xmath280 and @xmath281 , '' phys .
lett.b * 408 * ( 1997 ) 98 [ arxiv : hep - th/9705221 ] .
w. i. taylor and m. van raamsdonk , `` multiple d0-branes in weakly curved backgrounds , '' nucl .
b * 558 * ( 1999 ) 63 [ arxiv : hep - th/9904095 ] .
w. i. taylor and m. van raamsdonk , `` multiple dp - branes in weak background fields , '' nucl .
b * 573 * ( 2000 ) 703 [ arxiv : hep - th/9910052 ] .
m. r. douglas and g. w. moore , `` d - branes , quivers , and ale instantons , '' [ arxiv : hep - th/9603167 ] .
a. adams , j. polchinski and e. silverstein , `` do nt panic ! closed string tachyons in ale space - times , '' [ arxiv : hep - th/0108075 ] .
h. ooguri and c. vafa , `` two - dimensional black hole and singularities of cy manifolds , '' nucl .
b * 463 * ( 1996 ) 55 [ arxiv : hep - th/9511164 ] .
a. hanany and a. zaffaroni , `` on the realization of chiral four - dimensional gauge theories using branes , '' jhep * 9805 * ( 1998 ) 001 [ arxiv : hep - th/9801134 ] .
b. feng , a. hanany , y. h. he and n. prezas , `` discrete torsion , covering groups and quiver diagrams , '' jhep * 0104 * , 037 ( 2001 ) [ arxiv : hep - th/0011192 ] .
b. feng , a. hanany , y. h. he and n. prezas , `` discrete torsion , non - abelian orbifolds and the schur multiplier , '' jhep * 0101 * , 033 ( 2001 ) [ arxiv : hep - th/0010023 ] .
m. r. douglas , `` d - branes and discrete torsion , '' [ arxiv : hep - th/9807235 ] . m. r. douglas and b. fiol , `` d - branes and discrete torsion .
ii , '' [ arxiv : hep - th/9903031 ] .
j. ambjorn , y. m. makeenko , j. nishimura and r. j. szabo , `` lattice gauge fields and discrete noncommutative yang - mills theory , '' jhep * 0005 * ( 2000 ) 023 [ arxiv : hep - th/0004147 ] .
d. bigatti , `` gauge theory on the fuzzy torus , '' [ arxiv : hep - th/0109018 ] .
y. kimura , `` noncommutative gauge theories on fuzzy sphere and fuzzy torus from matrix model , '' prog .
* 106 * ( 2001 ) 445 [ arxiv : hep - th/0103192 ] .
k. dasgupta , s. hyun , k. oh and r. tatar , `` conifolds with discrete torsion and noncommutativity , '' jhep * 0009 * , 043 ( 2000 ) [ arxiv : hep - th/0008091 ] .
d. berenstein and r. g. leigh , `` discrete torsion , ads / cft and duality , '' jhep * 0001 * , 038 ( 2000 ) [ arxiv : hep - th/0001055 ] .
d. berenstein , v. jejjala and r. g. leigh , `` marginal and relevant deformations of n = 4 field theories and non - commutative moduli spaces of vacua , '' nucl .
b * 589 * , 196 ( 2000 ) [ arxiv : hep - th/0005087 ] .
d. berenstein , v. jejjala and r. g. leigh , `` noncommutative moduli spaces and t duality , '' phys .
b * 493 * , 162 ( 2000 ) [ arxiv : hep - th/0006168 ] .
a. hanany and b. kol , `` on orientifolds , discrete torsion , branes and m theory , '' jhep * 0006 * , 013 ( 2000 ) [ arxiv : hep - th/0003025 ] .
n. seiberg and e. witten , `` string theory and noncommutative geometry , '' jhep * 9909 * , 032 ( 1999 ) [ arxiv : hep - th/9908142 ] .
s. minwalla , m. van raamsdonk and n. seiberg , `` noncommutative perturbative dynamics , '' jhep * 0002 * ( 2000 ) 020 [ arxiv : hep - th/9912072 ] . m. hayakawa , `` perturbative analysis on infrared and ultraviolet aspects of noncommutative qed on @xmath282 , '' [ arxiv : hep - th/9912167 ] .
a. matusis , l. susskind and n. toumbas , `` the ir / uv connection in the non - commutative gauge theories , '' jhep * 0012 * ( 2000 ) 002 [ arxiv : hep - th/0002075 ]
. m. van raamsdonk , `` the meaning of infrared singularities in noncommutative gauge theories , '' [ arxiv : hep - th/0110093 ] . | we argue that the worldvolume theories of d - branes probing orbifolds with discrete torsion develop , in the large quiver limit , new non - commutative directions .
this provides an explicit ` deconstruction ' of a wide class of noncommutative theories .
this also provides insight into the physical meaning of discrete torsion and its relation to the t - dual b field .
we demonstrate that the strict large quiver limit reproduces the matrix theory construction of higher - dimensional d - branes , and argue that finite ` fuzzy moose ' theories provide novel regularizations of non - commutative theories and explicit string theory realizations of gauge theories on fuzzy tori .
we also comment briefly on the relation to @xmath0 , @xmath1 and little string theories .
* deconstructing noncommutativity * * with a * * giant fuzzy moose * allan adams and michal fabinger department of physics and slac stanford university stanford , ca 94305 - 4060 |
neutron stars in low - mass x - ray binaries ( lmxbs ) provide observational evidence for their rapid spins reluctantly , and via increasingly diverse phenomena .
highly coherent burst oscillations , occuring only around the peak of thermonuclear ( type - i ) bursts , were the first such phenomenon to be discovered , and since have been detected in @xmath2 sources ( e.g. * ? ? ?
continuous pulsations in the persistent emission occur even more infrequently , and occur at just above the burst oscillation frequency in those sources which exhibit both @xcite .
this result supports the hypothesis that the burst oscillation frequency traces the neutron star spin .
most recently , intermittent persistent pulsations have been detected in several sources , including one previously known burst oscillation source ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . it is uncertain why some sources show pulsations or burst oscillations and others do not . additionally , sources which exhibit burst oscillations do not do so in every burst .
in fact , the presence of oscillations can be as rare as 1 burst in 14 ( for 4u 1916@xmath0053 ; see * ? ? ? * ) .
the mechanism by which the burst oscillations are produced is yet another uncertainty .
although oscillations are observed at high ( fractional ) amplitudes early in some bursts , consistent with a spreading `` hot spot '' model ( e.g. * ? ? ?
* ) , the oscillations in the burst tail , when the burning must have spread to the entire stellar surface , are harder to explain .
the oscillations may instead ( or also ) arise from anisotropies in the surface brightness originating from hydrodynamic instabilities @xcite or modes excited in the neutron star ocean ( e.g. @xcite ; see also @xcite ) .
the low - mass x - ray binary exo 0748@xmath0676 is particularly well - studied .
this transient was discovered during _
exosat _ observations in 1985 @xcite , which also revealed the first thermonuclear bursts from the source as well as x - ray dipping activity .
synchronous x - ray and optical eclipses @xcite are observed once every 3.82 hr orbit .
a variety of low- and high - frequency variability has been characterised ( e.g. * ? ? ?
* ; * ? ? ?
* ) , notably including a 695 hz quasi - periodic oscillation .
more recently , absorption features in the summed x - ray spectra of bursts observed by _ xmm - newton _ were identified as redshifted lines from near the neutron star surface ( @xmath3 ; * ? ? ?
the narrowness of these features requires that the neutron star is rotating slowly , as rotation speeds @xmath4 hz will broaden the line profiles to the point where they are undetectable @xcite .
subsequently , @xcite detected a 45 hz peak in the summed power spectrum of 38 thermonuclear bursts , which they interpreted as a spin frequency sufficiently slow to give negligible broadening .
however , subsequent followup studies have failed to confirm the spectral line detection ( e.g. * ? ? ?
here we present analysis of recently observed bursts from exo 0748@xmath0676 which suggest that the neutron star spin is not 45 hz , but 552 hz .
we analysed observations of exo 0748@xmath0676 made with the proportional counter array ( pca ; * ? ? ?
* ) onboard the _ rossi x - ray timing explorer _
( _ rxte _ ) .
the pca consists of five identical , co - aligned proportional counter units ( pcus ) , sensitive to photons in the energy range 260 kev .
a passive collimator restricts incoming photons to a circular field - of - view with radius @xmath5 .
photon counts from the pca are processed independently by up to 6 event analyzers ( eas ) in a variety of configurations , permitting time resolution down to @xmath6 and up to 256 spectral channels . in an earlier paper we presented a detailed study of the properties of all thermonuclear bursts in public _
rxte _ data through 2007 june ( * ? ? ?
* hereafter g08 ) .
additional observations are continually being made public , and we are periodically analysing these newly available data to extend the g08 sample .
exo 0748@xmath0676 was observed intensively throughout 20067 to monitor low- and high - frequency quasi - periodic oscillations .
we analysed 223 additional observations to the g08 sample , and detected 67 additional thermonuclear bursts .
our burst detection and analysis procedures
are identical to those of g08 . one component of the analyses is a search for burst oscillations from sources that have not previously exhibited them .
for each burst , we extract a lightcurve over the full pca energy range and binned on @xmath7 .
we then search for oscillations while the burst flux is at least 10% of the peak value for that burst .
for exo 0748@xmath0676 , the search duration in all the bursts was typically in the range 1290 s , or 40 s in the mean .
the search is carried out by computing fast fourier transforms ( ffts ) of overlapping windows of data of length 1 , 2 , and 4 s , stepped by 0.25 s. for each power spectrum we search for excess power at frequencies @xmath8 hz , below which red noise from the burst rise and decay dominates .
we set a detection threshold corresponding to @xmath9 significance taking into account the number of trials for each window .
however , we consider a single detection at this confidence level , or detections in windows that overlap in time , insufficient to confirm a burst oscillation .
robust detections require exceeding our threshold in power spectra from multiple independent time windows , either in the same burst or other bursts from the same source , and at frequencies within a few hz .
we measured power far exceeding the noise level in power spectra of two bursts from exo 0748@xmath0676 , on 2007 january 14 14:08:44 ut ( mjd 54114.58941 ) and 2007 december 13 13:29:20 ut ( mjd 54447.56204 ) . in the first burst , a power spectrum of a 1-s window of data beginning 0.125 s before the burst start ( here defined as the time at which the burst flux first exceeded 25% of the peak flux , following g08 ) exhibited a peak leahy power of 59.68 at a frequency of 552 hz ( fig .
[ ffts ] ) . the second burst , occurring almost a year later , in a 2 s window of data beginning 0.5 s before the burst start reached a leahy power of 48.26 at a frequency of 552.5 hz .
while both these detections are highly significant ( with single trial probabilities of @xmath10 and @xmath11 ) we did not detect the signal in multiple independent ( non - overlapping ) time windows in either burst .
however , the correspondence of the detection frequencies in two separate bursts virtually guarantees the signal is real .
we estimated the likelihood of two such peaks separated by less than 1 hz arising by chance , as the product of the probabilities for each detection , multiplied by the probability related to their separation . assuming a uniform distribution for noise peaks over the full frequency range of 104000 hz , the probability for two such nearby detections is roughly @xmath12 . taking into account all trials in the blind search covering all 157 bursts from exo 0748@xmath0676 in the extended g08 catalog
the estimated null hypothesis probability is @xmath13 , equivalent to @xmath14 .
this estimate includes no correction for the lack of independence between the searches in overlapping time windows ; such correction would only increase the significance .
we also measured the distribution of noise powers in the absence of a signal for simulated lightcurves matching the observed 0.125-s lightcurve within each time window in which the oscillation was detected .
we confirmed that the noise powers are distributed as @xmath15 with two degrees of freedom , which supports our estimated significance .
thus , we conclude that the signal is a genuine burst oscillation .
we folded the lightcurve in each time window in which the 552 hz signal was detected initially , to measure the amplitude and harmonic content of the profile . since the power spectra in each case have rather limited frequency resolution , we folded the data on a grid of frequencies and chose the frequency value which maximised the variance in the folded profile .
the corresponding frequency value for the january 14 burst was 552.02 hz , while for the december 13 burst was 552.45 hz .
we tested for the presence of harmonic content from the power spectrum of the folded pulse profile , following @xcite .
the maximum leahy power for the 2007 january burst at @xmath16 the oscillation frequency was 6.47 , which was below our detection threshold .
the power at 3 or more times the oscillation frequency was comparable or smaller , and even smaller powers were measured for the december burst .
thus , we found no evidence for significant power at the 2nd or higher harmonics during either burst .
we fitted each profile with a sinusoidal model , giving fractional rms amplitudes of @xmath17% and @xmath18% for the january 14 and december 13 bursts , respectively .
the uncertainties were estimated by fixing the amplitude on a grid of values , re - fitting the model with the remaining parameters free , and determining the range for which the @xmath19 ( for 1-@xmath20 error ) . to determine the energy dependence of the oscillation we subdivided the pca energy band in five intervals adjusted to give approximately the same source counts in each .
we then created background - subtracted lightcurves for photons within each energy band and folded each lightcurve on the best - fit pulse frequency determined above .
the rms pulse fraction ( for a single sinusoid ) increased steadily from @xmath21% between 24 kev , to 20% above 9 kev , consistently in both bursts .
the arrival phase of the pulse varied weakly below @xmath22 kev , but appeared to arrive earlier by @xmath23 in phase ( @xmath24 ms ) above this energy .
we also computed oversampled ffts covering each burst to measure the evolution of the oscillation . in both bursts we found evidence of an increase in the signal frequency , which is characteristic of burst oscillations ( fig .
[ pds ] ) .
the signal appears initially at a lower frequency ( approximately 551 hz ) in the january 14 burst and increases in frequency more rapidly .
the overall frequency drift for the two bursts was no more than 2 hz ( 0.4% ) .
following the detection , we performed a narrowband search for the signal in all 154 other bursts detected by _ rxte _ from exo 0748@xmath0676 , and for which high time resolution data was available .
as in g08 , we computed ffts of each 1 s interval of data for the first 16 s of the burst , and searched for signals within 5 hz of the 552 hz oscillation frequency .
we considered a signal to be a detection if it had less than a 1% chance of occurring due to noise given the 160 trial frequencies searched for each burst , and also if it persisted for two adjacent ( independent ) time and frequency bins with a chance probability of @xmath25 , or if it occurred in the first second of a burst with a chance probability of @xmath26 . with these criteria we found two additional ( albeit weak ) detections , in bursts on 2007 may 19 12:55:55 ut and 2007 dec 5 12:15:34 ut . in the may 19 burst oscillations
were detected weakly in the peak ( defined as the interval during which the count rate exceeded 90% of the maximum ) and tail ( from the peak onwards ) , whereas in the december 5 burst the oscillations were detected in the rise , as with the initial detections .
we also revisited the analysis of @xcite , to measure the significance of the 45 hz signal in the full sample of 157 bursts observed by _ rxte _ from exo 0748@xmath0676 to date .
we first reproduced the analysis of @xcite with their original sample of 38 bursts and the same energy selection and lightcurve rebinning strategy , obtaining similar results for the 45 hz peak and its significance .
next , we applied the same technique to the larger sample of 157 bursts now available , but did not detect the signal ( using a detection threshold corresponding to @xmath27 single - trial significance ) . finally , noting that the 38 bursts used by @xcite all occurred at times when the persistent flux ( measured by the _ rxte_/asm daily average count rates ) was atypically low , mostly below @xmath28 count s@xmath29
therefore , we also searched the subset of 129 bursts from the full sample that also occurred during such low flux intervals and applied the same search technique .
however , the @xmath30 hz signal was again not detected .
there are substantial differences in the characteristics of the two burst oscillations ( 45 hz and 552 hz ) now detected in exo 0748@xmath0676 .
the 45 hz signal had a fractional amplitude of @xmath31% ( rms ) and was detected in the summed power spectrum of 38 bursts , rebinned by a factor of ( typically ) 64 or 128 to give a resolution of 1 hz , calculated from light curves selecting photons within the energy range 660 kev , and covering intervals of typically 64 or 128 s of the burst decay .
in contrast , the 552 hz signal has a fractional amplitude of 15% ( rms ) , was detected separately in unbinned power spectra calculated from 1- and 2-s light curves extracted over the full pca energy range ( @xmath3260 kev ) during the rise of two individual bursts , separated by 11 months .
additionally , the dynamic power density spectra give evidence for frequency evolution while the 552 hz signal was present , similar to that observed in other burst oscillation sources .
the key question is , which of these two signals traces the neutron star spin ?
we consider three alternatives .
first , assuming both signals are genuine , the 45 hz signal may arise from the neutron star spin , in which case the 552 hz signal may arise from a high - order ( @xmath3313 ) radial mode ( @xcite ; see also @xcite ) .
however , the appearance of this mode alone is difficult to explain ; one would expect the excitation of many different modes , rather than just two with widely - separated orders .
second , if both signals are real but the 552 hz signal instead indicates the neutron star spin , there is no known mechanism that can give rise to a 45 hz signal from the neutron star surface . as suggested by @xcite
, the 45 hz oscillation may instead arise in the boundary layer between the disk and neutron star .
third , the lack of a detection of the 45 hz signal in the larger sample of bursts detected since 2004 suggests that the 45 hz signal may have arisen from statistical fluctuations .
our analysis shows that including many more bursts in the power spectral sum has the effect only of reducing the detection significance to well below any conservative detection threshold .
the signal power is also quite sensitive to other parameters of the data selection .
thus , while it is possible that the 45 hz oscillation is transient and has not been detectable since , it is also difficult to rule out an origin in statistical noise with any confidence .
the properties of the 552 hz oscillation closely resemble those of the burst oscillations observed in other systems , that are believed to trace the neutron star spin to within a few hz .
in contrast , the properties of the 45 hz oscillation are quite distinct , especially its low frequency and the inability to detect it in individual bursts .
thus , we conclude that the 552 hz signal almost certainly traces the neutron star spin in exo 0748@xmath0676 ; since the signal appears to increase by up to 2 hz during the burst rise , we expect that the true spin frequency may be a few hz higher .
the energy dependence of the 552 hz oscillation amplitude in exo 0748@xmath0676 was similar to that of other burst oscillation sources @xcite , although the overall amplitudes were higher .
the suggestion of the hard ( @xmath34 kev ) photons arriving earlier than the soft photons is however contrary to previous observations . in that respect
we note that the previous analysis focussed principally on oscillations present throughout the tails of bursts , whereas the oscillations in exo 0748@xmath0676 were present only in the rise .
it is not known whether the two types of oscillations have characteristically different variation of pulse arrival time with energy .
the duty cycle of the 552 hz oscillation ( the number of bursts in which it was detected compared to the total number observed ) is extremely small at @xmath35% , the smallest yet for all the burst oscillation sources ( e.g. g08 ) .
the unprecedented scarcity of the oscillation raises the question of what was so unusual about the bursts that exhibited them .
compared to the entire sample of bursts observed by _ rxte _ from exo 0748@xmath0676 , the bursts on january 14 and december 13 had somewhat shorter timescales ( calculated as the ratio of peak flux to fluence ) @xmath3613 s , while the typical range is 1530 s. similarly , the rises were of shorter duration than average , at 24 s. shorter burst timescales suggest a smaller proportion of hydrogen in the burst fuel than usual .
correspondingly , while the fluences were rather typical , the bursts reached higher than average peak fluxes .
however , other bursts were observed with similar properties which did not exibit oscillations , so these properties do not uniquely determine their observeability .
the january 14 burst occurred only 11.4 min after the previous event , and had a fluence only about 77% lower .
such short recurrence time bursts are common for exo 0748@xmath0676 , and many examples have been detected by _
rxte _ ( e.g. g08 ) as well as _
xmm - newton _ @xcite .
however , the second burst in these pairs is typically much fainter than the first .
interestingly , it is also possible that the december 13 burst was the second in a closely - spaced pair , as a data gap ( due to the 90 min satellite orbit ) prevented observations until approximately 7 min before the event . again , timing analysis of other such examples did not reveal any additional oscillations , however .
we also compared the properties of the persistent emission at the time when the bursts with oscillations occurred , to other public _ rxte _ observations of the source ( from g08 ) .
the persistent flux in both observations was @xmath37 ( 2.525 kev ) , approximately equal to the 50th percentile value of the flux distribution .
similarly , the hard and soft spectral colors were in the middle of the observed range , at @xmath38 and @xmath39 , respectively .
thus , we found no evidence for an unusual spectral or intensity state at the time of the bursts with oscillations .
a 552 hz spin frequency for exo 0748@xmath0676 means that spectral features arising from the neutron star surface will be significantly doppler broadened , as well as reducing the central line depth to only @xmath40% of the continuum level ( e.g. * ? ? ?
narrow lines from such rapidly - rotating objects can only arise if the system is viewed at extremely low inclinations ; however , the eclipses and dipping activity in exo 0748@xmath0676unambiguously indicate a high system inclination .
thus , the narrow ( @xmath24 ) spectral features reported by @xcite could not arise from the neutron star surface .
this conclusion is corroborated by other recent work , including the absence of lines in a deeper _ xmm - newton _
spectrum @xcite , as well as difficulties explaining the inferred fe column @xcite . however , the narrow spectral features are difficult to identify otherwise ( e.g. * ? ? ?
* ) , and a satisfactory explanation remains elusive .
we thank tony piro , randy cooper , lars bildsten , al levine and mike muno for useful discussions .
this research has made use of data obtained through the high energy astrophysics science archive research center online service , provided by the nasa / goddard space flight center . | we report the detection of pulsations at 552 hz in the rising phase of two type - i ( thermonuclear ) x - ray bursts observed from the accreting neutron star exo 0748@xmath0676 in 2007 january and december , by the _ rossi x - ray timing explorer_. the fractional amplitude was 15% ( rms ) .
the dynamic power density spectrum for each burst revealed an increase in frequency of @xmath12 hz while the oscillation was present .
the frequency drift , the high significance of the detections and the almost identical signal frequencies measured in two bursts separated by 11 months , confirms this signal as a burst oscillation similar to those found in 13 other sources to date .
we thus conclude that the spin frequency in exo 0748@xmath0676 is within a few hz of 552 hz , rather than 45 hz as was suggested from an earlier signal detection by @xcite .
consequently , doppler broadening must significantly affect spectral features arising from the neutron star surface , so that the narrow absorption features previously reported from an _ xmm - newton _ spectrum could not have arisen there .
the origin of both the previously reported 45 hz oscillation and the x - ray absorption lines is now uncertain . |
the spatio - temporal extension and evolution of the interaction region in heavy - ion collisions are not directly observable .
indirect experimental access to its geometry and dynamics is possible through hanbury brown - twiss ( hbt ) intensity interferometry @xcite .
however , the interpretation of the measured hbt correlations is in general model dependent , and the question arises to what extent their interpretational ambiguity can be reduced by a refined analysis of the data .
in general , hbt radius parameters measure the gaussian widths ( second central moments ) of the source distribution in space - time @xcite .
it is the finite lifetime of the particle source in heavy ion collisions which complicates their interpretation . for a static boson emitting source ,
the hbt - radii have a _ unique _ interpretation in terms of geometrical source sizes . for dynamical sources like those created in heavy ion collisions , however ,
the hbt - radii measure certain linear combinations of the lifetime , the geometrical sizes and other space - time correlations @xcite . furthermore ,
if the source expands all hbt parameters become functions of the pair momentum @xcite . for
azimuthally symmetric sources , corresponding to heavy - ion collisions at zero impact parameter , there exist two different `` complete '' gaussian parametrizations for the correlation function : the cartesian parametrization with parameters @xmath0 , @xmath1 , @xmath2 and @xmath3 @xcite , and the yano - koonin - podgoretski ( ykp ) parametrisation with parameters @xmath4 , @xmath5 , @xmath6 and @xmath7 @xcite . in each case , for expanding sources , these parameters are functions of the pair momentum .
the @xmath8-parameters have the dimension of a length while @xmath7 is a velocity .
the cartesian parametrization has the additional difficulty that the value and the spatio - temporal interpretation of its parameters depend strongly on the longitudinal rest frame of the observer .
the ykp radius parameters @xmath4 , @xmath5 , and @xmath6 , on the other hand , are independent of the longitudinal velocity of the frame in which the particle momenta are measured .
the fourth ykp fit parameter , the yano - koonin velocity @xmath7 , singles out a specific longitudinal rest frame ( relative to the observer ) in which the spatio - temporal interpretation of the ( longitudinally boost - invariant ) ykp - radius parameters becomes particularly simple .
in fact , for `` transparent '' sources ( i.e. sources , for which particle emission occurs from the whole volume and is not surface dominated ) without collective transverse expansion , it was shown in @xcite that @xmath4 , @xmath5 , and @xmath6 give exactly the transverse , longitudinal and temporal widths respectively , of the source emission function in the yano - koonin frame where @xmath9 .
we will show here that in this case , up to small corrections from asymmetries of the source rapidity profile due to longitudinal expansion flow , @xmath7 coincides with the longitudinal velocity of the fluid element around the point of highest emissivity in the source , such that @xmath4 , @xmath5 and @xmath6 measure ( and cleanly separate ) the transverse , longitudinal and temporal lengths of homogeneity of the source in the rest frame of the emitter . as discussed in sec .
[ sec4a ] , the pair momentum dependence of @xmath7 allows to measure the longitudinal expansion of the source in a very direct way . for the class of models studied here
, we will also show ( see sec .
[ sec4d ] ) that in the absence of transverse collective expansion the ykp parameters show perfect @xmath10-scaling , i.e. , they are , for a given source , universal functions of the transverse mass @xmath11 of the particle pair , independent of the particle rest mass .
this can be tested by comparing @xmath12 , @xmath13 , and @xmath14 correlations .
the same is not true for the cartesian parameters which contain additional kinematic and frame dependent factors which distinguish between pairs of particles with different mass .
this type of @xmath15 scaling of the ykp parameters is violated for sources with transverse collective expansion ; also other types of transverse @xmath16-@xmath17-correlations , like e.g. those occurring in opaque sources @xcite , and final state effects like resonance decays after freeze - out @xcite , can break this scaling . for a detailed discussion of these effects
see refs .
@xcite ( opaque sources ) and @xcite ( resonance decays ) . one of the purposes of this paper is to investigate the specific effects on the correlator and the related corrections to the spatio - temporal interpretation of the ykp parameters introduced by transverse expansion flow in the source .
these will in general depend on the particular source model , and such an investigation thus must necessarily involve an extensive model study .
we investigate here numerically a simple parametrization of the source which implements the finite longitudinal , transverse and temporal extension and the longitudinal expansion of realistic heavy - ion generated sources and contains the transverse expansion flow as a tunable parameter .
we will see that strong transverse flow affects the interpretation of the ykp parameter @xmath6 as the lifetime of the source ( in its own rest frame ) and thus renders the extraction of the duration of particle emission from the correlation data more difficult and less quantitatively reliable .
on the other hand , it also spoils the @xmath10-scaling of the ykp radius parameters which opens the possibility for an independent estimate of the transverse flow velocity from comparison between pion and kaon correlations .
clearly , such an estimate will remain somewhat model - dependent , but the mechanisms isolated in the present paper should still be very useful for qualitative consistency checks between data and theoretical interpretation .
the investigation reported here has two aspects : an analytical and a numerical one . on the analytical level ,
we study the connection between the ykp parameters and the second space - time moments of the emission function , discussing the dependence of the latter on various geometric and dynamical features of particle emission .
this discussion is largely model - independent ; it is much more detailed than the short account given in @xcite and should thus serve as a general basis of understanding which can be used to qualitatively anticipate the behaviour of the ykp parameters also for other source models than the one studied here .
the numerical side of our study is , of course , model - dependent and our quantitative results must therefore be regarded with the necessary caution .
our paper is organized as follows : in sec .
[ sec2 ] , we shortly compare the cartesian and ykp parametrizations , thereby setting up our notation . in sec .
[ sec3 ] , we introduce a class of hydrodynamical models for the emission function . sec .
[ sec4 ] contains a general discussion and a detailed numerical study of these models .
we focus in particular on the effects of collective expansion flow on the ykp parameters .
the main results are summarized in sec .
we shortly recall the basic relations between the emission function @xmath18 , the measured two - particle correlation function @xmath19 , and the different gaussian parametrisations of this correlator in terms of cartesian or ykp radius parameters .
we start from the relation @xcite ( here written down for bosons ) @xmath20 here , the emission function @xmath21 is the ( wigner ) phase space density of the boson emitting sources @xcite and denotes the probability that a boson with momentum @xmath17 is emitted from the space time point @xmath16 .
it specifies the one - particle momentum spectrum @xmath22 as well as the two - particle correlation @xmath19 . the r.h.s . of ( [ 1.1 ] ) has to be evaluated at @xmath23 ( the average momentum of the particle pair ) and @xmath24 ( their corresponding relative momentum ) where the @xmath25 are on - shell .
the fourier transform in ( [ 1.1 ] ) does not have a unique inverse since the four components of the relative momentum @xmath26 are not independent , due to the on - shell constraint @xmath27 which follows from @xmath28 . in practice the analysis of hbt correlation data must therefore be based on a comparison with specific models for the emission function @xmath18 , with the aim of constraining the class of reasonable " model sources as far as possible .
an important tool for this procedure are the model - independent expressions for the hbt parameters @xcite which allow to calculate from an arbitrary emission function @xmath29 the characteristic parameters of the two - particle correlation function @xmath30 by simple quadrature .
experimentally , these hbt parameters are obtained via a multidimensional gaussian fit to @xmath19 in momentum space . to compute these gaussian parameters of the ( momentum ) correlation function @xmath30 it is sufficient to use the gaussian approximation of the ( space - time ) emission function @xmath29 , @xmath31 + \delta s(x , k ) \
, , \ ] ] neglecting @xmath32 @xcite . here , the @xmath33 denote the space - time coordinates relative to the effective `` source centre '' @xmath34 for pions with momentum @xmath35 , @xmath36 and @xmath37 is the inverse of the gaussian curvature tensor in ( [ 1.3 ] ) , adjusted such that the first term in ( [ 1.3 ] ) reproduces the rms width of the full source @xmath18 .
the ( @xmath35-dependent ) expectation values in these definitions are defined as space - time averages over the emission function : @xmath38 the correction term @xmath39 contains information on the deviation of the emission function @xmath18 from a gaussian form in coordinate space , i.e. on sharp edges , wiggles , secondary peaks , etc . for the class of models discussed in this paper , however , the contributions from @xmath39 are known to have little influence on the half width of the correlation function @xcite , and can be neglected .
then , the two - particle correlation function @xmath19 can be calculated analytically from ( [ 1.1 ] ) : @xmath40 \ , .
\end{aligned}\ ] ] it is fully determined by the @xmath35-dependent second space - time moments @xmath41 of the source ( the effective widths " @xmath42 or lengths of homogeneity " @xcite ) . in general , a gaussian parametrisation of @xmath19 is specified by selecting a particular choice of three independent components of the relative momentum @xmath26 and implementing in ( [ 1.6 ] ) the on - shell constraint @xmath28 accordingly .
this is usually done in a cartesian coordinate system with @xmath43 along the beam axis and @xmath35 lying in the @xmath16-@xmath43-plane .
one labels the @xmath43-component of a 3-vector by @xmath44 ( for _ longitudinal _ ) , the @xmath16-component by @xmath45 ( for _ outward _ ) and the @xmath46-component by @xmath47 ( for _ side - ward _ ) . the mass - shell constraint ( [ 1.2 ] )
reads @xmath48 where @xmath49 denotes ( approximately ) the velocity of the particle pair transverse to the beam direction , and @xmath50 its longitudinal component .
the standard cartesian parametrisation @xcite of the correlation function is obtained by using ( [ 1.7 ] ) to eliminate @xmath51 from eq .
( [ 1.6 ] ) .
this determines 6 cartesian hbt radius parameters @xmath52 in terms of the variances @xmath53 of the emission function : @xmath54 \ , , \nonumber \\
r_{ij}^2({\bf k } ) & = & \langle ( \tilde{x}_i-{\beta}_i\tilde{t } ) ( \tilde{x}_j-{\beta}_j\tilde{t})\rangle \ , , \quad i , j = s , o , l \ , .
\label{1.8 } \end{aligned}\ ] ] for an azimuthally symmetric collision region , @xmath19 is symmetric with respect to @xmath55 @xcite .
then @xmath56 and @xcite @xmath57 \ , , \label{1.9}\ ] ] with @xcite [ 1.10 ] @xmath58 an alternative way of eliminating the redundant component of @xmath26 in ( [ 1.6 ] ) leads to the yano - koonin - podgoretskiparametrisation@xcite of @xmath19 , @xmath59 \ , .\ ] ] this is based on replacing in eq .
( [ 1.6 ] ) @xmath60 and @xmath61 in terms of @xmath62 , @xmath51 , and @xmath63 . here
, @xmath64 is a ( @xmath65-dependent ) 4-velocity with only a longitudinal spatial component : @xmath66 this parametrisation has the advantage that the ykp parameters @xmath67 , @xmath68 , and @xmath69 extracted from such a fit do not depend on the longitudinal velocity of the observer system in which the correlation function is measured ; they are invariant under longitudinal boosts .
the model - independent expressions for these ykp - parameters are most conveniently given in terms of the notational shorthands @xcite [ 1.13 ] @xmath70 where @xmath71 and @xmath72 for azimuthally symmetric sources such that @xmath73 .
in terms of these expressions one finds , i.e. as long as expression ( [ 1.14a ] ) for the velocity @xmath7 is defined .
the alternative forms of @xmath74 and @xmath75 given in @xcite are only valid if additionally @xmath76 . for a detailed discussion
see @xcite . ]
[ 1.14 ] @xmath77 for non - vanishing transverse pair momentum @xmath78 , the cartesian ( [ 1.8 ] ) and the ykp ( [ 1.11 ] ) parametrisations are equivalent and it is instructive to compare them .
the cartesian parameters can be calculated from the ykp ones via @xcite [ 1.15 ] @xmath79 later we will see that , for the explicit source models studied in this paper , in most cases the yano - koonin velocity @xmath7 is very close to the longitudinal pair velocity @xmath50 .
if this is true ( and one should be careful not to use the following expressions without first checking this ) eqs .
( [ 1.15]b , c ) simplify to [ 1.16 ] @xmath80 there is a slight subtlety for @xmath81 . in this limiting case
, the on - shell constraint ( [ 1.2 ] ) reads @xmath82 and can not be used to eliminate in ( [ 1.6 ] ) @xmath60 and @xmath61 in terms of @xmath83 , @xmath51 and @xmath84 . hence , strictly speaking , the ykp parametrisation exists only for @xmath85 . in practice , however , this does not lead to complications since the @xmath86 limit is well - defined for all ykp - parameters ( see sec .
[ sec4b ] ) .
the relations ( [ 1.15 ] ) provide a powerful consistency check on the experimental fitting procedure of the correlation radii .
they show that both parametrizations contain exactly the same spatio - temporal information
. however , certain space - time characteristics of the source are more directly accessible in a particular parametrization .
this is especially the case for the duration of the particle emission process , the lifetime " of the source . to see this we return to the expressions ( [ 1.10 ] ) for the cartesian hbt radii . these mix spatial and temporal information on the source in a non - trivial and frame - dependent way .
their interpretation in various reference systems was analysed analytically @xcite for a large class of ( azimuthally symmetric ) model emission functions and compared with the numerically calculated correlation function @xcite . for these models ,
the difference @xmath87 is dominated by the first term on the r.h.s . and thus provides access to the lifetime @xmath88 of the source @xcite .
note , however , that the definitions ( [ 1.10 ] ) and ( [ 1.17 ] ) are not lorentz invariant , and that the lifetime @xmath89 extracted from eq .
( [ 1.17 ] ) thus depends on the analysis frame . furthermore , in practice the term @xmath90 turned out to be much smaller than the terms @xmath91 and @xmath92 which are the leading contributions to @xmath93 and @xmath94 , respectively @xcite . as a consequence ,
excellent statistics of the data with very small statistical errors of @xmath1 and @xmath0 are required to extract the small contribution @xmath95 .
this makes the extraction of a small source lifetime from the standard fit difficult .
successful attempts have been reported from low - energy heavy - ion collisions ( using 2-proton correlations ) where the measured lifetimes are very long : @xmath96 fm/@xmath97 in ar+sc collisions at @xmath98 mev @xcite and @xmath99 fm/@xmath97 in xe+al collisions at @xmath100 mev @xcite ( the latter is the typical evaporation time of a compound nucleus ) .
two - pion correlations at ultra - relativistic energies ( @xmath101 gev ) so far failed to yield positive evidence for a non - vanishing emission duration @xcite , except for the heaviest collision system pb+pb @xcite , but even there the effective lifetime is only a few fm/@xmath97 . for the ykp parametrization
the situation is different .
the parameters @xmath6 , @xmath102 and @xmath103 are invariant under longitudinal boosts and thus independent of the analysis frame .
the key to their space - time interpretation is provided by the fourth fit parameter @xmath104 .
it specifies a ( pair momentum dependent ) longitudinal reference frame , the yano - koonin ( yk ) frame which defined by @xmath105 resp .
@xmath106 ( see ( [ 1.14a ] ) ) , in which the space - time variances ( [ 1.14 ] ) for the ykp radius parameters simplify considerably . especially , for certain classes of source models including the one studied below , the terms proportional to @xmath107 , @xmath108 , and @xmath109 are small @xcite . neglecting these terms
one obtains @xcite in the yk frame [ 1.19 ] @xmath110 in neighbouring frames , the yano - koonin velocity can be calculated with the same approximations as @xmath111 note that in the yk frame the temporal structure of the source enters only in the parameter @xmath6 .
its leading conribution is given by the time @xmath112 during which particles of momentum @xmath35 are emitted in this frame .
in the ykp parametrisation @xmath113 is fitted directly and not obtained as the difference of two large fit parameters as in the cartesian fit . in practice , however , the extraction of @xmath6 from the ykp fit is still not easy . from eq .
( [ 1.11 ] ) it follows that in the yk frame @xmath6 must be extracted from the @xmath51-dependence of the correlator . due to the mass - shell constraint ( [ 1.2 ] ) the interesting range of @xmath51 is limited , especially for low - momentum pairs , and the sensitivity of the fit function ( [ 1.11 ] ) to @xmath6 is weaker than to the two other radius parameters .
@xmath6 values thus tend to come out with larger experimental error bars .
for our numerical study we have taken the model from ref .
@xcite with the emission function @xmath114 \ ; \exp \left[- { ( \tau-\tau_0)^2 \over 2(\delta \tau)^2 } - { r^2 \over 2 r^2 } - { ( \eta- \eta_0)^2 \over 2 ( \delta \eta)^2 } \right ] . \label{2.1}\ ] ] the first term specifies the shape of the freeze - out hypersurface , the second one is a lorentz - covariant boltzmann factor encoding the assumption of local thermal equilibration superimposed by collective expansion , while the last one has a purely geometrical interpretation .
the space - time coordinates in longitudinal and temporal directions are parametrised by the space - time rapidity @xmath115 $ ] and the longitudinal proper time @xmath116 . in the transverse direction ,
the radius is @xmath117 .
accordingly , the measure reads @xmath118 . the time
component of the pair momentum is set to the on - shell value @xmath119 .
this approximation was studied in detail in ref.@xcite where it was shown to be acceptable .
thus the pair momentum @xmath65 can be parametrised using the momentum rapidity @xmath120 $ ] and the transverse mass @xmath121 , @xmath122 we implement longitudinal and azimuthally symmetric transverse expansion of the source by parametrising the flow velocity in the form @xmath123 for the longitudinal flow rapidity we take @xmath124 independent of @xmath125 , i.e. we assume a bjorken scaling profile @xcite @xmath126 in the longitudinal direction .
for the transverse flow rapidity we take a linear profile of strength @xmath127 : @xmath128 the scalar product in the exponent of the boltzmann factor generates the @xmath16-@xmath65-correlation in our source .
it can then be written as @xmath129 please note that for non - zero transverse momentum @xmath130 , a finite transverse flow breaks the azimuthal symmetry of the emission function via the second term in ( [ 2.5 ] ) . for @xmath131 ,
the emission function is azimuthally symmetric for all @xmath78 .
also , it then has no explicit @xmath130-dependence , and @xmath15 is the only relevant scale . as will be discussed in sec .
[ sec4d ] this gives rise to perfect @xmath15-scaling of the ykp radius parameters in the absence of transverse flow , which is again broken for non - zero transverse flow @xcite .
besides @xmath127 , the model parameters are the freeze - out temperature @xmath132 , the transverse geometric ( gaussian ) radius @xmath8 , the average freeze - out proper time @xmath133 as well as the mean proper emission duration @xmath134 , the centre of the source rapidity distribution @xmath135 , and the ( gaussian ) width of the space - time rapidity profile @xmath136 . a rough spatial picture of the source at various fixed coordinate times can be gleaned from the figs .
1 and 2 in ref . @xcite ( although their source has sharp edges whereas ours is smoothed by gaussian profiles ) and from figs .
1 and 2 in ref . @xcite .
note that our parametrization of @xmath18 does not allow for the case of opaque sources where the emission is surface dominated @xcite . in this case ,
the contribution of @xmath137 may become negative and large @xcite which might alter the argument following ( [ 1.17 ] ) .
a detailed study of such opaque sources is presented in ref .
@xcite . on the basis of a comparison with the preliminary data of ref .
@xcite the authors of that study conclude that the source created in pb+pb collisions at the cern sps is rather transparent and not opaque ; its qualitative features are well described by the model presented here @xcite .
we did our calculations for pions ( @xmath138 mev/@xmath139 ) and kaons ( @xmath140 mev/@xmath139 ) .
resonance decays @xcite are not discussed here but is deferred to a separate publication @xcite .
the calculations presented here are meant to illustrate general properties of the ykp parameters ; no attempts to compare with data will be made .
in this section we present a quantitative study of the ykp fit - parameters . since the ykp - parameter @xmath4 is identical to the `` side '' radius of the cartesian parametrisation , @xmath141 , its interpretation is obvious and independent of the longitudinal velocity of the reference frame .
hence we focus subsequently on the remaining three parameters @xmath142 , @xmath143 , and @xmath7 .
unless stated otherwise , the numerical calculations below are done with the set of source parameters @xmath144 mev , @xmath145 fm , @xmath146 , @xmath147 fm/@xmath97 , @xmath148 fm/@xmath97 . according to eq ( [ 1.12 ] ) , the ykp fit parameter @xmath7 is a longitudinal velocity . in this subsection
we give a detailed discussion of the reference frame specified by @xmath7 and establish its relation to several other commonly used reference frames .
their definitions are : * * cms * : the centre of mass frame of the fireball , specified by @xmath149 . * * lcms * ( longitudinally comoving system @xcite ) : a pion ( kaon ) pair - dependent frame , specified by @xmath150 . in this frame
, only the transverse velocity component of the pion ( kaon ) pair is non - vanishing . * * lsps * ( longitudinal saddle - point system @xcite ) : the longitudinally moving rest frame of the point of maximal emissivity for a given pair momentum . in general
, the velocity of this frame depends on the momentum of the emitted particle pair . for symmetric sources the point of maximal emissivity ( `` saddle point '' ) coincides with the `` source centre '' @xmath34 defined in ( [ 1.4 ] ) . in this approximation , for a source like ( [ 2.1 ] ) , the lsps velocity is given by the longitudinal component of @xmath151 .
* * yk * ( yano - koonin frame @xcite ) : the frame for which the ykp velocity parameter vanishes , @xmath152 .
again , this frame is in general pair momentum dependent .
these four frames are quite different in nature . the velocities ( or rapidities ) of the cms and lcms frames can be easily determined experimentally , the first from the peak in the single particle rapidity distribution , the second from the longitudinal momentum of the measured pion pair . however , the velocity of the lsps is in a sense more interesting : from its definition it is directly related to the longitudinal expansion velocity of the source .
in fact , longitudinal expansion of the source leads to a characteristic dependence of the lsps velocity @xmath153 on the pair rapidity .
this is most easily seen by considering two extreme fireball models : \(1 ) if the source does not expand , all its elements move with the same velocity ( rapidity ) , namely that of the cms . then there is no kinematic difference between different parts of the fireball , and the saddle point is @xmath65-independent and given by the peak of the space - time distribution of the source .
thus the rapidity of the lsps , defined as @xmath154 $ ] , is _ independent _ of the pair rapidity @xmath155 and identical to the rapidity of the cms .
this behaviour of @xmath156 arises automatically if the source has no @xmath157 correlations , i.e. if the emission function factorizes , @xmath158
. then all space - time characteristics of the source , including the lsps velocity , are determined by @xmath159 alone and do not depend on @xmath65 .
hence @xmath156 is independent of @xmath155 .
factorization is , however , not necessary for this behaviour : non - vanishing @xmath160 correlations generated , e.g. , by temperature gradients do nt induce a @xmath155-dependence of @xmath156 as long as the source does not expand longitudinally .
\(2 ) if we set in eq . ( [ 2.1 ] ) @xmath161 , we recover the bjorken model @xcite for a longitudinally infinite source with boost - invariant longitudinal expansion .
the only @xmath162-dependence then comes from the thermal boltzmann factor , and the longitudinal saddle point obviously lies at @xmath163 .
since for the bjorken scaling profile @xmath162 coincides with the longitudinal fluid rapidity , this implies that in this case the rapidity of the lsps is _ identical _ with the pair rapidity , @xmath164 , i.e. the lsps coincides with the lcms .
it is obvious from this discussion that knowledge of the function @xmath165 would allow to distinguish between these two scenarios : a non - expanding source would yield @xmath166 while a source with boost - invariant longitudinal expansion gives @xmath167 .
realistic models are expected to lie in between these two extremes .
unfortunately , the lsps - velocity @xmath168 can not be measured .
this is clear from its definition as the longitudinal flow velocity evaluated at the point of maximum emissivity .
as discussed above , this point is approximately given by the source centre @xmath34 which itself is unmeasurable : it drops out @xcite from both the single and two - particle spectra which are invariant under a translation of the source centre ( even if @xmath35-dependent ! ) .
the only velocity we can measure , namely from an ykp fit of the two - particle correlator , is the yano - koonin ( yk ) velocity @xmath104 resp .
the associated rapidity @xmath169 $ ] .
it is therefore very gratifying to know that for sufficiently rapidly expanding systems the two velocities are always closely related ( although in general not identical ) . in particular
, we will show below that for the class of models ( [ 2.1 ] ) the function @xmath170 shares with @xmath165 the feature that it provides a direct signature for longitudinal expansion .
the close relationship between the ( measurable ) yk - velocity and the ( theoretical ) lsps - velocity for certain source models has been known for years . in ref .
@xcite a `` symmetric frame '' was introduced as the reference frame , in which the production process is symmetric in the beam direction . in this frame , the longitudinal extension and the lifetime of the source reach their extremal values @xcite . for moving , but non - expanding azimuthally symmetric sources podgoretski found in this way the parametrisation ( [ 1.11 ] ) , with @xmath35-independent parameters @xmath103 , @xmath102 , and @xmath6 , and with what we call the yano - koonin velocity @xmath7 being identical to the velocity of his `` symmetric frame '' . in this case
the yk - system also coincides with the rest frame of the source as a whole ( cms ) as well as with the lsps . that the coincidence between the yk and lsps systems is more generally valid for sources which are symmetric around their saddle point
@xmath34 has been observed in refs .
it thus holds for any emission function in the gaussian saddle - point approximation , due to the symmetry of the latter .
as the following paragraph will show , differences between the yk - velocity and the velocity of the lsps are only due to asymmetries of the source around its saddle point .
although such asymmetries usually exist for collectively expanding sources with finite geometric extension , they are generally small and can be treated perturbatively .
therefore , the yk - frame and the lsps - frame are usually very close to each other , @xmath171 . from the examples above
it is clear , however , that the same is not true for the lcms ( i.e. the longitudinal rest frame defined by the pair rapidity @xmath155 ) , and that generally @xmath172 .
let us now discuss the difference @xmath173 in more detail .
if it is small , so is @xmath30 when evaluated in the lsps - frame . from eqs .
( [ 1.18 ] ) and ( [ 1.13c ] ) we see that then in the lsps frame @xmath7 is given by @xmath174 where we expanded in first order of @xmath175 , @xmath176 , @xmath109 . the smallness of these terms was argued in @xcite and will be checked in the following subsection . the first and second term of ( [ 3.1 ] )
reflect the longitudinal asymmetry of the source around the saddle point ; they vanish for sources with longitudinal reflection symmetry @xmath177 . similarly ,
the second and third term vanish unless the reflection symmetry @xmath178 around the saddle point in the out "- direction is broken ( e.g. by transverse flow ) .
( asymmetry in @xmath179 is needed for the first and third terms to become non - zero . )
a non - zero value of the last term , finally , indicates the breaking of the out"-side " rotation symmetry ; this can again be caused by transverse source gradients as e.g. transverse flow ( see eq .
( [ 2.5 ] ) ) .
we conclude that the difference @xmath173 is entirely due to asymmetries of the source around the saddle point .
furthermore , we will show in sec .
[ sec4b ] that the last three terms in ( [ 3.1 ] ) are small for small values of the transverse flow rapidity @xmath127 and/or small values of @xmath130 .
( they vanish for @xmath180 . ) in these limits the difference between @xmath7 and @xmath153 is dominated by the longitudinal source asymmetry , and @xmath7 is very accurately given by the ( leading ) first term in ( [ 3.1 ] ) , see eq .
( [ 1.18 ] ) .
note that in our model the breaking of the longitudinal reflection symmetry is due to the non - symmetric rapidity profile of the emission function for @xmath181 . for a quantitative discussion we plot in fig .
[ f1 ] the yano - koonin rapidity @xmath182 as well as the difference @xmath183 as functions of @xmath15 and @xmath155 .
all rapidities are given relative to the cms .
one sees that for large values of @xmath130 the agreement of @xmath182 with @xmath156 is almost perfect .
also , in this limit both rapidities approach the value of @xmath155 , i.e. the yk and lsps systems coincide with the lcms .
the reason for this is that for large @xmath130 the boltzmann term in the emission function ( [ 2.1 ] ) becomes sharply peaked around the point @xmath16 where the fluid velocity agrees with the pair velocity ; the geometric terms in the emission function are much smoother and can be neglected .
the relevant term is the first term in eq .
( [ 2.5 ] ) , and thus the relevant variable is the transverse mass @xmath15 .
hence this kinematic region starts for kaons at smaller values of @xmath130 than for pions ( see fig .
. for small values of @xmath15 , the difference @xmath184 increases , and both begin to lag behind the lcms rapidity @xmath155 .
still , the yk frame is closer to the lcms than is the lsps .
the fact that the yk and lsps systems track each other so closely implies that the linear rise of the yk rapidity with the pair rapidity @xmath155 reflects nothing but a similar rise of the lsps rapidity with @xmath155 . as argued above
, the latter is a direct indication for longitudinal expansion of the source .
however , it should be noted that this expansion need not necessarily be of hydrodynamic nature .
the same feature would be generated by a source consisting of free - streaming pions and resonances which were created at an initial proper time @xmath185 through a boost - invariant production mechanism @xcite , suffering no further re - scattering .
it is easily seen that the strict correlations between coordinates and momenta in a free - streaming gas again lead to a linear dependence of the `` source rapidity '' @xmath182 on the pair rapidity , with @xmath15-independent unit slope .
in fact , it is possible to simulate this situation with the emission function ( [ 2.1],[2.3 ] ) by setting @xmath132 and @xmath127 to zero , i.e. by eliminating the thermal smearing of the momenta and the transverse collective flow .
( of course , this would also result in vanishing ykp radius parameters , because pions of fixed rapidity @xmath155 can come from only a single point in the source . ) a linear rise of @xmath182 with @xmath155 ( with approximately unit slope ) was recently observed by the gibs collaboration in dubna @xcite by analysing pion correlations from mg+mg collisions at 4.4 @xmath186 gev/@xmath97 .
they interpreted their result as evidence for rapid longitudinal expansion of the source .
the data were averaged over the transverse pair momentum @xmath130 .
the data sample was taken with a `` central '' trigger , but since @xmath187 mg is a rather small nucleus with a large surface to volume ratio it is not clear what fraction of the participating nucleons were stopped to become part of a thermalized fireball .
the strong linear increase of @xmath182 with @xmath155 could thus also reflect to some part the free - streaming expansion of the pion sources created in the periphery of the nuclear reaction .
preliminary results of the na49 collaboration at cern for pb+pb collisions at 158 @xmath186 gev/@xmath97 also show a rise of @xmath182 with the pair rapidity @xmath155 @xcite . here
, however , the @xmath155-dependence of @xmath182 does not appear to be quite linear , and its slope is less then 1 . with all due caution with regard to the still preliminary nature of these results
, this may indicate genuine hydrodynamic longitudinal expansion at a somewhat slower rate than resulting from our longitudinal scaling profile .
na49 have also looked separately at a subsample of pairs with @xmath188 mev/@xmath97 , showing that for them the yk rapidity appears to rise more rapidly with @xmath155 than in the @xmath130-averaged sample , in agreement with theoretical expectations for a thermalized source ( see .
[ f1]a ) .
the @xmath15-dependence of @xmath182 can be understood along the same lines .
[ f1]b shows that @xmath170 approximates @xmath155 better with increasing transverse mass . in the limit @xmath189 the bosons
can be emitted only from the source element which moves exactly with the same rapidity , hence @xmath190 . in the opposite limit @xmath191 the boltzmann factor in ( [ 2.1 ] ) becomes a smooth function of @xmath16 , and the emission function is dominated by the gaussian geometric terms . in this limit one
thus expects the yk rapidity to approach the value @xmath192 .
the numerical results of fig .
[ f1]b show that the value of the pion mass is already large enough for the boltzmann part of the emission function to become important . as a result
the @xmath15-dependence of @xmath182 is weak in the entire range which can be covered by pions , and even weaker for kaons . in this subsection
, we study quantitatively the correction terms of eqs .
( [ 1.19 ] ) and ( [ 3.1 ] ) which may compromise the approximation ( [ 1.18 ] ) for @xmath7 and the simple interpretation of @xmath102 and @xmath6 as longitudinal and temporal widths of the emission function .
since the geometric interpretation of @xmath102 and @xmath6 refers to the yk frame , our analysis will also be done in this frame .
let us first focus on the central rapidity region @xmath193 .
then @xmath194 , i.e. , the four reference frames listed in sec .
[ sec4a ] coincide .
furthermore , the source is symmetric in the longitudinal direction and thus @xmath195 .
the only non - vanishing corrections thus arise from the terms @xmath196 and @xmath197 . in fig .
[ f2 ] they are plotted as a function of @xmath130 for different values of the scaling parameter @xmath198 for the transverse flow . without transverse flow ( i.e. for @xmath199 ) the source is azimuthally symmetric ( see eq .
( [ 2.5 ] ) ) which implies @xmath200 . also , the source is reflection symmetric in the out - direction , @xmath201 . for non - zero
transverse flow the correction terms are generally non - zero , and they grow with increasing @xmath127 . note that , for fixed @xmath127 and @xmath130 , the correction terms are considerably smaller for kaons than for pions .
this can be understood as follows : as discussed after eq .
( [ 2.5 ] ) , for @xmath131 the source depends only on @xmath15 , and thus even for non - zero transverse flow everything to zeroth order still scales with @xmath15 .
we will show below that for expanding sources the regions of homogeneity , which effectively contribute to the correlation function , are generically decreasing functions of @xmath15 .
this is thus also true for the correction terms .
since at fixed @xmath130 the value of @xmath15 is larger for kaons than for pions , the corresponding correction terms are smaller in absolute terms ( although not necessarily relative to the leading contributions ) . the @xmath78-dependence of the correction terms at non - zero transverse flow @xmath202 can also be easily understood .
the rise of @xmath203 for increasing transverse momentum ( fig .
[ f2]a ) is due to the azimuthal symmetry breaking by the second term of eq .
( [ 2.5 ] ) which increases both with @xmath78 and @xmath127 .
it agrees with the findings of ref .
@xcite but , as pointed out in @xcite , contradicts the behaviour seen by pratt in the first of refs .
@xcite for an infinitesimally thin spherically expanding shell where @xmath109 decreases with increasing @xmath130 .
a similar behaviour is seen @xcite in hydrodynamical simulations where freeze - out occurs along an infinitesimally thin freeze - out hypersurface ; there also @xmath204 first decreases very rapidly with increasing @xmath130 , then saturates and slightly increases again without , however , ever turning positive .
the strong decrease of @xmath204 with @xmath130 appears to be an artefact of the idealization of an infinitesimally thin expanding shell ; in @xcite it was shown to be much less visible for opaque sources with a finite thickness of the emitting surface layer , returning to the here observed rise already for a rather modest surface thickness .
different from @xmath109 , the variance @xmath196 reaches an extremum and then decreases again for very large @xmath130 ( fig .
this results from an interplay between the increasing breaking of the @xmath205 reflection symmetry , which tends to increase the value for @xmath196 , and a decreasing homogeneity length in space - time rapidity @xmath162 which affects the @xmath206 part of this variance . in the model - independent expressions ( [ 1.19 ] ) and ( [ 3.1 ] ) for the ykp parameters , the correction terms discussed above are divided by @xmath207 and @xmath208 , respectively . from appendix [ app ] we know that the ratios remain finite in the limit @xmath209 .
still , the corrections to the ykp parameters could become sizeable , depending on how slowly the variances @xmath210 and @xmath109 vanish in this limit . in fig .
[ f2]b , d we show that the correction terms actually remain small even after dividing them by the appropriate powers of @xmath211 .
thus , at least at mid - rapidity , the leading order approximations ( [ 1.19]b , c ) are seen to be generally very good .
the largest correction comes from the difference @xmath212 in eq .
( [ 1.19c ] ) .
a more detailed discussion of its effects on @xmath6 will follow in the next subsection .
we now proceed to a discussion of the correction terms for @xmath181 .
we define @xmath213 as the rapidity of the pair in the cms . in fig .
[ f3 ] we compare the correction terms for @xmath214 to those for @xmath215 .
of course , at @xmath214 the yk frame no longer coincides with the cms , see sec .
[ sec4a ] .
since the transverse variances are not affected by longitudinal boosts nor do they depend on @xmath155 ( see appendix [ app ] ) , @xmath216 does not change with the pair rapidity . however , at a given value of @xmath130 the transverse pair velocity @xmath217 becomes smaller , since @xmath218 increases with the pair rapidity @xmath219 .
the correction term @xmath220 thus increases with @xmath219 , especially at low @xmath130 .
a similar effect is seen in the plots for @xmath221 .
while @xmath196 decreases with increasing @xmath219 , the ratio with @xmath211 actually increases by about a factor 2 at small @xmath130 . since in the forward rapidity region
the source becomes non - symmetric under reflection @xmath222 , @xmath223 and @xmath224 are non - zero , but small .
we conclude this subsection with a discussion of the sensitivity of the correction terms to the longitudinal extension of the source , which is parametrised by @xmath136 .
we again put @xmath225 , so that @xmath226 . from the explicit expressions given in appendix [ app ]
it is clear that the transverse variances @xmath109 are independent of @xmath227 .
therefore only the size of @xmath196 changes . in fig .
[ f4 ] we show its @xmath130-dependence for different values of @xmath136 . due to our assumption of freeze - out along a hyperbola of constant proper time @xmath133 ( smeared by an amount @xmath228 ) , an increase of @xmath136 causes a larger effective extension of the source both in the @xmath43 and in the time direction .
this becomes especially important at small @xmath130 where the geometrical factors in the emission function ( [ 2.1 ] ) play an important role .
this explains the relatively large @xmath227-dependence of @xmath108 at small @xmath130 and the weaker dependence at large @xmath130 where the source distribution is dominated by the boltzmann term which does not depend on @xmath136 . in this subsection
we combine the results from the previous subsection for the correction terms with the leading contributions in order to arrive at a quantitative understanding of the longitudinal and temporal ykp radius parameters @xmath102 and @xmath6 , and in particular of their dependence on the pair momentum @xmath35 . in fig .
[ f5 ] we show @xmath6 and @xmath5 together with their approximations @xmath229 , @xmath230 , for pion pairs with rapidity @xmath215 and @xmath214 , respectively , as functions of @xmath78 . for vanishing transverse flow both approximations
are seen to be exact , in agreement with the discussion from the previous subsection . for non - zero
transverse flow the approximation @xmath231 remains exact for pairs with @xmath225 .
the reason is that for such pairs the yk rapidity relative to the cms is zero , and thus the longitudinal velocity @xmath50 ( which multiplies the correction terms in ( [ 1.19b ] ) ) of the pair in the yk frame vanishes . from fig .
[ f5]b one sees , however , that also for forward rapidity pairs at @xmath232 the correction terms stay below 10% at all values of @xmath78 .
the situation is not quite as good for @xmath6 .
here one sees apparently strong differences between @xmath6 and @xmath233 as soon as the transverse flow is switched on . from fig .
[ f2 ] it is clear that the ( in our case positive ) correction term @xmath234 is the culprit and dominates the difference . for a transverse flow of @xmath235 as shown in fig .
[ f5]a this term becomes ( in the experimentally accessible @xmath130 range ) larger than 1 ( fm/@xmath97)@xmath236 and thus comparable to the leading term @xmath237 .
however , the numerical results shown in this figure actually correspond to a rather extreme situation .
first , the assumed transverse flow rapidity @xmath238 is large ; the heavy ion data at ags and cern energies for si- and s - induced reactions seem to require smaller values @xcite .
second , the difference between @xmath6 and @xmath233 is small at low transverse momenta and becomes large only at large @xmath130 ; in that range the leading term @xmath233 is essentially given by the source parameter @xmath228 ( see discussion below ) which in pb+pb and au+au collisions @xcite ( where @xmath127 may be larger than for the smaller systems analyzed so far ) seems to be bigger than the 1 fm/@xmath97 assumed here tends to be negative @xcite . in this case , the deviation of @xmath6 from @xmath239 will have the opposite sign @xcite , and @xmath75 usually even turns negative for small values of @xmath130 . for a detailed study
we refer to @xcite . ] .
[ f5]a shows that the effective source lifetime @xmath240 is a strong function of the pair momentum @xmath35 : it is largest at small rapidity @xmath219 and transverse momentum @xmath130 and decreases with increasing @xmath219 and/or @xmath130 .
its asymptotic value for large @xmath35 in the cms is , not unexpectedly , given by the variance ( [ a11 ] ) of the proper time distribution of our source ( [ 2.1 ] ) .
but why is it larger for pairs with smaller momenta @xmath35 ? from fig .
[ f5]b it is clear that the longitudinal region of homogeneity @xmath102 is a decreasing function of the pair momentum @xmath35 .
the reason for this is the same as the similar decrease of @xmath2 in the cartesian fit and well understood @xcite as a consequence of the strong longitudinal expansion of the source .
this expansion introduces a longitudinal velocity gradient , and the longitudinal length of homogeneity is given by the inverse of this gradient multiplied by a `` thermal smearing factor '' @xcite .
the latter reflects the statistical distribution of the particle momenta around the local source fluid velocity , and for a thermal distribution the spatial region over which this thermal smearing is effective decreases with increasing pair momentum .
this causes the shrinking of the longitudinal homogeneity length with @xmath35 . since for different pair momenta @xmath6 measures the source lifetime in different yk reference frames , the freeze - out `` hypersurface ''
will in general appear to have different shapes for pairs with different momenta . only in our model , where freeze - out occurs at fixed proper time @xmath133 ( up to a gaussian smearing with width @xmath228 ) , is it frame - independent .
it is thus generally unavoidable ( and here , of course , true in any frame ) that freeze - out at different points @xmath43 in the source will occur at different times @xmath179 in the yk frame .
since a @xmath43-region of size @xmath102 contributes to the correlation function , @xmath102 determines how large a domain of this freeze - out surface ( and thus how large an interval of freeze - out times in the yk frame ) is sampled by the correlator .
this interval of freeze - out times combines with the intrinsic gaussian width @xmath228 to yield the total effective duration of particle emission .
it will be largest at small pair momenta where the homogeneity region @xmath102 is biggest , and will reduce to just the variance of the gaussian proper time distribution at large pair momenta where the longitudinal ( and transverse ) homogeneity regions shrink to zero .
another interesting feature of fig .
[ f5 ] is that at large @xmath130 both @xmath102 and @xmath6 are independent of the pair rapidity @xmath155 .
this is a consequence of our boost - invariant longitudinal velocity profile and need not remain true for systems with different longitudinal expansion . as argued before , at large @xmath15 the space - time shape of the source is dominated by the boltzmann term and becomes insensitive to the gaussian geometric factors .
the hbt radii thus only see the local velocity gradients which in our case are invariant under longitudinal boosts . at large @xmath15 pion pairs with different rapidities
@xmath155 thus all see the same local source structure , and the ykp radii become @xmath155-independent .
we close this subsection with a discussion of the dependence of @xmath102 and @xmath6 on the other source parameters . since the rapidity dependence does not change qualitatively from what has already been discussed , we concentrate on zero rapidity pion pairs , @xmath225 . for the transverse flow
we choose a non - zero , but moderate value of @xmath241 . in fig .
[ f6 ] we show the dependence on the longitudinal size of the source which is parametrised by @xmath227 .
one sees that at large transverse momenta neither @xmath102 nor @xmath6 and @xmath237 are affected by the width @xmath136 of the gaussian geometric factor , in line with the arguments above . at small transverse momenta ,
both radii increase monotonically with @xmath227 .
this means that , at low @xmath130 , @xmath102 becomes sensitive to the global longitudinal geometry of the source and no longer only reflects the local longitudinal velocity gradients .
[ f6]a is very interesting : for small @xmath227 the longitudinal length of homogeneity is limited by the longitudinal geometry , and @xmath6 never has a chance to probe a large region of the proper time freeze - out surface .
hence @xmath89 is limited to the variance of the proper time distribution @xmath242 in the source ( see ( [ a2],[a11 ] ) ) , independent of @xmath130 .
as @xmath136 increases , at small @xmath130 the longitudinal length of homogeneity increases too , and @xmath6 receives an additional contribution from the time variation ( in the yk rest frame ) along the freeze - out surface inside a longitudinal region of size @xmath102 .
thus the rise of the effective source lifetime @xmath89 at small @xmath130 is an indirect measure for the longitudinal geometric size of the source .
unfortunately , the detailed quantitative dependence is model - dependent .
[ f7 ] shows what happens when the width of the proper time distribution @xmath242 ( eq . ( [ a2 ] ) ) is changed .
increasing @xmath228 by 1 fm/@xmath97 , @xmath6 also increases by about 1 fm/@xmath97 ( slightly less at large @xmath130 : according to eq .
( [ a11 ] ) it increases from 0.89 fm/@xmath97 for @xmath243 fm/@xmath97 to 1.49 fm/@xmath97 for @xmath244 fm/@xmath97 . ] ) , while @xmath102 increases more at small @xmath130 and less at large @xmath130 .
the increase of @xmath102 is due to the decrease of the longitudinal velocity gradient ( which for a boost - invariant profile is @xmath245 ) with @xmath125 . as the time distribution @xmath242 becomes wider , larger proper times are probed by the emitted pions resulting in larger longitudinal homogeneity regions . changing the average freeze - out time @xmath133 rather than its spread @xmath228 has qualitatively similar consequences ( see fig .
[ f8 ] ) , only that @xmath6 at sufficiently large @xmath130 again reduces to the same small variance of the time distribution @xmath242 .
note that , at small @xmath130 , @xmath6 increases both with increasing @xmath136 and increasing @xmath228 ; this supports our claim that it is `` sensitive '' to the total longitudinal extension of the source @xmath246 .
however , the relation is not linear ( in particular in our numerical results a doubling of @xmath227 is seen to have less effect than a doubling of @xmath133 ) , making it hard to use in practice .
it is obvious that the longitudinal velocity gradient ( which decreases by a factor 2 when @xmath133 is doubled ) has a stronger influence on @xmath6 and @xmath102 than the geometrical width in @xmath162 . in this subsection
we compare pion and kaon correlation functions .
we discuss the @xmath15-scaling of the ykp radius parameters and its breaking by transverse collective flow . at the end of the subsection
we formulate a program how to extract transverse flow from the @xmath15-dependence of the ykp radius parameters . in fig .
[ f9 ] we compare , for central rapidity pairs @xmath225 , the three ykp radius parameters for pion and kaon pairs , as functions of @xmath130 . the left column shows a source without transverse expansion , in the right column the transverse flow rapidity was set to @xmath238 .
the onset of transverse flow has two qualitative effects : _
( i ) _ the transverse radius acquires a @xmath130-dependence @xcite , and _
( ii ) _ @xmath6 and @xmath233 begin to deviate from each other , as discussed in the previous subsection . ( the equality @xmath247 remains exact because we are studying pion pairs at @xmath225 . ) the effects of flow on @xmath102 and @xmath233 are seen to be weak , for both pions and kaons .
note also that at small @xmath130 the kaon radii are generically smaller than the pion radii , with or without transverse flow .
this is also seen in experiment @xcite .
however , except for the change in the rest mass we have changed no parameters in the emission function , so the difference must be entirely kinematic . indeed , it just reflects the fact that for thermalized sources like ( [ 2.1 ] ) the leading dependence on the particle rest mass is through the variable @xmath248 . as discussed after eq .
( [ 2.5 ] ) the source ( [ 2.1 ] ) depends _ only _ on @xmath15 if it does not expand transversally ( @xmath249 ) . in this case
the correction terms in eqs .
( [ 1.19]b , c ) vanish exactly , and @xmath250 , @xmath251 , and @xmath252 are also functions of @xmath15 only .
this is shown in the left column of fig .
[ f10 ] where the ykp radii for a source without transverse flow are plotted as functions of @xmath15 and seen to exactly coincide for pions and kaons in the common @xmath15-range .
since in the absence of transverse flow the boltzmann factor in ( [ 2.1 ] ) has no transverse gradients ( we have assumed a constant temperature @xmath132 ) , the transverse radius @xmath103 is @xmath15 independent and equal to the transverse geometric ( gaussian ) radius @xmath8 .
it was pointed out in refs .
@xcite that transverse temperature gradients can also cause an @xmath15-dependence of the transverse radius @xmath103 ; but since the source remains in this case a function of @xmath15 only , the @xmath15-scaling of the ykp radii persists ; it can only be broken by transverse flow .
the breaking of the @xmath15-scaling by transverse flow is shown in the right column of fig .
[ f10 ] , for @xmath238 .
it has two origins : the emission function itself is no longer a function of @xmath15 only ( see ( [ 2.5 ] ) ) , and the now non - vanishing correction terms in ( [ 1.19]b , c ) depend on @xmath253 and thus on both @xmath15 _ and _ the rest mass @xmath254 .
it is obvious that the scaling violations induced by the pion - kaon mass difference are weak and require very accurate measurements .
furthermore , one may be worried that resonance decay contributions to the correlation radii @xcite ( which we havent discussed here ) lead also to a breaking of the @xmath15 scaling , because they affect pions more than kaons , and this may make it difficult to isolate the transverse flow effects .
we refer to the detailed discussion of resonance decays in the context of the model source ( [ 2.1 ] ) in ref .
that study shows , however , that their influence on the @xmath15-dependence of the transverse radius parameter @xmath103 is weak @xcite .
furthermore , resonances tend to increase all three hbt radii ( in particular the effective lifetime @xmath6 ) , while the @xmath15-scaling violations from transverse flow have the opposite sign for @xmath102 and @xmath255 .
detailed dynamical studies of the freeze - out process have shown that the transverse gradients of the temperature across the freeze - out surface tend to be small @xcite .
so the experimentally observed @xmath15-dependence of the transverse radius @xcite is presumably due to transverse flow @xcite .
it was shown in ref .
@xcite that the strength of the @xmath15-dependence of @xmath103 increases monotonously with the strength @xmath127 of the transverse expansion .
alber @xcite has suggested to quantify the strength of collective flow by fitting the hbt radii to a power law in @xmath15 , @xmath256 and using the magnitude of the extracted ( negative ) power as a flow measure .
he found @xmath257 for @xmath2 and smaller values for @xmath258 , with a tendency to increase for larger collision systems @xcite .
he interpreted this as a signature for strong longitudinal and weaker transverse expansion , the latter becoming more important for larger systems . in fig .
[ f11 ] we study the possible conclusions from such an exercise when applied to the results from our model .
the left column shows double logarithmic plots for @xmath103 and @xmath102 as functions of @xmath15 .
obviously the assumption of a power law dependence is well justified for @xmath102 but somewhat marginal for @xmath103 .
@xmath259 can not be approximated by a power law at all . in the right column
we show the extracted powers as a function of @xmath127 , the scale parameter for the transverse flow .
since @xmath103 is not well represented by a power law , the extracted slope depends somewhat on the fit region , as indicated for the two sets of curves in fig .
altogether it is , however , clear that for pions the power @xmath258 increases approximately linearly with @xmath127 and for kaons somewhat more strongly .
but even for large transverse flow rapidities @xmath260 the power remains below 0.2 .
in contrast , the corresponding power @xmath261 in a fit @xmath262 is already 0.55 in the absence of transverse flow , reflecting the strong boost - invariant longitudinal expansion .
( note that the decrease of @xmath102 with increasing @xmath15 is faster than the @xmath263-law suggested in ref .
@xcite see also ref .
@xcite for a discussion of this point . ) as the transverse flow is switched on , @xmath261 changes much more weakly than @xmath258 , showing that @xmath102 is mostly sensitive to the longitudinal flow while @xmath103 is only affected by transverse expansion .
again , kaons are affected by the transverse flow more strongly than pions .
we have presented a numerical study of the yano - koonin - podgoretski fit parameters for the two - particle correlation function .
our starting point were the recently derived model - independent expressions for the ykp parameters in terms of second order space - time variances of the source emission function .
these expressions allow for an easy evaluation of the ykp parameters as functions of the pair momentum @xmath35 and for detailed parameter studies .
we exploited them for a class of hydrodynamic models describing locally thermalized and collectively expanding sources , and we studied the dependence of the ykp parameters on the longitudinal and temporal extension of the source as well as on its longitudinal and transverse expansion velocity . in the context of such models it has been argued previously that in a certain approximation ( which becomes exact in the absence of transverse @xmath16-@xmath17-correlations of the source as , e.g. , induced by transverse expansion flow ) the ykp parametrisation achieves a perfect factorisation of the longitudinal and transverse spatial and the temporal extensions ( `` lengths of homogeneity '' ) of the source , in the comoving frame of the emitting fluid element .
the velocity of this emitting fluid element is then given by the fourth ykp parameter ( the yk velocity ) .
here we have shown numerically that , within these models , these features are preserved even in the presence of transverse flow .
the transverse radius parameter @xmath264 gives the effective transverse size of the source , the longitudinal radius parameter @xmath265 its effective longitudinal size , both in the local rest frame of the emitter as seen by pairs with momentum @xmath35 .
also , the ykp parameter @xmath266 provides a direct estimate of the effective emission duration for particles with momentum @xmath35 ; this estimate is quite accurate as long as the average transverse flow rapidity remains below 0.5 and the `` lifetime parameter '' @xmath228 is not too small ( @xmath267 fm/@xmath97 ) .
we also showed analytically and numerically that the yk velocity obtained from the ykp fit is indeed approximately equal to the longitudinal velocity of the emitting fluid element ( the lsps velocity ) , and that the small differences between these two velocities can be understood quantitatively in terms of asymmetries of the source around the point of maximum emissivity ( its `` saddle point '' ) .
this enables us to interpret the rise of the yk rapidity with the pair rapidity @xmath155 as a direct consequence of the longitudinal expansion of our source , and the @xmath15-dependence of the slope of the function @xmath170 as a signature for the thermal smearing of the particle momenta in the fluid rest frame . in ref .
@xcite we further showed that the slope of the function @xmath170 is nearly independent of the transverse flow rapidity @xmath127 .
thus the yk rapidity is manifestly dominated by the _ longitudinal _ expansion and hardly affected by the transverse expansion at all .
the latter causes , on the other hand , an @xmath15-dependence of the transverse radius parameter @xmath103 , which in turn is completely unaffected by the longitudinal expansion . in the present model study , the ykp parametrisation thus not only leads to a factorisation of the ( transverse and longitudinal ) spatial and temporal aspects of the source _ geometry _ , but it also cleanly separates its transverse and longitudinal _
dynamics_. the sensitivity of the ykp radius parameters to the transverse expansion of the source was investigated quantitatively in sec .
[ sec4d ] .
while the longitudinal radius parameter @xmath102 is affected very little by the transverse flow ( its strong @xmath15-dependence arises from the strong longitudinal flow ) , the transverse radius shows a considerable dependence on @xmath127 ( but none to the longitudinal flow ) .
furthermore , transverse flow breaks the exact @xmath15-scaling of the ykp radius parameters which we showed to exist for @xmath131 ( see also @xcite ) . as explained in sec .
[ sec4d ] , both effects can be combined for a quantitative extraction of the mean transverse expansion velocity from the ykp radius parameters .
the results from a comprehensive analysis @xcite of resonance decay contributions to the correlation function indicate that they do nt jeopardise such a program .
we would not like to close without remarking that the class of source models studied here is restricted in one crucial aspect : if the source is `` opaque '' @xcite , ( i.e. the particle emission is strongly surface dominated rather than being distributed over the whole emission region with a gaussian geometric weight as assumed here ) , the resulting differences between the variances @xmath91 and @xmath92 lead to much larger contributions in eqs .
( [ 1.17 ] ) , ( [ 1.19 ] ) and ( [ 3.1 ] ) than found in the present study . in ref .
@xcite it is shown that this affects strongly the interpretation of measured ykp parameters , in particular of the `` temporal '' parameter @xmath75 , which typically becomes strongly negative for small @xmath130 due to the negative contribution from @xmath268 in this case . however , that study also shows that presently available data on ykp radii from pb+pb collisions @xcite do not show any such indications for opaqueness of the source and favor models with volume dominated emission .
the na44 data @xcite for hbt radii from heavy ion collisions with pb targets at the cern sps , which within the standard cartesian parametrization seem to be consistent with @xmath269 , are more difficult to interpret because of the non - trivial shape of the acceptance window of this experiment in the @xmath270 plane and the lack of information on the @xmath65-dependence of these parameters .
also , the important consistency relations ( [ 1.15 ] ) so far have only been checked by the na49 collaboration @xcite whose data are in qualitative agreement with the numerical results presented here @xcite . a quantitative comparison with the experiments will be presented as soon as finalized data become available .
this work was supported by grants from daad , dfg , nsfc , bmbf and gsi .
we gratefully acknowledge discussions with h. appelshuser , s. chapman , d. ferenc , p. foka , m. gadzicki , k. kadija , h. kalechofsky , m. martin , p. seyboth and c. slotta .
thanks the cern theory group for their warm hospitality .
using cylindrical coordinates @xmath271 with @xmath272 , we can write the emission function ( [ 2.1 ] ) as @xmath273 with @xmath274 where we defined @xmath275 the @xmath276 and @xmath125 integrations can be done analytically . we use @xmath277 and define @xmath278 the variance of the @xmath125-distribution @xmath242 is @xmath279 defining further @xmath280 we find for the non - vanishing moments up to second order @xmath281 for @xmath282 the @xmath283-averages have to be done numerically . for convenience
we also give @xmath284 \ , , \\
\label{a24 } \langle \tilde x^2 - \tilde y^2 \rangle & = & \left\langle r^2\ , { i_2(b(r))\over i_0(b(r ) ) } \right\rangle _ * -\left\langle r\ , { i_1(b(r))\over i_0(b(r ) ) } \right\rangle_*^2 \ , .
\end{aligned}\ ] ] please note that for small arguments @xmath285 .
thus for small @xmath127 and/or @xmath130 , @xmath196 and @xmath109 vanish linearly and quadratically , respectively . on leave of absence from institute of particle physics , hua - zhong normal university , wuhan , china .
m. gyulassy , s.k .
kauffmann , and l.w .
wilson , phys .
c*20 * , 2267 ( 1979 ) .
d. boal , c.k .
gelbke and b. jennings , rev .
phys . * 62 * , 553 ( 1990 ) .
s. chapman , p. scotto , and u. heinz , phys .
lett . * 74 * , 4400 ( 1995 ) .
s. chapman , p. scotto , and u. heinz , heavy ion physics * 1 * , 1 ( 1995 ) .
s. chapman , j.r .
nix , and u. heinz , phys .
c*52 * , 2694 ( 1995 ) .
wiedemann , p. scotto and u. heinz , phys .
c*53 * , 918 ( 1996 ) .
u. heinz , b. tomik , u.a .
wiedemann , and y .- f .
wu , phys .
b*382 * , 181 ( 1996 ) .
m. herrmann and g.f .
bertsch , phys .
c*51 * , 328 ( 1995 ) .
s. pratt , phys .
lett . * 53 * , 1219 ( 1984 ) ; phys .
d*33 * , 1314 ( 1986 ) .
f. yano and s. koonin , phys .
* 78b * , 556 ( 1978 ) .
podgoretski , sov .
. phys . * 37 * , 272 ( 1983 ) .
gibs coll .
anikina et al . , dubna preprint jinr e1 - 95 - 311 ( 1995 ) .
h. heiselberg and a.p .
vischer , los alamos eprint archive nucl - th/9609022 and nucl - th/9703030 .
schlei , u. ornik , m. plmer , and r. weiner , phys .
b*293 * , 275 ( 1992 ) ; j. bolz , u. ornik , m. plmer , b.r .
schlei , and r. weiner , phys .
b*300 * , 404 ( 1993 ) ; and phys .
rev . d*47 * , 3860 ( 1993 ) .
b. tomik and u. heinz , in preparation .
u. heinz , nucl .
a*610 * , 264c ( 1996 ) .
u.a . wiedemann and u. heinz ,
_ resonance decay contributions to hbt correlation radii _ , los alamos eprint archive nucl - th/9611031 .
e. shuryak , phys .
b*44 * , 387 ( 1973 ) ; sov .
. phys . * 18 * , 667 ( 1974 ) .
s. chapman and u. heinz , phys .
b*340 * , 250 ( 1994 ) .
s.v . akkelin and y.m .
sinyukov , phys .
b*356 * , 525 ( 1995 ) .
t. csrg and s. pratt , in _ proceedings of the workshop on relativistic heavy ion physics at present and future accelerators _ ,
budapest , 1991 , edited by t. csrg et al .
( mta kfki press , budapest , 1991 ) , p. 75 .
rischke and m. gyulassy , nucl .
a*607 * , 479 ( 1996 ) .
lisa et al . , phys .
. lett . * 71 * , 2863 ( 1993 ) .
lisa et al . , phys .
c*49 * , 2788 ( 1994 ) .
na35 coll .
, t. alber et al . , z. phys .
c*66 * , 77 ( 1995 ) ; + na35 coll . ,
t. alber et al . , phys .
* 74 * , 1303 ( 1995 ) .
na44 coll .
, h. beker et al .
lett . * 74 * , 3340 ( 1995 ) .
k. kadija ( na49 coll . ) , nucl .
a*610 * , 248c ( 1996 ) .
bjorken , phys .
rev . d*27 * , 140 ( 1983 ) .
u. heinz , b. tomik , u.a .
wiedemann , and y .- f .
wu , heavy ion physics * 4 * , 249 ( 1996 ) .
s. chapman and j.r .
nix , phys .
c*54 * , 866 ( 1996 ) .
u. heinz , in _ qcd phase transitions _
, ed . by h. feldmeier et al . ,
proceedings of the 25th international workshop on gross properties of nuclei and nuclear excitations , hirschegg , austria , jan .
13 - 18 , 1997 ; gsi report issn 0720 - 8715 , los alamos eprint archive nucl - th/9701054 .
t. csrg and b. lrstad , phys .
c*54 * , 1396 ( 1996 ) ; and nucl .
a*590 * , 465c ( 1995 ) .
b. schlei and u. heinz , in preparation . k.s .
lee , u. heinz , and e. schnedermann , z. phys .
c*48 * , 525 ( 1990 ) ; e. schnedermann and u. heinz , phys .
69 * , 2908 ( 1992 ) ; e. schnedermann , j. sollfrank , and u. heinz , phys .
c*48 * , 2462 ( 1993 ) .
p. braun - munzinger , j. stachel , j.p .
wessels , and n. xu , phys .
b*344 * , 43 ( 1994 ) ; j. stachel , nucl .
a*610 * , 509c ( 1996 ) .
n. herrmann ( fopi coll . ) , nucl .
a*610 * , 49c ( 1996 ) .
a.n . makhlin and yu.m .
sinyukov , z. phys .
c*39 * , 69 ( 1988 ) .
e802 coll .
, v. cianciolo et al . , nucl .
a*590 * , 459c ( 1995 ) .
u. heinz , k.s .
lee , and e. schnedermann , in _ the nuclear equation of state _ , edited by w. greiner and h. stcker , nato asi series 216b , p. 385
( plenum , new york , 1990 ) .
u. mayer , phd thesis , university of regensburg , 1995 , unpublished .
na49 coll .
, t. alber et al . , nucl .
a*590 * , 453c ( 1995 ) ; t. alber , phd thesis , mpi fr physik , mnchen ( 1995 ) , unpublished .
sinyukov , s.v .
akkelin , and a.yu .
tolstykh , nucl .
a*610 * , 278c ( 1996 ) .
a. franz ( na44 coll . ) , nucl .
a*610 * , 240c ( 1996 ) . | the yano - koonin - podgoretski ( ykp ) parametrisation of hanbury brown - twiss ( hbt ) two - particle correlation functions opens new strategies for extracting the emission duration and testing the longitudinal expansion in heavy - ion collisions .
based on the recently derived model - independent expressions , we present a detailed parameter study of the ykp parameters for a finite , hydrodynamically expanding source model of heavy - ion collisions . for the class of models studied here ,
we show that the three ykp radius parameters have an interpretation as longitudinal extension , transverse extension and emission duration of the source in the ykp frame .
this frame is specified by the fourth fit parameter , the yano - koonin velocity which describes to a good approximation the velocity of the fluid element with highest emissivity and allows to test for the longitudinal expansion of the source .
deviations from this interpretation of the ykp parameters are discussed quantitatively . |
we build up a matlab model to simulate a direct current ( dc ) network shown in fig .
[ fig.3.dc for simulink ] to illustrate the flow tracing process .
the flow quantity @xmath142 is given by the electric current @xmath143 in this model .
nodes 1 and 2 are two nodes with current sources where @xmath144a and @xmath145a , respectively .
the resistances of resistors are randomly chosen within the set of integer numbers [ 1,10 ] , shown in tab .
[ tab.3.resistance of resistors ] . the sink flow leaving from the sink nodes 9 and 10
are measured by the current scopes as @xmath146a and @xmath147a .
the current directions are shown in fig .
[ fig.3.dc for simulink abstract ] .
next , we show how to calculate the source - to - sink hidden currents from the current source @xmath148 and @xmath149 to the sink @xmath150 and @xmath151 by different methods .
.resistances of the resistors in fig .
[ fig.3.dc for simulink ] . [ cols="<,^,^,^,^,^,^",options="header " , ] from the matlab simulation results of the dc network , the downstream extended incidence matrix , @xmath59 , is and the downstream contribution matrix , @xmath152 , is @xmath153 we also obtain , from the experiments , that @xmath154 , @xmath155 , @xmath156 , @xmath157 , @xmath158 and @xmath159 . thus , we calculate @xmath160 for @xmath161 and @xmath162 by @xmath163 , @xmath164 , @xmath165 , and @xmath166 .
we note that all these numbers coincide with that in tab .
[ tab.3.flow tracing result proportion sharing ] .
define the _ upstream extended incidence matrix _ , @xmath167 , by @xmath168 we know @xmath169 , implying , @xmath170 since @xmath171
, we have @xmath172 equations ( [ eq.3.upstream extended incidence matrix ] ) and ( [ eq.3.upstream equaiton ] ) imply @xmath173 where @xmath174^t$ ] and @xmath175^t$ ] . from @xmath176 , we have @xmath177_{ij } f_j^t\\ & = \sum_{j=1}^n\left[{\mathbf{k'^{-1}}}\right]_{ij}f_j^{out } \cdot\iota_j^t . \end{split}\ ] ] let @xmath178 be the _ upstream contribution matrix _ whose element , @xmath179_{ij}$ ] , is a _ upstream contribution factor _ indicating how much proportion of the total outflow at node @xmath6 is coming from node @xmath5 , i.e. , @xmath180 .
then , @xmath181 .
the upstream extended incidence matrix , @xmath167 , of the dc network is and the upstream contribution matrix , @xmath182 , is we also obtain @xmath183 , @xmath184 , @xmath156 , @xmath157 , @xmath158 and @xmath159 .
then , @xmath185 , @xmath186 , @xmath187 , and @xmath188 .
the results are the same as that in tab .
[ tab.3.flow tracing result proportion sharing ] .
figure [ fig.comparison of local and global before fs ] shows the experiment results of the local interaction strength and non - local interaction strength in different types of networks , including the er , ws and ba networks .
final results are taken by averaging the results of 100 time - points that are uniformly chosen in the time scale [ 10,20 ] , i.e. , @xmath189 and @xmath190 , where @xmath191 and @xmath192 are the values of @xmath140 and @xmath141 at the @xmath193 time - point .
the dynamic behaviour of the oscillators in these networks is described by the kuramoto model by assigning a small coupling strength , such that the oscillators are in an incoherent state . comparing the results in fig .
[ fig.comparison of local and global before fs ] with that in the paper when fs is present , we find that those pairs of nodes which are non - locally interacted when fs is not present also have non - local interactions when fs is present .
this suggests that the existence of non - local interaction between a pair of nodes strongly depends on the network topological features of the network rather than the coupling strength . | understanding the interactions among nodes in complex networks are of great importance , since it shows how these nodes are cooperatively supporting the functioning of the systems .
scientists have developed numerous methods and approaches to uncover the underlying physical connectivity based on measurements of functional quantities of the nodes states .
however , little is known about how this local connectivity impacts on the non - local interactions and exchanges of physical flows between arbitrary nodes . in this paper , we show how to determine the non - local interchange of physical flows between any pair of nodes in a complex network , even if they are not physically connected by an edge .
we show that such non - local interactions can happen in a steady or dynamic state of either a linear or non - linear network .
our approach can be used to conservative flow networks and , under certain conditions , to bidirectional flow networks .
research on complex networks has been attracting the attention of many scientists for several decades @xcite . if the topology of physical connections between any nodes in a complex network is known ( such as in a power grid ) , one wishes to understand how this topology of the connected nodes drives large - scale non - local behaviour of the network @xcite . to understand large - scale behaviour of complex networks , it is imperative to develop an approach capable of calculating how much physical flow goes from one node to another one through all possible existing paths , a quantity that we refer in this work as hidden " flow , since this quantity is usually inaccessible from measurements and it is unknown . in this paper
, we introduce the flow tracing method which is known in electrical engineering to track power flows @xcite .
this method applied to power grids not only provides the information of how much power is supplied by a particular generator to a particular consumer , but also helps energy companies to cost proper revenue to energy suppliers and to charge proper wheeling fee from consumers .
we avail from this approach to demonstrate how to calculate the hidden flow between any two nodes in conservative flow networks @xcite , by only requiring information about the adjacent flows between any two connected nodes .
this work thus provides a rigorous way to gauge the non - local interactions among nodes in a network .
the applicability of the method is enormous since flow networks can be used to model many complex networks , such as transportation networks , water networks , gas and oil networks and power grids .
furthermore , we extend the method to provide an instantaneous picture of how nodes interact non - locally in non - linear networks by constructing linear equivalent model to these networks .
we also discuss the application of the method to bidirectional networks , such as transportation networks .
a flow network is a digraph , @xmath0 , where @xmath1 and @xmath2 are the sets of nodes and edges , respectively .
the direction of an arbitrary edge in @xmath3 corresponds to the direction of the physical flow on it .
a flow network normally contains three types of nodes : ( 1 ) the source node [ e.g. , node 1 or 2 in fig . [ fig:3.flow ] ( a ) ] , which has a source injecting flow into the network ; ( 2 ) the sink node [ e.g. , node 3 or 4 in fig . [ fig:3.flow ] ( a ) ] , which has a sink taking flow away from the network ; ( 3 ) the junction node [ e.g. , node 5 in fig . [ fig:3.flow ] ( a ) ] , which distributes the flow .
we define @xmath4 to be the _ adjacent flow _ , or simply the _ flow _ , between nodes @xmath5 and @xmath6 , which is the measurable flow coming from nodes @xmath5 to @xmath6 through edge @xmath7 .
particularly , @xmath8 if node @xmath5 is not physically connected with node @xmath6 . in this paper
, we consider the conservative flow networks , which satisfy the following properties : ( 1 ) the adjacent flow from node @xmath5 to node @xmath6 is the opposite of that from node @xmath6 to node @xmath5 , i.e. , @xmath9 ; ( 2 ) the adjacent flow is conserved at a junction node , i.e. , @xmath10 , where node @xmath5 is a junction node ; ( 3 ) there is no loop flow representing a closed path in a flow network , where a loop flow is shown in fig . [ fig:3.flow ] ( b ) ;
( 4 ) there is no isolated node in the network , i.e. , every node must be connected to at least one other node in the network .
a path in a digraph @xmath3 from node @xmath5 to node @xmath6 , @xmath11 , is an alternating sequence of distinct nodes and edges starting from node @xmath5 and ending at node @xmath6 , in which the directions of all edges must coincide with their original directions in @xmath3 .
the _ hidden flow _ , @xmath12 , is defined to be the summation of the flows starting from node @xmath5 and arriving at node @xmath6 through all possible paths from node @xmath5 to @xmath6 .
normally , we can measure or calculate the adjacent flows in a flow network , but it is not easy to obtain the hidden flows , a quantity typically not accessible through measurements .
we find the calculation of hidden flows based on the information of adjacent flows , in a conservative flow network , by the flow tracing " method .
define the _ node - net exchanging flow _ at node @xmath5 by @xmath13 if node @xmath5 is a source node , we have @xmath14 ; we denote @xmath15 by @xmath16 as the amount of the _ source flow _ being injected into the network from a source at node @xmath5 .
we set @xmath17 if node @xmath5 is a sink node or a junction node .
if node @xmath5 is a sink node we have @xmath18 ; we denote @xmath19 to indicate the amount of the _ sink flow _ leaving the network from the sink at node @xmath5 .
we set @xmath20 if node @xmath5 is a source node or a junction node .
assume there is a positive flow from node @xmath5 to node @xmath6 , denoted by @xmath21 .
we use @xmath22 to indicate @xmath4 as an _ outflow _ from node @xmath5 arriving at node @xmath6 , and @xmath23 to represent @xmath4 as an _ inflow _ at node @xmath6 coming from node @xmath5 .
thus , @xmath24 .
@xmath4 can be positive , negative or zero in a flow network .
however , we restrict any outflow or inflow at a node to be a non - negative number .
this means that , if @xmath25 , we force @xmath22 and @xmath23 to be zeros .
analogously , @xmath25 means @xmath26 , we have @xmath27 to denote the outflow from node @xmath6 to node @xmath5 and @xmath28 to be the inflow at node @xmath5 from node @xmath6 .
define the _ total inflow _ at node @xmath5 by @xmath29 and the _ total outflow _ at node @xmath5 by @xmath30 in a conservative flow network , the total inflow of a node is equal to its total outflow , i.e. , @xmath31 .
we assume @xmath32 , @xmath33 , meaning that each node in a flow network must exchange flow with other nodes , i.e. , no node is isolated .
the _ proportional sharing principle ( psp ) _
@xcite states that for an arbitrary node , @xmath34 , with @xmath35 inflows and @xmath36 outflows ( fig .
[ fig.3.proportion node ] ) in a conservative flow network , ( 1 ) the outflow on each outflow edge is proportionally fed by all inflows , and ( 2 ) by assuming that node @xmath5 injects a flow @xmath37 to node @xmath34 , and node @xmath6 takes a flow @xmath38 out of node @xmath34 , we have that the _ node - to - node hidden flow _ from node @xmath5 to node @xmath6 via node @xmath34 is calculated by @xmath39 or by @xmath40 equations ( [ eq.3.proportional sharing principle down ] ) and ( [ eq.3.proportional sharing principle up ] ) result in the same value of @xmath12 , since @xmath41 .
equation ( [ eq.3.proportional sharing principle down ] ) represents the _ downstream flow tracing _
method , where we start tracing the hidden flow from a source node @xmath5 to a sink node @xmath6 , by using the percentage , @xmath42 , to indicate the percentage of @xmath37 that goes to @xmath6 .
equation ( [ eq.3.proportional sharing principle up ] ) denotes the _ upstream flow tracing _
method , where we trace the flow from a sink node @xmath6 to a source node @xmath5 , by knowing the proportion of @xmath38 is provided by @xmath37 .
we just deal with the downstream flow tracing in the main context of this paper and explain the upstream flow tracing by an example in the supplementary material @xcite . with @xmath35 inflows and @xmath36 outflows . ]
define the _ downstream coefficient _ at node @xmath34 for the outflow @xmath38 by @xmath43 to indicate the proportion of the outflow at edge @xmath44 to the total outflow at node @xmath34 .
define the _ upstream coefficient _ at node @xmath34 for the inflow @xmath37 by @xmath45 denoting the proportion of the inflow at edge @xmath46 to the total inflow at node @xmath34 .
then the calculation of @xmath12 can be simply expressed by @xmath47 or @xmath48 .
define the _ sink proportion _ and _ source proportion _ at node @xmath34 by @xmath49 respectively , where the sink proportion , @xmath50 , indicates the proportion of the sink flow to the total outflow at node @xmath34 , and the source proportion , @xmath51 , indicates the proportion of the source flow to the total inflow at node @xmath34 . by defining the sink proportion and source proportion , we are now able to calculate the source - to - sink hidden flow from a source at node @xmath5 to a sink at node @xmath6 denoted by @xmath52 . from eq .
( [ total inflow ] ) , we know that @xmath16 is a part of @xmath53 , where @xmath16 is the source flow at node @xmath5 . from eq .
( [ sink and source proportion ] ) , we know the proportion of @xmath16 to @xmath53 . according to the psp
, we can then calculate the source - to - sink hidden flow by @xmath54 .
the source - to - sink hidden flow is different from the node - to - node hidden flow between two nodes . for example , in fig .
[ fig:3.flow ] ( a ) , @xmath55 indicates how much hidden flow from the 4-unit source at node 1 to the 3-unit sink at node 4 , which is different from @xmath56 that indicating how much hidden flow goes from node @xmath57 to node @xmath58 . in the supplementary material @xcite
, we demonstrate how to use the downstream flow tracing method to trace hidden current flows in a dc network .
however , the downstream flow tracing is most suitable for small networks , because finding all the paths with different lengths between any pair of nodes is a huge amount of work in large sized networks . the extended incidence matrix , proposed in refs .
@xcite , solves this problem . the _ downstream extended incidence matrix _
, @xmath59 , in a flow network including @xmath60 nodes is an @xmath61 dimensional matrix , defined by @xmath62 transform eq ( [ total inflow ] ) to @xmath63 considering @xmath64 , we have @xmath65 from eqs .
( [ eq.3.downstream extended incidence matrix ] ) and ( [ eq.3.downstream equaiton ] ) , we have @xmath66 where @xmath67^t$ ] , and @xmath68^t$ ] .
@xmath69 is an invertible matrix @xcite , thus , @xmath70 , implying that , @xmath71_{ij } f_j^s,\ ] ] where @xmath72_{ij}$ ] is an entry ( @xmath73 row , @xmath74 column ) of the matrix @xmath75 .
equation ( [ eq.3.downstream matrix coefficient 1 ] ) indicates that the outflow of node @xmath5 , @xmath76 , is fed by every source @xmath77 . let @xmath78 be the _
downstream contribution matrix_. considering @xmath79 , we have @xmath80 for a source node @xmath6 with @xmath81 , @xmath82 indicates the node - to - node hidden flow from node @xmath6 to node @xmath5 , i.e. , @xmath83 . to demonstrate that the hidden flow can be calculated similarly regardless of the nature of the nodes , we introduce an _ equivalence principle _ and treat any sink or junction node as a hypothetical source node , without altering the original network topology and flows according to what is explained next .
if node @xmath6 is a sink node or junction node with a total inflow @xmath84 and @xmath85 , we treat node @xmath6 as a hypothetical source node with @xmath86 , where the hypothetical source takes the place of all the edges injecting flows into @xmath6 . by this treatment
, we have not changed the flows in the network , but we can hypothetically treat node @xmath6 as a source node with @xmath87 , such that the node - to - node hidden flow from node @xmath6 to node @xmath5 can also be calculated by @xmath83 . from our analysis , @xmath88_{ij}$ ] is a _ donwstream contribution factor _ indicating how much hidden flow goes from node @xmath6 to @xmath5 , i.e. , @xmath89 for any pair of nodes .
then , @xmath90 indicates the source - to - sink hidden flow from node @xmath6 to node @xmath5 .
an example is given in the supplementary material @xcite explaining how to apply the downstream extended incidence matrix to carry out flow tracing .
we now formally conclude and extend our work to analyse the nonlocality study in complex networks .
let @xmath5 , @xmath6 , @xmath35 , @xmath36 , @xmath91 , @xmath92 be different nodes in a conservative flow network , where node @xmath5 has a source , node @xmath6 has a sink , nodes @xmath35 , @xmath36 are connected by edge @xmath93 with an adjacent flow @xmath94 , and nodes @xmath91 , @xmath92 are connected by edge @xmath95 with an adjacent flow @xmath96 .
the non - local interaction calculation includes the following parts : ( 1 ) the node - to - node hidden flow from node @xmath5 to node @xmath6 , @xmath12 , is calculated by @xmath97 ( 2 ) the source - to - sink hidden flow from node @xmath5 to node @xmath6 , @xmath52 , is calculated by @xmath98 ( 3 ) the node - to - edge hidden flow from node @xmath5 to edge @xmath93 , @xmath99 , is calculated by @xmath100 ; ( 4 ) the edge - to - node hidden flow from edge @xmath93 to node @xmath6 , @xmath101 , is calculated by @xmath102 and ( 5 ) the edge - to - edge hidden flow from edge @xmath95 to edge @xmath93 , @xmath103 , is calculated by @xmath104 .
next , we extend these concepts to show how nodes interact non - locally in non - linear systems by constructing linear model analogous to the non - linear networks .
let the equation @xmath105 indicate a dynamic scheme describing the behaviour of @xmath60 coupled nodes , where @xmath106 is the dynamical variable of each node , @xmath107 is the isolated dynamic function , @xmath108 is the element of the laplacian matrix , and @xmath109 is an arbitrary coupled dynamic function .
we treat the system as a flow network by interpreting @xmath110 as the node - net exchanging flow at node @xmath5
. the value and sign of @xmath111 may change over time . if @xmath112 , we treat node @xmath5 as a source node at time @xmath113 and the source flow is @xmath114 . if @xmath115 , we treat node @xmath5 as a sink node at time @xmath113 and the sink flow is @xmath116 . if @xmath117 , we treat node @xmath5 as a junction node at time @xmath113 .
let @xmath118 be the adjacent flow from node @xmath5 to node @xmath6 .
if @xmath119 , we have @xmath120 as the outflow from node @xmath5 and @xmath121 as the inflow at node @xmath6 at time @xmath113 .
if @xmath122 , we have @xmath123 as the outflow from node @xmath6 and @xmath124 as the inflow at node @xmath5 at time @xmath113 .
by doing this interpretation , we are constructing an equivalent linear conservative flow network that instantaneously behaves in the same way as the non - linear network described by eq .
( [ eq.3.dynamic system ] ) .
this enables us to calculate the non - local interactions in the equivalent linear flow network which informs us about the non - local interactions in the original non - linear network .
we demonstrate the non - local interaction process in non - linear networks by tracing hidden flows in networks where the dynamics of nodes is described by a revised kuramoto model @xcite , as given by @xmath125 where @xmath126 is the coupling strength , @xmath108 is the entry of the laplacian matrix , @xmath127 and @xmath128 indicate the phase angle and natural frequency in a rotating frame , respectively . in this rotating frame , @xmath129 , when the oscillators emerge into _ frequency synchronisation ( fs ) _ for a large enough @xmath126 @xcite . in the fs state ,
all the node - net exchanging flows @xmath130 and all the adjacent flows @xmath131 are constants , since @xmath132 are constants .
let @xmath133 be a normalised variable in [ 0,1 ] indicating the local interaction strength between oscillator @xmath5 and @xmath6 , where @xmath134 is the maximum of all absolute values of adjacent flows . since @xmath9 , we have @xmath135 .
every hidden flow is traced by considering that flows are directed .
this implies that all the calculated hidden flows are non - negative and at least one of @xmath12 and @xmath136 is 0 .
we let @xmath137 be the non - local interaction strength between oscillator @xmath5 and @xmath6 , where @xmath138 is the non - zero one between @xmath12 and @xmath136 , and @xmath139 is the maximum of all hidden flows .
this definition of the non - local interaction strength allows us to compare @xmath140 and @xmath141 for the same pair of nodes in a network .
we construct three types of networks with 25 nodes , namely the erds - rnyi ( er ) @xcite , watts - strogatz ( ws ) @xcite and barabsi - albert ( ba ) models @xcite .
the dynamic behaviour of the nodes in these networks follows eq .
( [ eq.4 kuramoto model ] ) .
the natural frequencies of oscillators are set to be random numbers in [ 0,1 ] .
figure [ fig.comparison of local and non - local ] shows the comparison of the local interactions and the non - local interactions when the oscillators emerge into fs with a large enough @xmath126 .
figures [ fig.comparison of local and non - local ] ( a ) , ( b ) and ( c ) show the local interaction strengths , @xmath140 , for er , ws and ba networks , respectively .
figures [ fig.comparison of local and non - local ] ( d ) , ( e ) and ( f ) demonstrate the non - local interaction strengths , @xmath141 , for er , ws and ba networks , respectively .
figure [ fig.comparison of local and non - local ] ( d ) exposes some hidden interactions that fig . [ fig.comparison of local and non - local ] ( a ) does not show to exist in an er network . by comparing figs .
[ fig.comparison of local and non - local ] ( b ) and ( e ) , we see that a randomly rewired edge in a ws network not only produces interaction between the two adjacent nodes connected by this edge , but also creates functional clusters among nodes close to the two adjacent nodes .
so , complex systems , such as social networks that are modelled by a ws network @xcite can in fact be better connected than previously thought .
we constructed the ba network by assigning smaller labels to nodes with larger degrees . both figs .
[ fig.comparison of local and non - local ] ( c ) and ( f ) illustrate the strong interactions among the nodes with large degrees ( small labels ) .
figure [ fig.comparison of local and non - local ] ( c ) shows that the interactions between unconnected nodes with small degrees ( large labels ) are week or inexistent , though , such interactions are revealed in fig .
[ fig.comparison of local and non - local ] ( f ) . through this comparison
, we understand that two nodes in a network may strongly interact with each other even if they are not connected by an edge . in the supplementary material @xcite , we present a non - local interaction study for these networks when fs is not present . comparing the results of the experiments with and without the existence of fs , we find that those pairs of nodes which are non - locally interacting when fs is not present also have non - local interactions in the result when fs is present .
this suggests that , the existence of non - local interaction between a pair of nodes strongly depends on the network topological features of the network rather than the coupling strength .
our method can also be applied to a bidirectional flow network if the network can be separated into two independent unidirectional networks .
for example , under the assumption that all roads are bidirectional , we can separate the transportation network of a city into two networks .
one network includes all the left hand roads and the other one contains all the right hand roads . using our method
, we can trace the transportation flow between any two nodes in these two networks .
37ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( ) @noop * * , ( ) @noop _ _
( , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * ( ) @noop * * , ( ) in @noop _ _ , vol .
( , ) pp .
@noop * * , ( ) @noop * * , ( ) in @noop _ _ , vol .
( , ) pp . in @noop
_ _ ( , ) pp .
@noop * * , ( ) in @noop _ _ ( , ) pp . @noop * * , ( ) @noop * * , ( ) @noop ( ) in @noop _ _ , vol .
( , ) pp .
@noop in @noop _ _ ( , ) pp .
@noop _ _ ( , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , ) * supplementary material * |
low - lying quadrupole oscillations in nuclei have been studied extensively with all kinds of nuclear structure models . among the more popular are the geometrical model@xcite and the interacting boson model ( ibm)@xcite .
although rather different in their microscopic interpretation and formulation , both of them have been extremely successful in describing low - lying energy levels associated to quadrupole oscillations . in three different situations
both models provide analytical results that can be identified in the geometrical model as those corresponding to spherical@xcite , well quadrupole deformed @xcite and deformed @xmath0-unstable shapes@xcite .
the corresponding situations occur in ibm for the u(5)@xcite , su(3)@xcite and o(6)@xcite limits respectively .
connection between both models has been established by exploring the geometrical content of the ibm by using coherent states @xcite .
high - lying dipole states are studied in the context of the geometrical model by using the dynamic collective model within the hydrodynamical approach@xcite . in the ibm context
these states can be studied with the so called dipole boson model @xcite , which includes a dipole @xmath1 boson into the usual ibm space ( quadrupole @xmath2 and @xmath3 bosons ) .
the connection between both approaches , especially for the case of well deformed nuclei , has already been discussed@xcite . in this paper
we are interested in studying the coupling of the dipole mode to a @xmath0-unstable quadrupole form .
the results of the geometrical model will be presented and the study within the dipole boson model will be performed both in the laboratory frame and in the intrinsic frame projecting afterwards onto the laboratory frame .
the paper is structured as follows . in section 2
the coupling of the dipole mode to a @xmath0-unstable rotor is analysed within the geometrical model . in section 3 the coupling of the dipole mode to an o(6 ) quadrupole deformed nuclei is investigated in the interacting boson model .
this is done in two steps , first we study the problem in the intrinsic frame and then the projection to the laboratory frame is performed .
comparisons of exact results and those obtained in the intrinsic frame plus projection for different sets of parameters are presented and discussed . in section 4
examples of calculations for photonuclear cross - section are presented , and the different approximations compared .
finally , in section 5 the paper is summarised and the main conclusions assessed .
the interaction between dipole and quadrupole oscillations has been studied in the geometrical model .
the coupling to the quadrupole degree of freedom splits the energy of the dipole mode , and within the linear coupling in the quadrupole amplitude@xcite the energies of the three resulting dipole resonances as seen from the intrinsic frame are @xmath4~~ ; ~~\mu=-1,0,1~~,\ ] ] where @xmath5 is the unperturbed dipole energy , @xmath6 is the restoring force parameter , and @xmath7 is the dipole quadrupole coupling coefficient . a plot of these energies with respect to the unperturbed value @xmath5 is given in fig .
1 as a function of the asymmetry parameter @xmath0 , for a fixed value of the quadrupole deformation parameter @xmath8 a detailed structure of the dipole absorption line can be obtained from a numerical diagonalization of the coupled dipole
quadrupole hamiltonian . a simple model in the strong coupling approximation for extracting the dipole strength distribution from the intrinsic values is provided by the procedure of averaging the three contributions over the different quadrupole @xmath8 and @xmath0 deformations , according to the ground state distribution .
the relative strength of the dipole transition probability @xmath9 per unit of energy is in this way proportional to @xmath10 where @xmath11 is the wave function of the ground state resulting from the quadrupole energy surface .
the specific case of @xmath0-unstable rotor occurs when the intrinsic energy surface is isotropic in the @xmath0 variable , while displaying a sharp minimum for deformation @xmath12 . assuming the extreme situation of a very sharp minimum
, we can in first approximation keep fixed the value of @xmath12 and only average over the @xmath0 variable . in this adiabatic picture
one obtains the probability distribution shown in fig .
2 . the distribution is symmetric around @xmath5 , with three maxima , two at the edges ( at energies @xmath13 ) and one in the center ( at energy @xmath5 ) , with @xmath14 and with zeros occurring at energies @xmath15 .
the resulting total splitting @xmath16 is therefore directly proportional to the quadrupole deformation @xmath12 , as in the case of a static quadrupole deformation .
note that although the probability at the edges is 3 times the one at the center , the total probability is anyway exactly equally divided into the three bumps .
it may be worth noticing that , aside from the broadening of the three peaks , the predicted situation in the @xmath0-unstable case is actually rather similar to the prediction associated with a static quadrupole deformed system with @xmath0 equal to 30 degrees .
this similarity of the two situations is a feature that also appears for other observables and has been discussed elsewhere@xcite .
to include the dipole degrees of freedom in the usual ibm , one negative parity dipole @xmath1boson ( @xmath17 ) has to be included in addition to the positive parity quadrupole @xmath2 ( @xmath18 ) and @xmath3 ( @xmath19 ) bosons . for a boson system with @xmath20 quadrupole bosons plus @xmath21 dipole boson ,
the hamiltonian can be written as @xmath22 where @xmath23 is the usual ibm hamiltonian for quadrupole degrees of freedom , @xmath24 is the dipole hamiltonian given by @xmath25 and @xmath26 is the interaction between quadrupole and dipole degrees of freedom .
the operator @xmath27 is the @xmath1-boson number operator , which in this case is @xmath28 or @xmath21 , and @xmath5 is the unperturbed dipole energy . for the interaction a simple quadrupole - quadrupole form
is assumed @xmath29 where @xmath30 is the usual ibm quadrupole operator @xmath31 and @xmath32 the operators @xmath33 ( where @xmath0 stands for @xmath2 , @xmath3 or @xmath1 bosons ) are introduced so as to have operators with the appropriate properties under spatial rotations .
the simplest form for the dipole operator is @xmath34 where @xmath35 is an overall normalization parameter , which will be assumed equal to unity in the applications .
first the laboratory ibm calculation is presented .
the quadrupole @xmath36 hamiltonian for the o(6 ) limit can be written in the form @xmath37= b ~ c_2[o(5 ) ] + c ~ c_2[o(3 ) ] + d ~ c_2[o(6 ) ] \label{hamo6}\ ] ] where @xmath38 stands for the quadratic casimir operator of the indicated algebra as defined in ref .
energies associated with this @xmath36 hamiltonian can be obtained analytically to be @xmath39 where @xmath40 , @xmath41 and @xmath42 are the labels of the irreducible representations of the algebras o(6 ) , o(5 ) and o(3 ) respectively .
the coupling to the dipole oscillations and the calculation of transition probabilities is done numerically by using the computer codes gdr and gdrt , respectively @xcite .
the states belonging to the @xmath43 representation are favoured in energy and the relative position of the states belonging to other representations is governed by the parameter @xmath3 . in the o(6 ) limit these representations are not connected by the quadrupole @xmath44 operator if @xmath45 is taken in eq .
( [ quadrb ] ) , and they therefore remain unmixed ( and characterized by a good value of @xmath40 ) also with the inclusion of the dipole - quadrupole coupling . consequently the dipole operator will only connect the ground state with states of the @xmath43 representation , which will therefore be the only ones contributing to the @xmath46 distribution .
this implies that , as far as the dipole distribution is concerned , the calculation is completely insensitive to the value of the parameter @xmath3 . for the other parameters , in order to favour the comparison with preceding results obtained within the simple scheme based on the geometrical model , the parameter @xmath47 has been put equal to zero in order to minimize the splitting due to the angular momentum , a feature not accounted for by the unprojected intrinsic state .
the results obtained for the energies of dipole states and the corresponding @xmath46 are shown in fig .
3 ( left panels ) for different values of the parameter @xmath48 in front of the o(5 ) casimir operator . at variance with the previous distribution displayed in fig .
2 we have now a discrete distribution due to the finite number of bosons and consequently finite number of states .
note that the pure su(3 ) limit with @xmath49 gives a dipole strength distribution with only two lines , since only two @xmath50 states can be obtained by coupling the dipole boson with the ground state rotational band . in the o(6 ) limit there are instead @xmath51 @xmath50 states connected to the ground state by the dipole operator , namely all those originated by the @xmath52 and @xmath53 states appearing in the @xmath54 band .
we can however compare the envelope of this discrete distribution with the continuous distribution given by the geometrical model in its simplest adiabatic version . in order to facilitate comparisons an appropriate discretization of fig . 2 for the case of n=15 bosons , an unperturbed dipole energy @xmath5 = 15 mev , and a dipole - quadrupole coupling @xmath55=0.2 mev ( same parameters as in fig .
3 ) is presented in fig .
4 . in the limit of very small values of @xmath48 , namely in strong - coupling situations ,
the splitting due to the different values of @xmath41 within each o(6 ) representation is small compared to the coupling . as a consequence the patterns obtained in figs .
3 and 4 are similar , since the mixing between the different states induced by the quadrupole - dipole interaction is large . at the other extreme
, for values of @xmath48 large with respect to the dipole - quadrupole coupling the pattern observed in fig .
3 becomes very asymmetric , tending to concentrate the transition strength onto the lowest state , washing out any resemblance with the one obtained in the geometrical model , fig .
4 , and giving rise to a situation close to the pure spherical case .
in fact in this weak - coupling limit , because of the large @xmath41 splitting , only the lowest @xmath50 states have large overlaps with the initial s d ground state . for a better understanding of the situation we will study next the problem within the intrinsic frame in the ibm .
let us now study the problem of the coupling of quadrupole and dipole degrees of freedom in the intrinsic frame , within the ibm description .
the basic idea of the intrinsic - frame formalism in this case is to consider that the pure quadrupole states of the ground `` band '' are globally described by a boson condensate of the form @xmath56 where the basic boson is given by @xmath57~~ , \label{ground2}\ ] ] @xmath8 and @xmath0 being obtained by minimizing the energy surface . in the case of su(3 ) one has @xmath58 and @xmath59 , and the intrinsic state is fully representative of the ground - state rotational band . in the case of o(6 ) , instead , one has @xmath60 , and the intrinsic state , which gives rise to a @xmath0-independent energy surface , is associated with all states of the `` band '' corresponding to the @xmath43 representation . for the dipole part , the corresponding building blocks are @xmath61 , @xmath62 and @xmath63 .
thus , the three states @xmath64 provide with an intrinsic basis where the dipole - quadrupole states can be studied . in this basis , the pure @xmath44 and @xmath1 parts of the hamiltonian are diagonal and the only non - diagonal contributions come from the interaction term .
the matrix elements needed in order @xmath65 are @xmath66 where @xmath67 have already been calculated @xcite , @xmath68~ , \label{qsd1}\ ] ] @xmath69~~. \label{qsd2}\ ] ] the @xmath30 matrix elements not specified are zero .
the matrix elements of @xmath70 can be easily computed to be @xmath71 all the remaining relevant matrix elements for our study are zero .
the parameter @xmath72 is the structure constant in the quadrupole operator @xmath30 .
the su(3 ) case ( studied in ref .
@xcite ) is obtained for @xmath73 and the o(6 ) limit corresponds to @xmath45 .
the dipole energies are obtained by diagonalising the interaction @xmath26 in the basis @xmath64 .
they are analytically given by @xmath74 ~,\label{enerp1 } \\
\lambda_- & = & \epsilon_1 - \kappa ~\left [ \frac{1}{\sqrt{6}}~ q^{(sd)}_{0}(\beta , \gamma ) - q^{(sd)}_{2}(\beta , \gamma ) \right ] ~,\nonumber\end{aligned}\ ] ] which provide the dependence of the dipole energies on the ibm deformation parameters @xmath75 . with the choice @xmath73
, corresponding to the su(3 ) limit , eqs .
( [ enerp1 ] ) become equivalent to eqs .
( 9b ) in ref .
@xcite with terms linear and quadratic in @xmath8 .
the corresponding energies are shown in fig .
the o(6 ) limit corresponds instead to @xmath45 . in this case
( [ enerp1 ] ) have only a linear term in @xmath8 , and the corresponding energies are shown in fig .
these latter energies are equal to those obtained in the preceding section for the geometrical model , once the correspondence is made between deformation parameters and coupling strengths in the two models , such that @xmath76 as in the case of the geometrical model , a simple way of obtaining the dipole strength distribution from the ibm intrinsic state is to average the intrinsic energies over @xmath0 . in analogy to the intrinsic energies , the resulting distribution will show , _ mutatis mutandis _
, a pattern identical to the one shown in fig . 2 for the geometrical model .
we have seen so far that the straightforward use of the intrinsic state in the ibm without any projection technique gives results similar to the ones obtained in the geometrical model .
these results are consistent with those obtained in the laboratory frame only for boson hamiltonians that give rise to splittings in the elements of each o(6 ) multiplet which are much smaller than the splitting due to the dipole - quadrupole coupling . in all other more general cases , the effect of the breaking of degeneracy of each multiplet is essential , and proper treatment of the projection to the laboratory system from the intrinsic state is needed . in this subsection
it is therefore shown how to project the intrinsic state for the ground - state representation @xmath43 into the @xmath41 component . with this
we would like to show that the calculations in the intrinsic frame , after projecting after on @xmath41 , reproduce the laboratory results .
in the o(6 ) case the projector is known @xcite , and the important feature is that it produces exactly the laboratory states @xmath77 , belonging to the representation @xmath43 .
we have considered states of the form @xmath77 coupled to @xmath78 to total @xmath79 .
the states @xmath77 are obtained by applying the projector @xmath80 ( bs functions @xcite ) to the intrinsic ground state @xmath81 as given in eqs .
( [ ground1],[ground2 ] ) . for each value of @xmath41
there is only one state ( either @xmath82 or @xmath83 ) to be considered .
thus , for a given @xmath20 there will be @xmath84 dipole states . in order to diagonalize the hamiltonian in the @xmath79 subspace
, we have included diagonal ( trivial ) and non diagonal terms .
for the latter ones we have used the results of the matrix elements of the interaction @xmath85 in the intrinsic frame ( [ qqdef]-[qp ] ) and made the appropriate integrations on @xmath0 and the euler angles @xmath86 to go from the intrinsic to the laboratory frame .
the resulting matrix to be diagonalized in the lab connects states of the same @xmath41 ( diagonal terms ) and states of @xmath41 with @xmath87 as known .
we have done full calculations up to @xmath88 for the case @xmath45 and found simple expressions for the non - diagonal matrix elements that allow calculations for any value of @xmath20 .
these expressions are ( we will use the notation @xmath89 for the states , with the selection rule @xmath90 = @xmath91 ) @xmath92 @xmath93 where @xmath94 .
we have checked that these expressions coincide , in leading order in 1/n , with those associated with the corresponding `` exact '' states for @xmath95 and @xmath96 . with these matrix elements
we have done a series of calculations and the resulting dipole @xmath97 distributions are compared in fig .
3 ( right panels ) with the exact results in the laboratory frame for different values of @xmath48 . in fig .
4 the corresponding distribution obtained from the unprojected intrinsic state ( equal for all values of @xmath48 ) is plotted for comparison . in all the cases
we have fixed the number of bosons to 15 and the intensity of the quadrupole ( @xmath44)-quadrupole ( @xmath1 ) interaction to the value @xmath98 mev .
we have changed the intensity of the @xmath99 term by acting on the @xmath48 term in eq .
( [ hamo6 ] ) . in the following points we comment on the results given in fig .
* \i ) @xmath100 mev .
this gives an excitation energy of the @xmath101 of 0.0012 mev , almost degenerate case .
the pattern of the @xmath102 in the exact calculation @xcite ( left lowest panel ) is as mentioned before ( three maxima , two at the edges with @xmath102 about 3 times the value of the maximum at the center ) . in our projected calculation
( right lowest panel ) we obtain similar pattern with three bumps but now the maxima at the edges are lower than in the full lab calculation and the maximum at the center is higher than the one in the exact calculation . in this way
we obtain three maxima of roughly the same height .
the two minima are more or less in the same places in both calculations .
it is remarkable that the number of states in the projected calculation close to the edges is larger than in the full lab calculation and the number of states close to the center is smaller . in that way the summed @xmath102 values in each of the three bumps
is the same in both calculations .
the energy separation between edges is also about the same in both calculations . *
\ii ) @xmath103 mev .
this gives an excitation energy of the @xmath101 of 0.012 mev .
now the patterns are more similar in both calculations , the population of the lowest part being favoured in both cases .
the results start to deviate appreciably from the unprojected ones .
* \iii ) @xmath104 mev .
this gives an excitation energy of the @xmath101 of 0.120 mev .
now the patterns of the laboratory and the projected intrinsic state are very close , with strong deviation from the pure results of the intrinsic state .
* \iv ) @xmath105 mev .
this gives an excitation energy of the @xmath101 of 1.2 mev .
now only the 3 - 4 lowest states are populated and energies and @xmath102 s are practically identical in both calculations . from this comparison we conclude that our intrinsic calculation followed by projection on @xmath41 gives a good approximation to the laboratory results .
it is worth noticing that in all cases the agreement is excellent except for the almost degenerate case , which is surprising since in that case the intrinsic state is expected to be a good approximation ( see fig .
3 lowest left panel and fig .
4 ) . this can be due either to the projection method or to the intrinsic trial wave function . in this case
the projection method is known to be exact for the large @xmath20 limit , thus the problem must come from the trial wave function . as mentioned above , in the limit of very small values of @xmath48 ( strong - coupling situations )
the splitting of the different values of @xmath41 within each o(6 ) representation is small compared to the coupling .
consequently , the trial wave function proposed as @xmath64 with @xmath81 given by eqs .
( [ ground1],[ground2 ] ) could be not completely appropriate in this case .
the coupling to the additional dipole boson has in fact destroyed the full @xmath0-unstability of the system ( cf .
5 ) , leading to an additional variation of the order of 1/@xmath20 in the basic boson of the condensate .
the situation is similar to that obtained , for example , in the coupling of an odd particle to a su(3 ) boson core , where the additional particle slightly shiftes the position of the minimum of the energy surface in the @xmath8-@xmath0 plane .
it should be noted that realistic cases of o(6 ) nuclei correspond to values of @xmath48 around @xmath106 mev , far from degeneration in @xmath41 . we have checked this by making a set of calculations for a fixed and reasonable value of @xmath48 and changing the interaction @xmath55 . in those cases agreement between laboratory calculations and intrinsic plus
projection is fine for any reasonable value of @xmath55 . only for extremely large and unphysical values of @xmath55
some discrepancies between both calculations start to appear . once the energy spectra and the dipole transition strengths
have been discussed we present in the next section the cross sections for absorption of unpolarized @xmath0 radiation in the gdr region .
dipole strength distributions are traditionally measured in photoabsorption processes .
although photoabsorption cross - sections directly reflect the dipole strength , some of the features associated with this strength may disappear , be masked or modified by the effect of the finite widths .
for this reason , once the distribution of the individual dipole states has been analyzed in the preceding section , we prefer in this section to compare results obtained within the ibm dipole model in the laboratory frame , unprojected and projected intrinsic frame directly for photoabsorption cross - sections .
the cross section for photoabsorption is in fact given by @xcite @xmath107 where @xmath108 are the widths of the dipole states at energy @xmath109 .
the dipole widths are either taken as a constant or assumed to have a power law dependence on the energy , @xmath110 where @xmath111 is the energy of the unperturbed dipole state and @xmath112 its width .
both @xmath112 and @xmath113 are used as adjustable parameters when one is fitting experimental data . in fig . 6
we present the cross - sections obtained for the two extreme ( @xmath48=0.00015 mev and @xmath48=0.15 mev ) dipole strength distributions corresponding to fig .
full lines give the results for the case @xmath48=0.00015 mev ( almost @xmath41-degenerate case ) while dashed lines are for @xmath48=0.15 mev . in both cases @xmath114 mev and the intensity of the quadrupole ( sd)-quadrupole ( p ) interaction ( @xmath55 ) is 0.2 mev .
left panels give the laboratory results and right panels the intrinsic plus projection on @xmath41 results .
we have done two sets of calculations which select different widths for the dipole states .
upper panels correspond to the case in which the dipole widths are assumed to have a power law dependence on the energy .
since we are not working with an specific nucleus , we have taken @xmath112= 0.026 @xmath115 ( mev ) and @xmath113=1.91 .
these values come from one global parametrization of experimental photoabsorption cross sections for @xmath116 @xcite .
that implies widths ranging from about 3.5 to 6.0 mev for @xmath109 ranging from 13 to 17 mev , respectively .
lower panels represent equivalent cases but taking the dipole widths as a constant equal to 2.5 mev .
first of all we should realize that the actual shape of the photoabsorption cross - section is very much dependent on the widths assumed .
the dashed lines ( case with @xmath48=0.15 mev ) in all panels still reflect contributions from the two lower states since they are separated by about 3 mev .
by contrast , all the states contribute to the almost @xmath41 degenerate case ( full lines , @xmath48=0.00015 mev ) since now the states differ by less than 0.5 mev .
thus the image on the photoabsorption cross - section of the shape of the dipole strength distribution in fig .
3 is diluted , even showing a plateau when we use the biggest widths ( upper panels ) . in all the cases , independently of the election of @xmath48 and the dipole widths assumed ( either constant or energy dependent ) , a good agreement between the laboratory and @xmath41-projected intrinsic frame calculations
is observed when comparing left and right panels , although as in fig . 3 as the absolute value of the ratio b/@xmath55 decreases its quality becomes a bit worse .
it is worth noticing that the calculations for the case of unprojected geometrical adiabatic situation given in fig .
4 are indistinguishable from those plotted with full line on the right - hand - side panels ( @xmath48=0.00015 mev ) as expected .
we have studied the problem of the coupling of the quadrupole and dipole oscillations in @xmath0-unstable nuclei in the framework of the ibm . we have obtained and explained the results in the laboratory by studying the problem in the intrinsic frame taking at higher order the non - degeneracy in @xmath41 . in the @xmath41-degenerate limit the dipole mode total splitting ( 2@xmath117 ) turns out to be proportional to the dipole - quadrupole coupling strength , the number of sd bosons considered , n , and the quadrupole deformation .
furthermore , the dipole mode splits in n+1 lines and its distribution is symmetric around the unperturbed dipole energy .
this symmetry is destroyed as soon as we allow non - degeneracy in @xmath41 , giving rise to a very asymmetric distribution . in the weak coupling limit
the transition strength is concentrated onto the lowest states .
since this limit seems to correspond to the actual situation in typical o(6 ) nuclei , @xmath118pt or @xmath119ba for instance , the explicit treatment of the non - degeneracy in @xmath41 is crucial . in this case
the results from the geometrical model in the adiabatic limit are quite different from the dipole boson model ones . in this sense , the o(6 )
limit is different to the su(3 ) case in which the rotational excitation energy is low and can be ignored in first approximation and , consequently , the geometrical model in the adiabatic limit gives an appropriate description .
calculations of photoabsorption cross - sections using the intrinsic state in the ibm followed by projection on @xmath41 give a good approximation to the laboratory results .
no experimental data on o(6 ) nuclei are available and they could be valuable to elucidate whether the dipole strength distribution in actual o(6 ) nuclei is in agreement with the @xmath41-projected dipole boson model results presented here .
this work was supported in part by a spanish italian cicyt - infn agreement and by the spanish dgicyt under project number pb98 - 1111 .
part of this work was done while e.g.l .
was at sevilla university as a marie curie fellow with contract erbfmbict-983090 .
we acknowledge useful discussions with f. iachello and a. leviatan .
we thank p. van isacker for providing us with the ibm codes gdr and gdrt . | the coupling of the high - lying dipole mode to the low - lying quadrupole modes for the case of deformed @xmath0-unstable nuclei is studied .
results from the geometrical model are compared to those obtained within the dipole boson model .
consistent results are obtained in both models .
the dipole boson model is treated within the intrinsic frame , with subsequent projection onto the laboratory frame . as an application ,
calculations of photonuclear cross - sections in @xmath0-unstable nuclei are presented . *
pacs numbers : 21.60.ev , 21.60.fw , 21.10.re * |
simple two - dimensional ( 2d ) ising - like models may present a complex behavior with rich phase diagrams .
a complex behavior may appear even in face of simple competing interactions .
in general , a complex behavior results from the frustration phenomenon caused by these competing interactions at different length scales . the 2d ising model with nearest - neighbor ferromagnetic exchange and dipolar interaction has been the focus of considerable interest because of its magnetic properties @xcite .
the hamiltonian model is defined by the ising lattice model with nearest - neighbor exchange interaction @xmath2 and dipolar contribution of strength @xmath3 , which is rewritten as @xmath4 where @xmath5 . here
, we have adopted the convention @xcite of summing over all distinct pairs of lattice spins at distances @xmath6 to define the constant @xmath7 .
the distances @xmath6 are measured in units of lattice . the competition between the exchange and dipolar interactions results in a phase diagram that is rather sensitive to the @xmath8 ratio @xcite . for @xmath9 ,
the model presents antiferromagnetic ground - state characterized by stable checkerboard like configurations @xcite . for larger @xmath10 values ,
the ground state changes to striped phases .
these phases are characterized by magnetic domains displayed in stripes of alternating spins perpendicular to the lattice plane .
each stripe corresponds to a pattern of upward or downward spins , whose width @xmath11 increases with @xmath10 @xcite .
as the temperature increases , for fixed couplings , an order - disorder transition takes place until the system reaches its usual paramagnetic phase . this theoretical aspect has been observed in experiments with ultrathin ( i.e. , a few atomic layers ) metal films on metal substrates @xcite below certain temperatures as a consequence of a reorientation transition of their spins .
apart from short - range interactions , long - range dipolar interactions are also present between the magnetic moments in these systems . as a matter of fact , even for the simple ferromagnetic ising model
, one should always include a long - range @xmath12 repulsive interaction between the magnetic moments in addition to the short - range exchange interaction term .
theoretical studies of magnetic systems have avoided the inclusion of the relatively small dipolar interaction compared with the exchange interaction . however , as we have mentioned the dipolar interaction plays an important role for quasi 2d - magnetic systems in characterizing metal films @xcite .
furthermore , from a theoretical viewpoint , a long - range repulsive interaction @xmath13 for @xmath14-dimensional systems , with @xmath15 , in addition to the usual nearest - neighbor ferromagnetic interaction , leads to the absence of ferromagnetism for all temperatures @xcite .
efforts have been made to determine the phase diagram as a function of the coupling @xmath10 from mean field approximation and numerical simulations @xcite .
analytical and monte carlo ( mc ) calculations have shown the following picture : for a fixed coupling @xmath16 , the phase diagram initially exhibits the alternating striped spin configurations as described above , characterizing the so called smectic striped phase . as the temperature increases
, the initial picture describes a transition into the tetragonal phase @xcite characterized by states with orientationally disordered stripes but still preserving some of this structural form , which tends to the completely disordered paramagnetic state .
recent numerical results related to this striped - tetragonal phase transition indicate a continuous transition for @xmath17 @xcite , a clear first - order transition for @xmath0 @xcite , a likely weaker first - order transition for @xmath18 @xcite , and a continuous transition for @xmath19 @xcite and @xmath20 @xcite . on the other hand , a theoretical approach developed by abanov _
@xcite , based on a continuum limit for the hamiltonian , has predicted a new domain phase in the phase diagram .
in addition to the striped and tetragonal phases , there would be an intermediate ising nematic phase , whose existence would depend on the parameters of the model .
different methods have been used to study the criticality of this model . as summarized above
, the order of the phase transition between the low temperature smectic striped and the high temperature tetragonal phases is still under debate and seems to depend on the coupling @xmath10 .
simulations have been hampered because large lattice size simulations are very cpu timeconsuming due to the dipolar term . in spite of these limitations ,
recent literature works have presented very satisfactory results , including numerical evidence about the intermediate ising nematic phase @xcite .
the conclusion seems to be that the nematic phase can be observed in a narrow range of temperatures and is @xmath10-dependent .
this amounts to observing not only a single peak in the specific heat , initially corresponding to the expected striped - tetragonal phase transition , but a richer phase diagram comprising the striped - nematic and the ising nematic - tetragonal transitions .
this new scenario has been identified , in particular , for @xmath21 , corresponding to stripes of width @xmath0 in the ground state . in this paper , we perform extensive mc simulations to study the phase transitions involved in this model for @xmath21 .
we present convincing numerical evidence for an ising nematic phase for this coupling and discuss the calculation of the critical exponents in face of the constraint related to lattice sizes necessary for observation of the true phenomenology of this model .
our study also allows us to figure out how reliable the simulations performed with a local update algorithm are . using histogram reweighting techniques @xcite and the patching procedure @xcite to combine data obtained from simulations at various temperatures ,
we evaluate the specific heat and the order parameter susceptibility over a wide range of temperatures , and the free - energy barrier associated with the coexisting phases .
this analysis procedure enables us to explore the existence of distinct maxima as a continuous function of the temperature and thus , to estimate the finite - size critical temperatures with high precision .
our results show that simulations need to be performed for lattice sizes @xmath22 at least as large as @xmath23 so that the ising nematic phase can be clearly observed . consequently
, the conclusions presented in the literature for smaller lattice sizes can be misleading .
moreover , to ensure the correct determination of the physical quantities and their error bars , one needs to have a reasonable number of independent measurements .
thus , to access how the long - range dipolar interaction may affect physical estimates , we also evaluate the integrated autocorrelation time in energy time series @xcite . in sec .
2 , we briefly review the histogram reweighting and patching techniques which we use to estimate average energies , orientational order parameter and their response functions , specific heat and susceptibility , as a function of temperature .
formation of the nematic phase is clearly exhibited as we increase the lattice size . in all cases , we estimate the specific heat maxima and finite - size critical temperatures by including jackknife error estimates in the analysis @xcite .
our histograms for the energy distributions present double - peak structure leading to the existence of domain walls between the states of each phase .
thus , we also evaluate the free - energy barriers at both phase transitions .
the integrated autocorrelation time analysis is presented in sec . 3 , which is followed by a brief outlook and our conclusions in sec .
we consider the two - dimensional system described by the hamiltonian in eq .
( [ eq : hamiltonian ] ) , where the spin variables @xmath24 are assumed to be aligned perpendicular to the square lattice with @xmath25 sites .
we have used the standard metropolis algorithm to generate configurations for lattice sizes @xmath26 and @xmath27 , with @xmath21 .
we carried out our simulations with periodic boundary conditions to minimize border effects due to the dipolar term in the hamiltonian .
thus , all distances @xmath6 must include sites in the infinitely replicated simulation box in both directions .
the boundary conditions add an infinite sum over all images of the simulation box to the dipole term of the hamiltonian .
we use ewald summation technique to compute this infinite sum .
this technique splits the infinite sum over all images of the system into two quickly converging sums : the direct sum , which is evaluated in the real space , the reciprocal sum , carried out in the reciprocal space , and a self - interaction correction term @xcite .
we have set the ewald parameter @xmath28 to 3.5 in all simulations .
this parameter determines the rate of convergence between the two sums .
our mc markov chains are initialized with random starts . for thermalization ,
@xmath29 sweeps were discarded for @xmath30 , and @xmath31 sweeps for @xmath32 and 72 .
our checks have shown that after @xmath33 sweeps the system does not present considerable changes in the energy .
a detailed analysis about the time evaluation can be found in ref .
for each temperature , measurements rely on @xmath34 sweeps for @xmath35 .
our largest lattice size @xmath36 relies only on @xmath37 sweeps because even a 10 times larger simulation would not present a reasonable number of independent measurements in face of our estimates from the integrated autocorrelation time calculations for smaller lattice sizes depicted in fig .
however , the patching procedure attenuates this drawback because we simultaneously match simulations at different temperatures @xmath38 to obtain the final physical estimates . .
data produced at temperatures @xmath38 and used for patching and reweighting . @xmath39 : finite - size critical temperatures for the phase transitions observed for system sizes @xmath22 defined by the maximum of specific heat @xmath40 as shown in figures 1(a ) and 1(b ) . [ cols="<,<,<,<,<,<",options="header " , ] the characterization of the order of the observed thermodynamic phase transitions can be achieved by means of fss analysis . in this context , the specific heat peak is described by the relation , @xmath41 the fss relation for the peak in the susceptibility is given by @xmath42 the first - order character is related to the critical exponent @xmath43 @xcite . from the hyperscaling relation @xmath44 , and assuming a first - order transition , one obtains @xmath45 and @xmath46 . since the phenomenology of this model reveals itself only for rather large lattice sizes , our data still present limitations for a reliable fss analysis . as we are going to show in the next section , any intensive mc simulations with local update algorithms will suffer severe limitations .
however , our data allows one to draw some conclusions .
first of all , let us verify the possible scenario found by the fss analysis . to this end
, we must only consider the last four results for the maximum of @xmath47 in table 1 and the maximum of @xmath48 in table 2 , corresponding to @xmath49 and 72 .
since the peaks for both specific heat and susceptibility related to the first transition seem to converge toward a constant value , we may assume @xmath50 and @xmath51 .
in this case , logarithmic corrections should be taken into account to describe their maximum .
the supposed values for the critical exponents do not support the first - order transition character . if @xmath47 does not present a divergent behavior , nor has appreciable finite - size effects , then the hypothesis of a kosterlitz - thouless type transition must be considered @xcite .
the hypothesis of a kt type transition has been discussed as an alternative to the possibility of the first - order character found from mc studies in ref .
@xcite . here , in spite of the high precision values for the physical estimates
, our results for the temperature and response functions may be biased when one considers estimates from the two largest lattice sizes , mainly the @xmath27 case , hampering further conclusions . in particular , the lower precision in the specific heat at @xmath52 when compared with
the results for the second peak at @xmath53 is related to strong autocorrelations in the striped - nematic transition , as presented in the next section .
the fss analysis for the second transition yields @xmath54 , @xmath55 and @xmath56 for @xmath57 .
these results clearly support the second order character for the nematic - tetragonal phase transition as first suggested theoretically by abanov @xcite , via the continuous model , and by pighn and cannas @xcite from a mean field approach .
however , we are uneasy about these numerical findings because of their very small godness - of - fit . for @xmath27 ,
in particular , we may argue that @xmath47 evaluation presents significant bias since its value does not follow the main trend , leading the scaling relation to an unlikely fit
. however , this picture does not change if we discard @xmath27 .
we obtain roughly the same numerical estimates for the exponents because @xmath58 has comparatively large error bar .
now , we will analyse the energy density distributions and , as explained below , we include the calculation of free energies and interface tensions , in our case linear tensions , to gather further information about the possible nature of the phase transitions . in fig .
5 we show energy density distributions from data obtained at temperatures @xmath38 . here
, we exhibit only energy distributions from simulations at temperatures close to the critical ones . in fig .
5(d ) we include two distributions for each lattice size because none of the sampled temperatures are close enough to the critical ones to display their true critical distributions .
figure 6 shows all histograms of energy distributions for @xmath59 .
let us initially consider the first transition .
figures 5(a ) and ( c ) present histograms with clear double - peak structures close to the critical temperatures @xmath52 .
this is a typical first - order phase transition behavior where a latent heat is responsible for the existence of those peaks at energies @xmath60 and @xmath61 , corresponding to the striped and nematic phases , respectively .
those states are separated by a minimum at @xmath62 corresponding to domains describing the coexistence of both phases . here
, we can evaluate the free energy barrier @xmath63 for this temperature - driven transition .
although it seems clear we face a first - order transition , it remains to be seen whether the linear tension is not negligible in the thermodynamic limit . to this end
, we evaluate the free energy @xcite @xmath64 at @xmath65 , where the finite lattice critical point @xmath66 is defined such that the histogram peaks have equal heights @xmath67 .
we proceed performing patching and reweighting of histograms @xcite , in order to evaluate @xmath63 at both transitions .
this leads to the results in fig .
7 for both transitions . in this figure ,
the solid and dashed lines are the linear extrapolation in @xmath68 , @xmath69 where @xmath70 is the linear tension and @xmath71 stands for possible fss correction . here
, we assume the two - gaussian approximation @xcite to describe the coexisting phases in this model . this procedure yields @xmath72 and @xmath73 for the striped - nematic transition
. obtained from simulations at temperatures @xmath38.,scaledwidth=80.0% ] now , looking at the normalized energy density distributions for the second transition , fig .
5(b ) and ( d ) , we also observe the characteristic double - peak behavior although the histograms in fig .
5(b ) have less pronounced peaks .
we observe higher peaks for @xmath59 and 72 after reweighting to the critical point where @xmath74 .
for this transition one obtains @xmath75 and @xmath76 .
the second transition is clearly weaker compared with the first one .
just to place some comparative figures to clarify the meaning of these numbers , one has @xmath77 for 2d potts model with 10 states and @xmath78 for this model with 7 states .
these numbers are exact results .
the first - order character of the model with 10 states is easily identified and is sufficient to work with lattice sizes up to @xmath79 .
it is well known that this size is too small for such identification in the case of @xmath80 .
long - range interactions always impose strong finite - size effects .
although our analysis procedure based on reweighting techniques with multiple histograms extracts the maximum information available in the mc data , some aspects still prevent us from definitive conclusions about the nature of the phase transitions . in all cases
, the bulk correlation length @xmath81 plays an important role .
actually , @xmath81 and the extension of finite systems @xmath22 are the only relevant length scales .
the validity of the fss ansatz requires that @xmath82 .
the first transition seems to be of kt type or even of first - order nature . for a kt type transition
, it is predicted that @xmath81 has the following behavior @xcite at @xmath83 @xmath84 where @xmath85 is the reduced temperature . for first - order transitions
@xmath81 remains finite and for second - order ones it behaves as @xmath86 .
thus , at @xmath83 the correlation length diverges exponentially , while divergence is the power law in a second - order phase transition .
unfortunately , a dedicated work would be required to clarify such scenario @xcite .
if present , the kt type transition would be responsible for eliminating the long - range order in the striped phase through the formation of unbound dislocations @xcite .
the reason for considering a kt type transition is the absence of divergences or significant finite - size effects in the response functions .
however , this absence may be justified by the enormous difficulty in sampling configurations at this transition , as revealed later in our analysis of integrated autocorrelation times . for the striped - nematic ( ) and nematic - tetragonal ( - - - ) transitions.,scaledwidth=80.0% ] our results from the interface tension analysis for the nematic - tetragonal transition indicate a very weak first - order phase transition .
this transition has been described as second - order from the fss analysis of the response functions specific heat and susceptibility .
therefore , the first - order character exhibited by the histograms for the larger lattice sizes @xmath59 and 72 does not seem to be caught by this fss analysis .
this implies that the fluctuations in the energy and order parameter are somehow damped in our data .
this may result from the small lattice sizes we work with , where the true first - order character is not effective and , presumably , combined with the need of a larger number of independent configurations in both phases . actually , according to our results the effect of the free - energy barrier should be more severe between the striped and nematic phases . in turn
, reliable calculation of the free - energy barriers requires enough configurations at @xmath87 that are exponentially suppressed by the boltzmann factor in canonical mc simulations .
local mc updates are not efficient , and the produced data correspond to successive states of a markov chain . this chain may be highly correlated , introducing bias to the estimates of the physical quantities and unreliable variances if the sampling is not large enough .
also , related to systems with long - range interactions , the presence of a very slow approach to equilibrium introduces a further degree of complexity .
this has been highlighted in a recent study by cannas _
@xcite on the dipolar model .
they found stronger metastabilities associated with the striped - nematic transition compared with the nematic - tetragonal one . here
, we complement our study by evaluating the integrated autocorrelation time , to assert the effectiveness of the collected data at each temperature @xmath38 with local update algorithms . to this end
, we define the normalized autocorrelation function for the energy time series @xcite , @xmath88 where @xmath89 is the energy standard deviation .
the integrated autocorrelation time @xmath90 , @xmath91 estimates the number of independent data points in a long sequence of @xmath92 mc measurements . actually , twice this value is the proper quantity to effectively evaluate this independence .
it is usually expected that the autocorrelation function decays as a power law . however ,
this behavior is changed by a temperature dependent factor near critical points @xcite .
our calculations of @xmath93 are shown in fig . 8 for lattice sizes @xmath94 and @xmath95 , where we have introduced the notation @xmath96 to describe the sample lengths , @xmath97 .
we obtain @xmath98 measurements ( @xmath99 ) for temperatures close to the finite - size critical point @xmath100 and smaller values for temperatures in the tetragonal phase . for the @xmath101 case ,
8(b ) shows a larger value : @xmath102 for @xmath103 in the striped phase and @xmath104 for @xmath105 , in the vicinity of the specific heat maximum .
again , as expected , smaller values are observed for samples in the tetragonal phase .
figure 8(c ) displays the results for @xmath23 , whose specific heat data present two critical temperatures . for the temperature inside the striped phase ( @xmath106 ) ,
one obtains a larger integrated autocorrelation time compared with the one achieved at the striped - nematic transition @xmath107 . in the nematic phase , one obtains @xmath108 . by increasing the temperature toward the next critical point @xmath109 ,
one obtains @xmath110 . as the temperature increases further ,
it is observed a strong decrease in @xmath111 , reaching values as small as 45 for @xmath112 .
figure 8(d ) shows huge numbers for this autocorrelation time , larger than @xmath113 for data collected in the striped and at the striped - ising nematic transition . in the ising nematic phase
, one has @xmath114 . at the ising nematic - tetragonal transition one
obtains @xmath115 , and a smaller value 60 at the temperature @xmath116 is achieved in the tetragonal phase . here , we could try to describe autocorrelation times for the energy observable as a simple power law @xmath117 , where @xmath118 is the associated dynamic critical exponent .
a rough estimate gives @xmath119 and @xmath120 for the striped - nematic and nematic - tetragonal phase transitions , respectively .
these are large numbers since local mc algorithms generally have a dynamic critical exponent @xmath121 @xcite . for non - local update algorithms , such as the swendsen - wang cluster algorithm , this exponent
is further reduced to @xmath122 in the pure 2d ising model . from the autocorrelation time analysis
, we may infer the existence of very long - lived states near and at the striped - nematic transition .
these states may lead to stronger bias in the determination of the response functions at this transition than at the nematic - tetragonal transition , where @xmath90 are orders of magnitude smaller .
our systematic analysis by reweighting techniques in multiple histograms describes the existence of two phase transitions , which correspond to a two - step melting process leading to a disordered state .
this long - range interaction model is described by a complex phase diagram with lines corresponding to striped - nematic and nematic - tetragonal phase transitions , depending on the coupling @xmath10 .
we have seen that the true phenomenology of this model can be observed for large lattice sizes only , and that the controversial nematic phase is detected in a narrow range of temperatures .
as @xmath10 increases , one observes larger stripe width .
as realized for @xmath0 , we also expect that for larger values of @xmath11 , it will be necessary to increase further the system size to obtain reliable results .
this can be explained by the need of inserting dislocations in the striped - domain structure .
the formation of bound dislocation pairs proliferates as the temperature increases , leading to a phase transition .
our numerical results clearly show the locations of the phase transitions , which are strongly lattice size dependent .
moreover , the response functions also have strong finite - size effects , which frustrate any convincing evaluation of the critical exponents from simple fss analysis .
the histogram analysis has shown a strong first - order character for the striped - nematic transition . however
, this character may introduce strong metastabilities , leading to bias in our simulation .
this is because the sampling may spend most of the time in one of two phases , mainly in the case of our largest lattice sizes .
according to our integrated autocorrelation time calculations , we have learnt that to obtain @xmath123 independent configurations for @xmath27 , about @xmath124 and @xmath125 mc sweeps are required for the striped - nematic and nematic - tetragonal transitions , respectively .
we have attenuated this limited sampling by matching simulations at different temperatures but close to each other .
however , it is not just a matter of how long one needs to perform the simulations .
our analysis has also revealed the need for larger lattice sizes , so that the character of the phase transitions is exposed .
it is likely that the lattice sizes are not large enough compared to the correlation lengths .
this because our critical exponent estimates from fss indicate continuous transitions , in contrast with the results for the free - energy barriers .
the asymptotic behaviors of the free - energy barriers between domains indicate that the interface tensions do not vanish in the thermodynamic limit .
another controversial point still remains , the possibility of a kt type transition for the nematic - tetragonal transition .
our results for the response functions do not discard this possibility .
the integrated autocorrelation time calculations show stronger critical slowing down in both phase transitions when compared to the pure 2d ising model . from this analysis
we may conclude that there are very long - lived states near and at the striped - nematic phase transition as opposed to the nematic - tetragonal transition , where @xmath90 present smaller orders of magnitude .
these results are very important to figure out how reliable the mc sampling for this system with long - range interaction is .
moreover , the use of important techniques for mc data analysis establishes the limits one can reach with local update algorithms .
more efficient results to obtain a deeper understanding of the complex striped - nematic - tetragonal phase transitions seem possible with a variant of wollf cluster algorithm @xcite or by means of generalized ensemble methods .
the authors acknowledge support by fapesp and capes ( brazil ) .
k. debell k , a. b. macisaac and j. p. whitehead , _ rev . mod .
* 72 * ( 2000 ) 225 .
p. m. gleiser , f. a. tamarit , s. a. cannas and m. a. montemurro , _ phys .
_ b * 68 * ( 2003 ) 134401 . s. a. pighn and s. a. cannas , _ phys .
_ b * 75 * ( 2007 ) 224433 .
e. rastelli , s. regina and a. tassi , _ phys .
_ b * 76 * ( 2007 ) 054438 .
a. b. macisaac , j. p. whitehead , m. c. robinson and k. debell , _ phys .
_ b * 51 * ( 1995 ) 16033 .
a. giuliani , j. l. lebowitz and e. h. lieb , _ phys .
_ b * 74 * ( 2006 ) 064420 ; _ phys .
_ b * 76 * ( 2007 ) 184426 .
o. portmann , a. vaterlaus and d. pescia , _ nature _ * 422 * ( 2003 ) 701 ; _ phys .
* 96 * ( 2006 ) 047212 .
m. biskup , l. chayes and s. a. kivelson , _ commun .
* 274 * ( 2007 ) 217 .
p. m. gleiser , f. a. tamarit and s. a. cannas , _ physica _ d * 168 - 169 * ( 2002 ) 73 . i. booth , a. b. macisaac , j. p. whitehead and k. debell , _ phys .
* 75 * ( 1995 ) 950 .
s. a. cannas , d. a. stariolo and f. a. tamarit , _ phys .
_ b * 69 * ( 2004 ) 092409 .
m. casartelli , l. dallasta , e. rastelli and s. regina , _ j. phys . a : math .
* 37 * ( 2004 ) 11731 .
j. arlett , j. p. whitehead , a. b. macisaac and k. debell , _ phys .
_ b * 54 * ( 1996 ) 3394 .
e. rastelli , s. regina s and a. tassi , _ phys .
_ b * 73 * ( 2006 ) 144418 .
a. abanov , v. kalatsky , v. l. pokrovsky and w. m. saslow , _ phys . rev .
b * 51 * ( 1995 ) 1023 . s. a. cannas , m. f. michelon , d. a. stariolo and f. a. tamarit , _ phys .
_ b * 73 * ( 2006 ) 184425 .
a. m. ferrenberg , r. h. swendsen , _ phys .
lett . _ * 61 * ( 1988 ) 2635 ; _ phys .
_ * 63 * ( 1989 ) 1195 .
n. a. alves , b. a. berg and s. sanielevici , _ nucl .
_ b * 376 * ( 1992 ) 218 .
a. d. sokal , _ monte carlo methods in statistical physics : foundations and new algorithms _ , lecture notes of the cours de troisime cycle de la physique en suisse romande , lausanne 1989 .
b. efron , _ siam rev .
_ * 21 * ( 1979 ) 460 ; b. efron and r. j. tibshirani , _ an introduction to the bootstrap _ , monographs on statistics and applied probability 57 ( chapman & hall , 1993 ) .
a. y. toukmaji and j. a. board jr .
, _ comput .
* 95 * ( 1996 ) 73 .
g. t. gao , x. c. zeng and w. wang , _ j. chem .
* 106 * ( 1996 ) 3311 .
s. a. cannas , m. f. michelon , d. a. stariolo and f. a. tamarit , _ phys .
_ e * 78 * ( 2008 ) 051602 .
n. schultka and e. manousakis , _ phys .
_ b * 49 * ( 1994 ) 12071 .
j. e. van himbergen and s. chakravarty , _ phys .
_ b * 23 * ( 1981 ) 359 ; j. e. van himbergen , _ phys . rev _
b * 25 * ( 1982 ) 5977 .
m. e. fisher and a. n. berker , _ phys .
_ b * 26 * ( 1982 ) 2507 .
k. decker , a. hasenfratz and p. hasenfratz , _ nucl .
_ b * 295 * ( 1988 ) [ fs21 ] 21 .
k. binder , _ phys .
rev . _ a * 25 * ( 1982 ) 1699 .
j. lee and j. m. kosterlitz , _ phys .
* 65 * ( 1990 ) 137 ; ibid . _ phys . rev .
_ b * 43 * ( 1991 ) 1268 ; ibid .
_ b * 43 * ( 1991)3265 .
n. alves , _ phys .
d * 46 * ( 1992 ) 3678 . m. s. s. challa , d. p. landau and k. binder , _ phys .
b * 34 * ( 1986 ) 1841 . j. m. kosterlitz , _ j. phys .
_ c * 7 * ( 1974 ) 1046 .
r. gupta and c. f. baillie , _ phys .
_ b * 45 * ( 1992 ) 2883 .
g. ossola and a. d. sokal , _ nucl .
_ b * 691 * ( 2004 ) [ fs ] 259 .
e. luijten and h.w.j .
bl@xmath126te , _ int
_ c * 6 * ( 1995 ) 359 ; ibid .
_ b * 56 * ( 1997 ) 8945 . | the two - dimensional ising model with nearest - neighbor ferromagnetic and long - range dipolar interactions exhibits a rich phase diagram .
the presence of the dipolar interaction changes the ferromagnetic ground state expected for the pure ising model to a series of striped phases as a function of the interaction strengths .
monte carlo simulations and histogram reweighting techniques applied to multiple histograms are performed to identify the critical temperatures for the phase transitions taking place for stripes of width @xmath0 on square lattices . in particular
, we aim to study the intermediate nematic phase , which is observed for large lattice sizes only . for these lattice sizes ,
we calculate the critical temperatures for the striped - nematic and nematic - tetragonal transitions , critical exponents , and the bulk free - energy barrier associated with the coexisting phases .
we also evaluate the long - term correlations in our time series near the finite - size critical points by studying the integrated autocorrelation time @xmath1 as a function of the lattice size .
this allows us to infer how severe the critical slowing down for this system with long - range interaction and nearby thermodynamic phase transitions is . |
we reconsider the two - dimensional @xmath0 model with random phase shifts and a symmetry - breaking field , which is described by the reduced hamiltonian @xmath1 @xmath0 spins @xmath2 are placed on the sites @xmath3 of a square lattice .
the coupling of nearest neighbors with a spin stiffness @xmath4 favors a relative angle prescribed by the quenched random phase shifts @xmath5 , which we assume to be gaussian distributed with zero mean and variance @xmath6 .
the global rotation symmetry of spins is broken by the field @xmath7 that selects @xmath8 equivalent favorable directions for the spins .
the model will be considered with uniform fields ( @xmath9 ) and with random local orientations @xmath10 of the field .
this model describes a variety of physical systems , ranging from disordered magnets@xcite to josephson - junction arrays@xcite and planar josephson - junctions.@xcite it is also closely related to surfaces of crystals with bulk disorder@xcite and even to periodic elastic media such as vortex lattices in the presence of a pinning potential,@xcite two - dimensional crystals with quenched impurities@xcite or adsorbates on disordered substrates.@xcite the study of the above and related models started already decades ago .
we briefly recall some marked steps in its history to provide the background for the present work .
the @xmath0 model in the absence of disorder and magnetic fields exhibits the kosterlitz - thouless ( kt ) transition from a phase with quasi - long - range order ( qlro , i.e. , algebraic decay of the spin - spin correlation ) to short - range order ( sro , i.e. , exponential decay of the spin - spin correlation ) at @xmath11,@xcite where the effective spin stiffness @xmath12 is renormalized by thermal fluctuations .
the transition was identified with an unbinding of vortex pairs .
uniform _ fields have been included by jos , kadanoff , kirkpatrick , and nelson.@xcite for @xmath13 they found that the phase with qlro is stable against such fields over a temperature range @xmath14 and that weak fields become relevant at low temperatures @xmath15 where they stabilize long - range order ( lro , i.e. finite magnetization ) .
for @xmath16 the transition resulting from the competition between the ordering tendency of the magnetic field and the disordering tendency of the thermal fluctuations could not be captured by a kt - like approach .
houghton , kenway , and ying@xcite found the phase with qlro to be unstable to magnetic fields with _ random _ orientation only for @xmath17 .
accordingly , a phase with qlro can be stable within a finite temperature range @xmath18 provided @xmath19 .
their study was extended by cardy and ostlund@xcite ( co ) to address the nature of the phase transitions and of the low - temperature phase .
their analysis implied that randomness of the direction of the magnetic field effectively generates a new type of disorder on large scales ( in the sense of a renormalization group ) , namely random phase shifts @xmath5.@xcite rubinstein , shraiman , and nelson@xcite ( rsn ) examined the @xmath0 model with random phase shifts in the absence of fields .
they found qlro to be stable against disorder below a certain critical strength and a phase diagram with reentrant shape .
paczuski and kardar@xcite ( pk ) added uniform fields and obtained a phase diagram composed of various phases with lro , qlro , and sro .
the transitions between these phases showed multiple reentrance .
recent advance in the understanding of the bond - disordered model without fields has led to a revised nonreentrant phase diagram , where the phase with qlro turned out to be more stable than found by rsn.@xcite the earlier underestimation of the order in the system was caused by a subtle _ breakdown of the fugacity expansion _ in the coulomb gas representation of the model at low temperatures.@xcite this breakdown was overcome in these recent approaches by nonperturbative techniques .
the features of the revised phase diagram have been confirmed by numerical simulations.@xcite the breakdown of the fugacity expansion is likely to affect _ all _ problems involving random phase shifts . in particular
one has to ask how the structure of the phase diagram , the nature of the phases and the nature of the phase transitions are modified .
it is also important to examine whether a similar breakdown occurs in a perturbative treatment of the fields .
the purpose of the present work is to reexamine model ( [ h ] ) in view of these questions .
the method of our choice is a mapping of model ( [ h ] ) on two coupled coulomb gases , `` vortices '' and `` charges '' ( sec .
thereby we take a pre - averaging over disorder using the replica trick .
we establish renormalization group flow equations for the model parameters to capture the large scale properties of the system ( sec .
iii ) . from the evaluation of these flow equations ( sec .
iv ) we obtain results that are significantly different from those obtained by co and pk for the special cases of ferromagnetic bonds with random fields and bond disorder with uniform fields .
our phase diagrams are illustrated in fig .
[ fig.schem ] for these two previously examined special cases .
we conclude with a discussion of our results and of consequences for physical realizations of the @xmath0 model ( sec .
appendices with some calculational details and a dictionary for a translation of notations from previous works to the present work are included .
in this section we specify the stochastic properties of the random phase shifts considered subsequently .
in addition , we transfer randomness in the orientation of the symmetry - breaking fields to the random phase shifts such that we need to deal only with ( correlated ) bond disorder . the latter model is then mapped onto two coupled coulomb gases .
eventually we arrive at a replica representation of the model that serves as basis for our further analysis . in order to be able to treat the model irrespectively of a randomness in the local field orientation ,
we decompose the field @xmath20 in fourier space into its longitudinal and transverse component , [ def.a ] @xmath21 both components ( polarizations @xmath22 ) are supposed to be gaussian distributed with zero average and to have variances ( over - lining denotes averaging over quenched disorder ) @xmath23 where we explicitly allow for @xmath24 .
then the random phase shifts have a variance @xmath25 and are in general correlated in space .
only in the special case @xmath26 the phase shifts are uncorrelated on different bonds of the lattice .
neglecting the effect of vortices and fields completely , the spin - spin correlation ( angular brackets denote a thermal average for a fixed disorder realization ) @xmath27 \rangle } \label{def.gamma}\ ] ] can be calculated in the spin - wave approximation and is found to decay algebraically , @xmath28 with an exponent @xmath29 . in this approximation @xmath30 depends only on the _ longitudinal _ disorder . as in ref .
we will also examine correlations of the edwards - anderson order parameter [ ea.corr ] @xmath31 between different replicas
@xmath32 of the system . in the spin - wave approximation
the exponent @xmath33 is completely independent of disorder.@xcite a convenient way of dealing with the randomness in the local field orientation @xmath10 is to eliminate the angles @xmath10 in favor of modified phase shifts @xmath34 by a substitution @xmath35 .
this transformation leaves the thermodynamic partition sum invariant , i.e. , @xmath36 therefore , the existence and location of phase transitions in the phase diagram are not affected by this manipulation .
in contrast , spin correlation functions are _ not _ invariant and can display qualitatively different large - scale behavior before and after the transformation . in order to restrict our analysis of random phase shifts to those with correlations ( [ corr.a ] ) even after the transformation , we limit the possible distributions of the angles @xmath10 to those having a gaussian distribution with correlations @xmath37 only such random orientations are transformed by the above mapping into longitudinal phase shifts with a variance @xmath38 equation ( [ corr.alpha ] ) implies correlations @xmath39 for @xmath40 in real space
. however , in the limit @xmath41 the relative fluctuations even of neighboring angles are that large that they can be considered practically as uncorrelated since they enter the hamiltonian only modulo @xmath42 . in this sense
the examination of the model ( [ h ] ) with random phase shifts and uniform field effectively includes the model with random symmetry - breaking fields in the limit @xmath43 .
we apply standard techniques@xcite to map model ( [ h ] ) onto a coulomb gas . for this purpose
we first use the villain approximation@xcite to replace both cosines in eq .
( [ h ] ) by a slightly different periodic potential and obtain an effective hamiltonian @xmath44 the partition sum of this effective model contains not only the integration over the angles @xmath45 but also a summation over the integer - valued vector field @xmath46 and the integer - valued scalar field @xmath47 .
the parameters of this effective model are related to the original parameters by @xmath48 and @xmath49 through the function @xcite @xmath50 since we are mainly interested in the stability of the phase with qlro ( i.e. , @xmath51 ) to symmetry - breaking fields ( i.e. , @xmath52 ) , we effectively operate in the limiting cases @xmath53 and @xmath54 . for small @xmath55
the sum over @xmath47 converges very slowly .
therefore it is of advantage to go over from the field @xmath47 to a conjugate integer - valued field @xmath56 by means of the poisson summation formula ( some calculational details are given in appendix [ app.cg ] ) .
after integrating out the angles @xmath45 in the partition sum we end up with a new effective hamiltonian for two coupled coulomb gases .
the first coulomb gas is the gas of _ vortices _ with a vorticity @xmath57 .
these vortices are located on the dual lattice , i.e. , on plaquette centers @xmath58.@xcite the _ `` charges '' _ @xmath59 constitute the second coulomb gas .
these coulomb gases are coupled to _
`` quenched vortices '' _ @xmath60 and _ `` quenched charges '' _ @xmath61 , which are generated by the transverse and longitudinal components of the bond disorder respectively .
the hamiltonian of the effective model reads [ h.rs ] @xmath62 it has been split into the interaction among vortices @xmath63 , the interaction among charges @xmath64 , the interaction between vortices and charges @xmath65 , and an excess term @xmath66 .
particles of the same type interact logarithmically and the coupling between the two gases is given by the angle @xmath67 enclosed by their relative distance and an arbitrary reference direction in the @xmath68 plane ( see appendix [ app.cg ] for some intermediate steps ) . due to the introduction of the charges by means of the poisson summation formula there are two imaginary - valued contributions to the hamiltonian : the coupling of thermal charges @xmath56 to quenched charges @xmath69 and the coupling of vortices @xmath70 to charges @xmath56 . however , the partition sum is real - valued since the imaginary parts cancel in pairs @xmath71
+ \exp[-{\cal h } ( \{n\ } , \{-n\})]$ ] .
both gases are dilute for @xmath72 and @xmath73 , where the core energies @xmath74 are large ( @xmath75 is a constant of order unity@xcite emerging from the lattice green function ) .
although these core energies are determined by @xmath76 and @xmath55 we will consider them formally as independent parameters in the subsequent treatment .
the partition function of this effective model ( [ h.rs ] ) reads @xmath77 the primed summation is constrained by neutrality conditions @xmath78 since the real part of the energy of nonneutral configurations diverges logarithmically with the system size .
the direct analysis of the partition sum ( [ z.rs ] ) in the spirit of the kt renormalization group ( rg ) is prevented by the broken translation invariance in the presence of the quenched vortices and charges .
since we are aiming at the calculation of disorder - averaged quantities , such as the correlation function ( [ def.gamma ] ) , we may perform the disorder average on the replicated partition sum already at this stage.@xcite for an integer number of replicas @xmath56 the hamiltonian of the replicated system is most conveniently expressed in a vector notation , where @xmath79 for @xmath80 is a replica charge at site @xmath81 and @xmath82 is a replica vortex at site @xmath58 : [ h.rep ] @xmath83 all parameters are incorporated in the replica - symmetric coupling matrices [ e.k ] @xmath84 hamiltonian ( [ h.rep ] ) is a generalization of that given in sec .
ii of co. in the limit @xmath43 we retrieve their model , i.e. , eq .
( [ h.rep.hc ] ) becomes equivalent to their eq .
( 2.15 ) . furthermore , in this limit the contribution of the charge core energy proportional to @xmath85 enforces an additional neutrality @xmath86 across the replicas , i.e. , the weight of nonneutral configurations vanishes in the partition sum .
then the vector @xmath87 can be decomposed into pairs of type @xmath88 at every site .
such pairs correspond to the vector charges of co. however , for finite @xmath85 the summation over charges is _ not _ restricted by this additional neutrality condition .
note that eq .
( [ h.rep ] ) contains an _ explicit _ coupling between vortices and charges .
the effective model under consideration is self - dual,@xcite i.e. , it is invariant under a mutual exchange of vortices and charges with a simultaneous substitution of parameters . for the unreplicated model ( [ h.rs ] )
these substitutions are @xmath89 this duality implies a relation @xmath90 for the partition sum,@xcite which simplifies our subsequent calculations because it enables us to relate renormalization effects arising from vortex fluctuations to renormalization effects arising from charge fluctuations or vice versa . in this way pk obtained recursion relations for the bond - disordered model _ with _ field from those of rsn _ without _ field .
we are going to use the same strategy starting from the corrected flow equations.@xcite we will proceed within the replica formulation ( [ h.rep ] ) , where the self - duality of the model is reflected by the substitutions @xmath91 the excess contribution ( [ h.x ] ) to the hamiltonian of the unreplicated system effectively modifies the action of the longitudinal disorder distribution .
the duality @xmath92 implies @xmath93 .
the latter relation is consistent with eq .
( [ dual ] ) if one keeps in mind the presence of the excess term in eq .
( [ dual.z ] ) .
this term gives rise to the denominator @xmath94 that becomes unity in the limit @xmath95 that eventually will be taken .
after the coulomb gas representation of our model has been established in the previous section we now turn to a scaling analysis for the relevance of vortices and charges .
we basically follow kosterlitz rg approach @xcite to the @xmath0 model and its recent generalization@xcite to the @xmath0 model with random phase shifts . to account for magnetic fields we have to include charges and their coupling to the vortices . following ref .
we introduce the notion of `` _ types _ '' of replica vortices and charges .
we consider charges at site @xmath96 and site @xmath97 as being of the same or of different type if @xmath98 or @xmath99 in analogy to the vortex types . from the core energy contribution to the hamiltonian ( [ h.rep ] ) we identify fugacities [ fuga ] @xmath100 for each type of particles . because of the replica symmetry of the matrices @xmath101 and @xmath102 different particle types have the same fugacity if they belong to the same `` _ class _
, '' that is , if they are of the same type after a permutation of replicas and/or an inversion of the replica vector ( such as @xmath103 ) . in the limit of vanishing field
all charge fugacities ( [ fuga]b ) go to zero .
the relevance of a weak field can be examined by looking at the scaling behavior of the charge fugacities .
as soon as the fugacity of some charge type increases under rescaling , the field constitutes a relevant perturbation .
analogously , the relevance of vortices signals the importance of topologically nontrivial spin configurations .
however , while a relevance of vortices indicates a tendency to reduced spin correlations on large scales , a relevance of charges expresses a tendency to enhanced spin correlations .
a rescaling of lengths @xmath104 and @xmath105 with an infinitesimal increase on a logarithmic length scale @xmath106 has the following effect on hamiltonian ( [ h.rep ] ) : the angles @xmath107 are invariant but the logarithms generate additive constants , @xmath108 .
using the neutrality conditions ( [ neutr ] ) the twofold spatial summations over these constants can be reduced to single summations .
then the rescaling effectively amounts to an additive flow of the core matrices @xmath109 , or equivalently to a multiplicative flow of the fugacities @xmath110 for a finite number @xmath56 of replicas we can now determine the stability of qlro to vortices and charges at given temperature @xmath111 and disorder strengths @xmath112 by looking for the most relevant class of particles .
we focus here on the charge sector , since the vortex sector has been covered in ref . .
the _ most relevant _ charge class is the one that minimizes @xmath113.\ ] ] the minimum is provided by charges of class @xmath114 for @xmath115 and by charges of class @xmath116 for @xmath117 .
the class `` @xmath114 '' is composed of all replica charges @xmath87 that contain an elementary charge in a single replica [ e.g. , @xmath118 and the class `` @xmath116 '' is composed of all replica charges @xmath119 that contain an elementary charge in one replica and an opposite elementary charge in a second replica [ i.e. , the co vector charges , such as @xmath120 .
let us consider for illustration vector charges that are composed only of @xmath121 elementary charges , thereof @xmath122 are positive and @xmath123 are negative .
the fugacity of such charges scales like @xmath124 \frac{p^2}{4 \pi}\right]\right\ } .\ ] ] figure [ fig.relev ] displays the stability boundary in the ( @xmath125 ) plane for various classes of charges , which are irrelevant on the high - temperature side of the corresponding ( full ) lines .
charge class @xmath114 is most relevant on the high - temperature side ( @xmath126 ) and charge class @xmath127 is most relevant on the low temperature side ( @xmath128 ) of the dashed line .
there is an important difference in the scaling analysis for charges and for vortices .
the region of the phase diagram , where _ all _ @xmath56-component charges are irrelevant , is _ independent _ of the number @xmath56 of replicas .
in addition , the boundary of this region is invariantly given for _ all _ integer @xmath56 by the charge classes @xmath114 or @xmath116 ( strictly speaking , the class @xmath116 exists only for @xmath129 ) .
this independence on @xmath56 indicates that the replica limit @xmath130 can be taken in a very simple way .
in contrast , the region where all vortices are irrelevant depends explicitly on @xmath56 and so does the class of the most relevant vortices at lower temperatures ( see fig . 2 of ref . ) .
therefore the replica limit will be easier to take for charges than for vortices .
the save proceeding for both cases is to consider the collective renormalization of the physical parameters by all classes of vortices and charges , which will be carried out subsequently .
we anticipate here that through this proceeding we obtain first closed expressions for the collective effects of charges , for which the perturbative expansion in small charge fugacities can be justified _
a posteriori_. the various terms in this expansion can again be identified with vector charge classes .
the leading terms correspond to the most relevant classes identified above and the stability region ( shaded area ) of fig .
[ fig.relev ] remains unchanged .
the renormalization of parameters on large scales due to fluctuations on small scales will now be calculated by integrating out pairs of particles from the partition sum following the approach of kosterlitz.@xcite since our model possesses the vortex - charge duality , it is sufficient to perform the actual calculations only in one particle sector .
the renormalization effects of the other sector are then obtained by a duality transformation .
we adopt the results for the vortex sector directly from ref . . however , we extend these previous calculations at two points .
first , we calculate the flow equations for the fugacities to higher order in the fugacities .
this extension is necessary to consider the parameter renormalizations not only within the phase with qlro , but also in the phase where magnetic fields are relevant .
this requires flow equations of higher order in the charge fugacities , which we can obtain through the duality mapping only from the terms of higher order in the vortex fugacities .
second , when the vortex dipoles are integrated out in the presence of charges , they renormalize the interactions among the charges ( and vice versa ) .
this feature was naturally absent in ref . , where fields were not included . for technical convenience we allow replica vortices to take positions in the continuous space and not only on the lattice .
the lattice provides only a cutoff ( taken to be unity ) for the minimal distance between vortices .
our present goal is to integrate out all configurations in the partition sum , where the smallest distance between two replica vortices ( of any type ) lies in the range @xmath131 .
we follow young s treatment of a vector coulomb gas,@xcite which can be applied directly to our vectors @xmath132 . during this procedure
two cases have to be distinguished : ( i ) the two close vortices are of opposite type , i.e. , @xmath133 ( in the continuous space subscripts to @xmath132 denote the particle label instead of the position ) and ( ii ) these two vortices are not of opposite type , i.e. , @xmath134 . in case ( i ) , where the pair of close vortices is composed of a vortex and its antivortex , they essentially annihilate .
the leading interaction with the other particles , which are approximately considered to be far away , is through the dipole moment of the pair . because of the polarizability of this dipole moment
the pair screens the interaction between the other particles.@xcite the screening of the interaction among vortices can be captured by a flow of the coupling @xmath135 where @xmath136 is the replica vortex density correlation at the cutoff distance [ see eq .
( 13 ) of ref . ] . due to the angular coupling between vortices and charges
there is an additional screening of the interaction among charges ( see appendix [ app.screen ] ) : @xmath137 besides the screening of these coupling matrices , the annihilation of such dipoles changes the free energy per unit area and per replica by @xmath138 in case ( ii ) the two close vortices appear ( in the eyes of the other particles , which are assumed to be far away ) essentially as a single vortex of type @xmath139 , i.e. the dominant interaction with the other particles is not given by the dipole moment of the pair but by the total vorticity .
therefore we can `` replace '' the two vortices by the single one , i.e. , we transfer the statistical weight of the entire particle configuration to a different particle configuration , where the two close vortices 1 and 2 are replaced by a vortex of type 3 and all other particles are unchanged . by analogy with young s
analysis@xcite [ leading to his eq .
( 75 ) ] we obtain @xmath140 the neutrality conditions ( [ neutr ] ) are preserved under such types of recombination . in order to establish the rg flow equations , we combine the contributions from scaling and screening . using the duality relations ( [ dual.rep ] ) we infer the contributions by charge fluctuations immediately from those by vortex fluctuations that have been given above .
thus , after collecting all terms , we find the complete flow equations for the replicated system [ flow.y ] @xmath141 and [ flow.param ] @xmath142 since we establish the flow equations for an infinitesimal change of the cutoff , we assume that we can simply add vortex and charge contributions , i.e. , that there are no terms mixing vortex and charge fugacities .
such a mixing could arise in principle if one implies a minimal cutoff distance not only among vortices and among charges , but also between vortices and charges .
we are inclined to believe that a hard - core interaction between vortices and charges is not necessary , because we do not expect divergences related to short distances of that type : due to the imaginary - valued nature of angular coupling ( [ h.vc ] ) in the hamiltonian it seems plausible that configurations with close vortex - charge pairs give no essential contribution to the partition sum . since not only the original couplings @xmath143 but also the particle density correlations @xmath144 are replica - symmetric , this symmetry holds also for the screened couplings .
even more , the renormalized matrices @xmath145 and @xmath146 can be reexpressed in terms of renormalized parameters @xmath76 and @xmath112 preserving relations ( [ e.k]a , b ) : equations ( [ flow.param]a,b ) are equivalent to [ flow.par ] @xmath147.\end{aligned}\ ] ] thereby the particle correlation matrices were decomposed into the `` connected '' and `` disconnected '' contributions according to @xmath148 for both vortices and charges . so far , the derivation of the rg flow equations for the replicated system have been conceptually straightforward . the less trivial part is to take the replica limit , i.e. , to perform an analytic continuation for @xmath130 .
the first step in this direction was the reduction of the @xmath143 matrix flow equation to the parameter flow equations ( [ flow.par ] ) .
we focus on these equations , since the physical large - scale properties of our system show up in the flow of these parameters . in order to send @xmath130 in these flow equations
, we have to overcome only one hurdle : to make @xmath56 an explicit parameter in the connected and disconnected correlations . here
again we follow the route of ref . , where the problem was solved for the vortex correlations , and treat charges on the same footing .
one easily realizes from the flow equations for the fugacities that large vortices ( charges ) with @xmath149 ( @xmath150 ) are less relevant than small vortices ( charges ) with @xmath151 ( @xmath152 ) .
therefore it is a good approximation to restrict the further calculation to the small particles ( this approximation is made for simplicity , the inclusion of large particles is possible ) .
the correlations can then explicitly be evaluated after introducing a auxiliary random variable @xmath153 .
its distribution is defined to be gaussian with average @xmath154_{\cal a}=0 $ ] and variance @xmath155_{\cal a}=1/2 $ ] ( averages over @xmath153 are denoted by @xmath156_{\cal a}$ ] ) .
we decompose the core energy matrices ( [ e.k]c , d ) into @xmath157 for vortices and charges ( one specifies by adding the corresponding subscripts `` v '' or `` c '' ) .
the initial values of @xmath158 given in eq .
( [ defs.e ] ) and @xmath159 next , we introduce weights ( proportional to the fugacities @xmath160 of the elementary particles ) @xmath161 and , combining eqs .
( [ fuga ] ) with ( [ def.c ] ) and its dual counterpart , we rewrite the correlations as @xmath162_{\cal a } , \\ c^{ab}&= & - \left[[(z_+ - z_-)^2 z^{n-2 } \right]_{\cal a},\end{aligned}\ ] ] where we take @xmath32 . in this representation
the number of replicas became an explicit variable that can be sent to zero , resulting in [ c.z ] @xmath163_{\cal a } , \\
c_{\rm con}&= & - \left[\frac{z_+ + z_- + 4z_+ z_-}{(1 + z_+ + z_- ) ^2 } \right]_{\cal a}.\end{aligned}\ ] ] then the replica limit of the flow equations ( [ flow.par ] ) for @xmath76 and @xmath112 can be taken right away , since now all dependences on @xmath56 are explicit .
the contributions to the flow of the free energy ( [ flow.param]c ) can be rewritten as@xcite @xmath164_{\cal a } \to [ \ln z_{\rm v}]_{\cal a},\ ] ] where in the last expression the replica limit has already been performed .
the analogous charge contribution is obtained by duality substitutions .
combining expressions ( [ flow.param]c ) , ( [ flow.par ] ) , ( [ c.z ] ) , and ( [ f.z ] ) the flow of the physical quantities @xmath76 , @xmath112 , and @xmath165 can be expressed through the fugacities @xmath166 for both types of particles . according to eq .
( [ defs.e ] ) these fugacities are small for large @xmath76 and small @xmath55 and one might be tempted to simplify the expressions ( [ c.z ] ) and ( [ f.z ] ) by a truncated expansion in these fugacities .
the early studies of the disordered @xmath0 model@xcite made use of this fugacity expansion . the invalidity of this expansion was recognized later on and was considered as a possible reason for the destruction of qlro even by infinitesimally weak disorder.@xcite however , it was shown recently@xcite for the model without fields that the mere breakdown of the fugacity expansion does not imply the relevance of vortices and the destruction of qlro . in other terms : in the vortex sector the closed expressions ( [ c.z ] ) and ( [ f.z ] ) are _ finite _ and they are to be evaluated nonperturbatively .
therefore it is a central issue of the present work to reexamine the validity of the fugacity expansion for charges that underlies refs . . to be explicit let us examine the `` smallness '' of the charge fugacities @xmath167 defined in eq
( [ def.z ] ) .
in contrast to the real - valued vortex fugacities they are complex - valued since @xmath168 is purely imaginary . their absolute value @xmath169 is small for weak fields _ independently _ of the disorder represented by the variable @xmath153 .
therefore it is safe to expand the expressions correlations ( [ c.z ] ) for charges _ before _ performing the average over @xmath153 .
again , this is in sharp contrast to the case of vortices , where the fugacities become arbitrarily large in the tails of the distribution of @xmath170 .
thus , the mechanism that led to the failure of the fugacity expansion for vortices is _ not _ effective for charges .
thus , we are entitled to use the fugacity expansion for charges to study the stability qlro to weak fields . we expand the charge correlations ( [ c.z ] ) [ and analogously the source of the flow of the free energy , the dual counterpart of eq .
( [ f.z ] ) ] to second order in the fugacities @xmath167 and perform the average over @xmath153 : [ cocol ] @xmath171}_{\cal a } & = & 2 y_{{\rm c}\uparrow}^2 -y_{{\rm c}\uparrow \downarrow}^2 - y_{{\rm c}\uparrow \uparrow}^2 + \cdots\end{aligned}\ ] ] we have replaced the terms in the series by vector charge fugacities using the identification ( [ fuga ] ) and substituted the symbols @xmath172 by the more illustrative arrow symbols .
the initial values of these fugacities are given by [ y.c.0 ] @xmath173 } , \\
y_{{\rm c}\uparrow \downarrow } & = & e^{-[1/h + ( \gamma p^2/2 \pi ) k^{-1 } ] } , \\ y_{{\rm c}\uparrow \uparrow } & = & e^{-[1/h + ( \gamma p^2/2 \pi ) ( k^{-1 } + 2 \sigma^l ) ] } .\end{aligned}\ ] ] all bare charge fugacities are determined by only three physical parameters @xmath76 , @xmath85 , and @xmath55 .
this relation is abandoned under renormalization , where the flow of @xmath76 and @xmath85 follows eq .
( [ flow.par ] ) and where we have a whole set of flow equations ( [ flow.y ] ) for the charge fugacities instead of a single flow equation for @xmath55 .
it is interesting to note that a fugacity with @xmath121 arrows is of order @xmath174 in the original field [ remember relation ( [ def.f ] ) ] .
this order @xmath175 is just the total number of positive / negative elementary charges introduced in eq .
( [ mo.re.c ] ) above .
since the fugacity expansion is valid for the charges , it is legitimate to analyze the stability of qlro by demanding that all classes of vector charges have to be irrelevant .
this analysis has been performed in sec .
[ sec.scal ] leading to the shaded stability area in fig .
[ fig.relev ] . as long as all charges are irrelevant ,
the second order terms in the flow equation ( [ flow.y]b ) play actually no role . for
a _ qualitative _ study of the flow of charge fugacities slightly inside the region where charges are relevant , we keep only the two charge classes @xmath176 and @xmath177
. this should be a reasonable approximation since it is always one of these two classes that brings about the instability of qlro at the low - temperature border of the stability region and that represent the most relevant charge class among all classes and for all parameters , as discussed after eq .
( [ mo.re.c ] ) .
therefore we truncate the charge fugacity flow equations ( [ flow.y]b ) to @xmath178 y_{{\rm c}\uparrow } - 2 \pi y_{{\rm c}\uparrow}y_{{\rm c}\uparrow \downarrow } , \\
\frac{d}{dl } y_{{\rm c}\uparrow \downarrow } & = & [ 2 - ( p^2 / 2 \pi ) k^{-1 } ] y_{{\rm c}\uparrow \downarrow } + 2 \pi [ y_{{\rm c}\uparrow}^2 - 2 y_{{\rm c}\uparrow \downarrow}^2 ] .\end{aligned}\ ] ] however , we stress that these truncated flow equations are incomplete , i.e. , these two selected fugacities are coupled to all other fugacities in the complete flow equations .
we have dropped also charge class @xmath179 , which also brings about only quantitative modifications . only for @xmath180 they are as relevant as charges @xmath116 and
should be kept , e.g. , to obtain the vanishing of the disconnected correlation ( [ cocol]a ) . concerning the processing of the vortex sector ,
we actually do nt go beyond ref . .
that is , we restrict our quantitative analysis to the region , where vortices are irrelevant and the second - order terms to the vortex fugacity flow equations ( [ flow.y]a ) can be discarded .
the linear flow equations for all @xmath181 can be reduced to flow equations for @xmath182 and @xmath183.@xcite avoiding a fugacity expansion , the vortex correlations @xmath184 can be evaluated asymptotically for large length scales and can be expressed in terms of an _ effective _ vortex fugacity @xmath185 , @xmath186}_{\cal a } & = & \frac 2 { \tau^*},\end{aligned}\ ] ] with a fraction of `` polarizable dipoles '' @xmath187 this fraction is unity for temperatures @xmath188 and goes to zero as the temperature drops below @xmath189 , where vortices start to freeze .
the initial value and the renormalization flow of the effective fugacity is given by@xcite [ y.v ] @xmath190_{\cal a } , \\
\frac{d}{dl } y_{\rm v } & = & [ 2-\pi \tau^ * k(1-\sigma^t \tau^ * k)]y_{\rm v}.\end{aligned}\ ] ] to summarize the above derivations we collect the full set of flow equations that constitute one of our main findings : [ flow.full ] @xmath191 , \\ \frac{d}{dl } y_{\rm v } & = & [ 2-\pi \tau^ * k(1-\sigma^t \tau^ * k ) ] y_{\rm v } , \\ \frac{d}{dl } y_{{\rm c}\uparrow } & = & [ 2 -(p^2 / 4 \pi ) ( k^{-1 } + \sigma^l ) ] y_{{\rm c}\uparrow } - 2 \pi y_{{\rm c}\uparrow}y_{{\rm c}\uparrow \downarrow } , \\ \frac{d}{dl } y_{{\rm c}\uparrow \downarrow } & = & [ 2 - ( p^2 / 2 \pi ) k^{-1 } ] y_{{\rm c}\uparrow \downarrow } + 2 \pi [ y_{{\rm c}\uparrow}^2 - 2 y_{{\rm c}\uparrow \downarrow}^2 ] , \\
\frac{d}{dl } { \cal f } & = & -\frac{2 \pi}{\tau^ * } y_{\rm v}^2 - \pi ( 2 y_{{\rm c}\uparrow}^2 - y_{{\rm c}\uparrow \downarrow}^2).\end{aligned}\ ] ] this set of flow equations has no longer an explicit duality between vortices and charges ( even if we ignore the second - order terms in the flow equations of the charge fugacities ) , because the approximations used to treat the coupling of vortices and charges to disorder have to be taken in a fundamentally different way , mainly because of the validity ( invalidity ) of the fugacity expansion for charges ( vortices ) .
in particular it was possible to retain only two charge classes , whereas the collective effect of all vortex classes is represented by the effective vortex fugacity and the appearance of the parameter @xmath192 ( that actually parametrizes the most relevant vortex class ) . however , if we specialize flow equations ( [ flow.full ] ) to the case without disorder , the duality reappears and we retrieve the flow equations of ref .
.@xcite if we switch off the field , i.e. , if we drop the charge sector , the flow equations ( [ flow.full ] ) are reduced per construction exactly to those derived recently.@xcite in the presence of random fields [ where @xmath193 and @xmath194 , see eq .
( [ y.c.0 ] ) ] , the flow equations reduce in the charge sector to those of co.@xcite however , the present flow equations allow also the consideration of finite @xmath85 .
the charge sector for uniform fields has structural similarities with the flow equations that have been given ( but that have not been evaluated ) in ref .
.@xcite the flow equations in ref .
miss the contributions from fugacities @xmath195 and the freezing of vortices for @xmath196 .
we now determine the large - scale properties of the model by integrating the flow equations ( [ flow.full ] ) .
thereby the initial parameters @xmath76 and @xmath112 are mapped for @xmath197 onto @xmath198 and @xmath199 .
the fugacities , which are initially given by eqs .
( [ y.c.0 ] ) and ( [ y.v]a ) , flow independently .
note that @xmath192 is also a flowing parameter which is determined by the flowing @xmath76 and @xmath200 through relation ( [ def.tau ] ) . from the flow equations one easily reads off how vortex and charge fluctuations renormalize the physical parameters qualitatively .
equation ( [ flow.full]a ) means that the effective temperature is increased by vortices but reduced by charges , corresponding to disordering tendency of vortices and the ordering tendency of charges .
vortices always increase the strength of disorder but the effect of charges is ambivalent in eq .
( [ flow.full]b ) .
the competition between the various terms leads to a rich structure of the phase diagram . for the clarity of the analysis we first break the problem into parts that we eventually assemble to the complete picture . for completeness and further reference
we recall the main properties of the system in the absence of fields:@xcite at temperatures below the kt transition temperature of the pure system , that is , @xmath201 , vortices are irrelevant even in the presence of random phase shifts with a strength below the critical value ( see fig .
[ fig.pdg.v ] ) @xmath202 the increase of @xmath203 and @xmath200 under renormalization implies that the phase with qlro is less extended in the @xmath204 plane of the unrenormalized parameters as compared to the @xmath205 plane of the renormalized parameters .
nevertheless , this shrinking is only a quantitative modification and qlro is stable for low temperatures and weak disorder ( see fig . 3 of ref .
it is evident from the dependence of the flow equations on the parameter @xmath192 that the physics within the phase with qlro is different for high temperatures and for low temperatures .
vortices start to freeze and the breakdown of the fugacity expansion sets in at temperatures below the line @xmath206 separating these two regimes .
the spin correlation function @xmath207 decays algebraically as in the spin wave approximation ( [ gamma.sw ] ) but with a renormalized exponent @xmath208 through the renormalization of @xmath76 and @xmath85 the value of @xmath209 depends implicitly also on the transverse disorder .
its value is @xmath210 at the high - temperature end of the transition line and drops continuously down to @xmath211 at the zero - temperature end of the transition line .
the exponent @xmath212 of the edwards - anderson spin correlation ( [ ea.corr ] ) assumes values in the range @xmath213 and also varies nonuniversally along the qlro / sro transition . when the transition line is approached from the high - temperature side , the correlation length diverges exponentially ( with exception of a special point at @xmath206@xcite ) with decreasing distance from the transition line as in the absence of disorder.@xcite within the qlro phase
glassy features are found in vortex density correlations for temperatures @xmath214.@xcite let us now consider the complementary case where we evaluate the effect of fields in the absence of vortices . taking into account renormalization effects ,
the linearized charge - fugacity flow equations indicate the irrelevance of fields at high enough temperature even for arbitrarily strong disorder : @xmath215 in the low - temperature region of the phase diagram , where the magnetic field is relevant , we can determine the critical exponent @xmath216 that defines the scaling relation @xmath217 between the magnetization @xmath218 and the field @xmath7 . from the linearized flow equations of the charge fugacities we determine the exponent @xmath219 in the scaling relation @xmath220 between the field amplitude and the system size @xmath221 . from the two charge classes under consideration we find @xmath222 from these two scaling relations one obtains at first sight two different values for @xmath219 .
the actual value of @xmath219 , which describes the relevance of @xmath7 , is given by the larger of the two values . therefore we find @xmath223 as special cases of eq .
( [ mo.re.c ] ) . to determine this exponent it is important to keep in mind that different vector charges are of different order in the field . at this point
one might question again whether charge classes different from the two considered ones could change the value of @xmath219 .
this is not the case and can be understood from an inspection of the @xmath219 values for the other classes .
these values are obtained by dividing the exponent in eq .
( [ mo.re.c ] ) by the power @xmath224 of the scaling relation between the fugacity and the field .
the different order of the charge fugacities in the field is also the reason that the separation line @xmath225 , where dominance switches between charges of class @xmath114 or @xmath116 , does not coincide with the separation line @xmath226 where the exponent @xmath219 has its nonanalytic dependence on @xmath203 and @xmath85 . from the assumption that the magnetization yields an extensive contribution to the free energy , @xmath227 and the definition of the magnetization @xmath228 we identify the exponent @xmath229 this expression reproduces the universal value @xmath230 for fields with @xmath231 at the kt transition of the pure system.@xcite at low temperatures , where the field is relevant , we find two different phases . for weak disorder @xmath232
the ordering tendency of the fields dominates over the disorder and lro is established . in the opposite case @xmath233
the disorder wins over the field and results in quasi - short - range order ( qsro ) characterized below .
the phase diagram , obtained from a numerical integration of the flow equations in the absence of vortices , is depicted in fig .
[ fig.pdg.c ] .
lro is found in a region @xmath234 and @xmath235 of the unrenormalized parameters . under renormalization both
charge fugacities saturate at a finite value and drive these parameters to zero , @xmath236 on large scales @xmath203 and @xmath85 decay only inversely proportional to the logarithm of the length scale , and so does the effective exponent @xmath237 .
therefore this fixed point corresponds indeed to a phase with lro , where @xmath238 for @xmath239 .
the fact that the fugacities saturate at a finite value @xmath240 is consistent with a very ( maybe infinitely ) strong field @xmath7 at the fixed point , since even an infinitely strong initial field would lead to finite fugacities , see eq .
( [ y.c.0 ] ) .
the analysis of the lro phase can not be taken quantitatively , since the flow of @xmath203 and @xmath85 entails the relevance of other charge classes that have been ignored in this analysis . however , we expect modifications to be only quantitative since the omitted charges are less relevant than the retained charges [ see the discussion near eq .
( [ mo.re.c ] ) ] .
qsro is found in a region @xmath241 and @xmath242 of the unrenormalized parameters and identified with the following asymptotic flow : [ fix.qsro ] @xmath243 since both @xmath112 increase without bounds , there is no fixed point in a strict sense .
according to our analysis in sec .
[ sec.scal ] _ all _ charge classes different from @xmath244 become actually irrelevant at temperature not too far below @xmath245 .
however , at significantly lower temperatures other classes will be relevant , e.g. , @xmath246 at @xmath247 .
the divergence of @xmath85 means that the system behaves asymptotically like in the presence of _ random _ fields . since both @xmath112
do nt couple back into the flow equations , their effect on the angle difference correlation function can be evaluated in the spin - wave approximation accounting for the scale dependence of @xmath85 , @xmath248 as found in refs . .
in the context of crystals grown on disordered substrates this behavior of the correlation function is called `` super - rough''.@xcite to the extent that only the class @xmath116 contributes to the renormalization , @xmath76 is not renormalized .
for the random field model this non - renormalization can be shown in general on the basis of a statistical symmetry.@xcite then the amplitude of the difference correlation in front of @xmath249 is universal in terms of the transition temperature @xmath250 , which is determined by @xmath251 , namely @xmath252 for @xmath253 ( our quantitative analysis has to be restricted to parameters , where the fixed - point value of @xmath254 is small .
otherwise higher orders in this fugacity can no longer be neglected ) .
our calculation agrees in the prefactor of @xmath255 with that of carpentier and le doussal.@xcite the increase of the difference correlation faster than logarithmically but slower than linearly in the length implies a faster than algebraic but slower than exponential decay of the spin correlation , @xmath256 i.e. the correlation length is still infinitely large .
to emphasize this we call the order quasi - short - ranged . in the qsro phase the competition between
the spin coupling energy and the magnetization energy leads to a state , which has less order than each of the contributions alone would have in the absence of the other contribution ( qlro or lro , respectively ) .
this competition results in a frustration typical for glassy disordered systems .
since the symmetry - breaking field does not induce a finite magnetization , the scaling relation and the exponent @xmath216 are actually meaningless in this phase .
the qsro phase can be considered as `` glass '' since the disorder ( @xmath85 ) dominates the physics on large scales .
in addition , the dominant fluctuations are vector charges of type @xmath244 , i.e. fluctuations that are correlated in different replicas and usually would imply a finite edwards - anderson order parameter .
however , we are not aware of such an order parameter in the strict sense for this phase .
but we observe that the edwards - anderson correlation ( [ ea.corr ] ) decays only algebraically with finite @xmath257 , whereas the spin correlation @xmath207 decays faster than algebraically .
this indicates that the conformations of @xmath45 are disorder dominated and that thermal fluctuations are not stronger than in the absence of disorder .
additional evidence for the glassy nature of this phase can be derived from dynamical properties,@xcite which are beyond the scope of the present work .
the transitions between the three phases lro , qlro and qsro are of different nature .
when the lro / qlro transition is approached from above , the spin correlation exponent reaches the universal value @xmath258,@xcite and @xmath259 when it is reached from below since the field becomes irrelevant .
these values are thus independent of weak disorder with @xmath260 . along the qsro / qlro transition
the exponent @xmath261 has a nonuniversal dependence on the disorder strength , and @xmath216 is not a meaningful quantity .
however , @xmath262 is universal along this line.@xcite the phase transition lro / qsro is located near @xmath263 , see fig [ fig.pdg.c ] .
the transition line found from the numerical integration of the flow equations is bent towards smaller @xmath85 and the bending decreases with decreasing field amplitude . in eq .
( [ x ] ) we have found the scaling exponent of the field to change along the line @xmath264 , which indicated that the instability of qlro against the field differs above and below this line . however , a change of the exponent for initial scaling of the field does not necessarily imply that the renormalization flow drifts to different fixed points .
we can provide an additional argument in support of this location of the transition line .
let us identify the difference between the phases with lro or qsro by the presence or absence of @xmath265 on large scales .
we further consider very weak fields such that @xmath76 and @xmath85 are very weakly renormalized ( on intermediate scales ) .
then we may use an adiabatic approximation for the charge fugacities , i.e. , we suppose the values of the fugacities to be determined through @xmath266 and @xmath267 as a function of slowly varying @xmath203 and @xmath85 . from the flow equations ( [ flow.full ] ) one obtains two different solutions @xmath268 , \\
y_{{\rm c } \uparrow}^2 \approx \frac { y_{{\rm c } \uparrow \downarrow}}\pi \left[1-\frac{p^2 \sigma^l}{4\pi}\right ] , \end{array } \right .
\\ \mbox{qsro } & \quad & \left\ { \begin{array}{l } y_{{\rm c } \uparrow \downarrow } \approx \frac 1 { 2\pi } \left[1-\frac{p^2}{4\pi k}\right ] , \\ y_{{\rm c } \uparrow}=0 , \end{array } \right.\end{aligned}\ ] ] which are precursors of the lro and qsro fixed points .
eventually these precursors flow towards the true fixed points because of the scale dependence of @xmath76 and @xmath85 .
these two solutions merge again at the line @xmath269 . from a linear stability analysis of the fugacity flow equations
one finds that the lro precursor is stable and the qsro precursor is unstable for @xmath270 and the qsro precursor is stable for @xmath271 .
the lro precursor is unphysical for @xmath272 since there @xmath273 .
although this analysis supports the horizontal location of the lro / qsro phase boundary in the limit of weak fields , we should like to remind again that the transition can be described only qualitatively since this analysis includes only two charge types and the other neglected types are expected to give quantitative corrections even to the borders of the lro phase , where the fixed point values of the charge fugacities remain finite .
it is worthwhile to point out that the irrelevance of fields in the present context is equivalent to the irrelevance of the periodic crystalline potential for the roughening transition of crystals with correlated substrate disorder ( the mapping between these two problems is the one discussed in sec .
[ sec.map ] , @xmath10 corresponds to the substrate conformation ) .
the latter problem has been examined in ref .
and the phase diagram obtained there ( fig . 2 therein )
is the analog to the onset of relevance of fields in our fig .
[ fig.relev ] . in that work
the longitudinal bond disorder originated from _ correlated _ @xmath10 as in eq .
( [ sig.til ] ) .
therefore the lro in the present context actually corresponds to a logarithmically rough surface profile that is locked into a configuration parallel to the disordered substrate .
the present lro / qsro transition corresponds there to a transition from a locked - in and logarithmically rough surface to a super - rough surface . recently
horovitz and golub have examined the same problem in view of its implications for josephson junctions.@xcite using a gaussian variational replica approach they found a topologically similar phase diagram .
but there are the following differences : the phase boundaries do not coincide exactly , since their approach does not take into account parameter renormalizations . in the phase where we find qlro and lro they also have qlro and lro
however , their phase with lro is split into two sub - phases , their `` josephson '' phase ( where charges @xmath116 are irrelevant ) and `` coexistence '' phase ( where charges @xmath116 are relevant ) .
we find charges @xmath116 to be relevant in the whole lro phase , i.e. the whole phase with lro has the character of the `` coexistence '' phase .
the boundary between their sub - phases coincides with the line @xmath274 , where charges of type @xmath116 become relevant according to the linear scaling analysis . however , both charges are relevant in the whole lro phase , their appearance is triggered by the second - order contribution @xmath275 to the flow equation ( [ flow.full]f ) of @xmath276 .
the most important difference is the region where we find qsro and they still find qlro that qualitatively influences the dependence of the critical josephson current on the junction area .
again , this difference is due to the presence of second - order terms in the flow equations , now terms of order @xmath277 to the flow equation ( [ flow.full]f ) .
the disagreements we find here are well - known from the random field @xmath0 model and have to be ascribed to the insufficiency of the gaussian variational ansatz to capture the physics represented by second - order terms to the flow equations .
we now discuss the phase diagram in the presence of vortices _ and _ charges . we distinguish the cases of uniform fields and random fields , since their effective longitudinal bond disorder is crucially different . for definiteness
we consider the case @xmath278 of random phase shifts that are uncorrelated on different bonds . the general case with finite @xmath279 deviates only quantitatively . as in the absence of disorder,@xcite a phase with qlro is stable only for @xmath13 . for weak disorder
there is a lro phase at the low - temperature side of the qlro phase . for @xmath280 lro is stable up to @xmath281 , whereas for @xmath282 lro is stable only up to @xmath264 .
figure [ fig.phadi ] depicts the phase diagram for @xmath283 , which is generic for all integer @xmath284 . as compared to the absence of vortices , one main effect of vortices
is the restriction of the qlro phase at higher temperatures , where sro sets in .
a second important effect is that the phase with qsro is turned into a phase with sro .
since @xmath285 increases logarithmically with the length scale in the absence of vortices , it reaches the value @xmath286 , where vortices become relevant . on small scales vortices
practically do not contribute to a renormalization of the parameters .
if we start from bare parameter values at a temperature slightly below the qsro / qlro transition , only charges @xmath116 are initially relevant and increase @xmath285 according to eq .
( [ fix.qsro ] ) .
then vortices become relevant only on an exponentially large scale@xcite @xmath287 on this length scale sro sets in , that is , @xmath288 represents the correlation length of the system . at disorder strengths
@xmath289 the phase diagram , judged by its order at largest scales , appears to be reentrant , i.e. , a phase sequence sro / qlro / sro can be obtained by a pure temperature change .
however , the difference between the low - temperature and the high - temperature part can be substantial in terms of the correlation length and probably also in terms of dynamic properties . in a remote analogy the high - temperature part compared to the low - temperature part of the sro phase should be as different as a gas compared to a very viscous liquid . due to the restriction of our analysis to small fugacities of vortices and
charges we can not rule out the possibility of an actual transition between two different phases at @xmath290 for @xmath291 . since the boundary of the qlro phase , which is well described in our approach , has a discontinuous slope at @xmath292 for @xmath293 one might expect that this point is a tricritical point that should be merged by and a third transition line . for @xmath294 the phase boundary lro / sro is determined by the competition of vortices and charges , which are both relevant already on smallest scales .
the location of this transition can therefore not be determined .
the phase diagram differs from the one found in ref . crucially at low temperatures , where the sro and qlro phases reached down to the point @xmath295 and @xmath296 and a reentrant phase sequence sro / qlro / lro / qlro / sro was possible by changing temperature .
although we still find a reentrance sro / qlro / sro the qlro phase ( as well as the lro phase ) has now a convex shape in the @xmath297 plane .
the discrepancy can be traced back to the overestimation of vortex fluctuations in a vortex fugacity expansion , which is avoided in the present work .
the phase diagram fig .
[ fig.phadi ] has been obtained by a numerical integration of the flow equations ( [ flow.full ] ) for two different amplitudes @xmath298 and @xmath299 of the field and for a fugacity parameter @xmath300 appropriate for the mapping of the original @xmath0 model onto the coulomb gases .
we find only a slight dependence of the phase boundaries on weak fields @xmath301 .
one could in principle evaluate the flow equations also for larger fields and would probably find a direct crossover from the glassy low - temperature sro region to the nonglassy high - temperature region as suggested by co. however , we do not pursue this issue further since for @xmath302 it is certainly no longer legitimate to neglect the less relevant charge classes .
the mapping discussed in section [ sec.map ] allows us to study random fields with spatially uncorrelated @xmath10 by considering @xmath303 in the flow equations . in this limit
the contributions proportional to @xmath265 simply disappear from the flow equations , in which @xmath304 then no longer enters and the flow of @xmath304 can be identified with the flow of @xmath85 after subtracting the transformation term @xmath305 .
randomness in the field induces only a few changes in comparison to the case of uniform fields .
the schematic phase diagram for the system with random fields and random phase shifts is depicted in fig .
[ fig.schem]b , which is deformed only slightly by finite vortex fugacities and field amplitudes .
in contrast to the case of uniform fields , the lro phase disappears and the low - temperature qlro / sro boundary is located at @xmath306 down to @xmath296 .
thus a phase with qlro now exists for @xmath307.@xcite the spin correlation function has a finite exponent ( [ eta.inf ] ) ( involving the original @xmath85 and not @xmath308 since we are interested in the correlation of @xmath45 and not of @xmath309 ) . for temperatures @xmath310 the cardy - ostlund charges @xmath116 become relevant and still generate transverse bond disorder according to eq .
( [ fix.qsro ] ) ( which also describes the further decrease of longitudinal bond disorder ) .
therefore vortices become relevant again on scales ( [ l_v ] ) for @xmath310 .
to summarize , we have performed a renormalization group analysis of the model ( [ h ] ) in the villain approximation that allowed us to represent the model by two coupled coulomb gases .
the breakdown of the fugacity expansion for the vortex gas@xcite was treated nonperturbatively ( the flow equations contain the effective vortex fugacities @xmath185 that are nonperturbative functions of the original vortex fugacity @xmath311).@xcite we have shown that an analogous breakdown does _ not _ occur for an expansion in the charge fugacities and have identified two most relevant classes of vector charges out of an infinite set .
the renormalization group flow equations for the interacting coulomb gases have been established and evaluated for the limit where both gases are dilute , i.e. , for low temperatures and weak fields . from these flow equations
we have determined the structure of the phase diagrams , in particular the domains of stability of the phase with qlro . due to the breakdown of the fugacity expansion for the vortices
this phase was found to be more stable against disorder than found previously.@xcite the critical exponents @xmath209 , @xmath312 , and @xmath216 have been given in eqs .
( [ eta.inf ] ) , ( [ overeta.inf ] ) , and ( [ delta.inf ] ) . in a strict sense
our approach is valid only for the characterization of the qlro phase , where vortices and fields are asymptotically irrelevant .
nevertheless we have attempted to complete the picture of the phase diagram by examining how vortices and fields become relevant , which was taken as indication for the lro , qsro , or sro nature of the neighboring phases .
thereby we were also able to locate qualitatively the lro / qlro transition in the case of uniform fields .
this transition was found to merge one corner of the phase with qlro .
its location is in qualitative agreement with that found in a recent gaussian variational approach,@xcite where vortices had not been considered . on the high - temperature side of the qlro phase vortices
become relevant on relatively short length scales and immediately lead to an increase of the effective temperature and a suppression of the charge fugacities . on the low - temperature side of the qlro phase charges
become relevant first . for weak longitudinal disorder ( @xmath313 ) the flow converges to a lro fixed point ,
otherwise the flow tends on intermediate scales towards the qsro fixed point before it eventually crosses over to the sro fixed point .
since the crossover from qsro to sro occurs at finite values of the charge fugacities , this behavior might be modified by higher orders in the charge fugacities , which we are not able to incorporate systematically at present .
this difficulty excludes also the calculation of phase diagrams for uniform fields with @xmath314 and random fields with @xmath315 in the present framework .
co have suggested the possibility of a first - order transition between the glassy low - temperature sro and the based on the nonglassy high - temperature sro . to our understanding
this scenario was based on low - temperature properties of the vortex fugacity flow equation that are absent in the present analysis , where the vortex fugacity expansion has been avoided .
we speculate that there could be a transition from a low - temperature sro phase to a high - temperature sro phase that merges the upper left corner of the qlro phase .
it could be possible that this transition line persists even at very strong disorder and ends up at @xmath316 and @xmath295 .
this point corresponds to the so - called gauge glass model that represents a zero - temperature critical point.@xcite however , for such a transition line we have no evidence other than the fact that corners usually occur only at a three - phase coexistence and that the gauge glass fixed point should have a continuation for weaker disorder ( finite @xmath285 )
. it would be interesting to examine this scenario further , which probably can be achieved only by numerical methods .
the studied model has physical applications in various fields .
the most direct application is to @xmath0 magnets , where random phase disorder can be provided by nonmagnetic impurities through the dzyaloshinskii - moriya interaction and local crystal fields break the spin rotation symmetry.@xcite in the case when vortices are excluded , the model also describes crystal surfaces grown on a disordered substrate.@xcite such disorder can also be related to dislocations that leave the crystal at the surface.@xcite the lro / qsro transition described above corresponds to a disorder driven transition from a rough crystal surface to a super - rough phase . in the context of planar josephson junctions@xcite
lro corresponds to a finite critical current density even for arbitrarily large contacts and qsro corresponds to a critical current density that decays faster than algebraically with the contact area .
the phase diagrams obtained here should also be qualitatively similar to those of crystalline films on periodic substrates with disorder@xcite and of vortex lattices.@xcite the transitions occur between phases that differ in the degree of structural order .
however , these systems are more complicate to analyze in a similar framework since they have a two - component displacement field instead of the one - component phase of model ( [ h ] ) .
in addition , the periodicity of the lattice implies that both the particle interactions and the pinning energy have the same wavelength , i.e. , @xmath231 .
uniform and disordered fields then correspond to pinning that is commensurate or incommensurate with the crystal . since @xmath231 it appears likely that there is no phase where the effects of pinning and the proliferation of dislocations are irrelevant on large scales and that the strongly relevant fields and dislocations can not be well described by a direct generalization of the present framework .
we are grateful to j. kierfeld and t. nattermann for stimulating discussions and a critical reading of the manuscript .
s.s . acknowledges support from the deutsche forschungsgemeinschaft under project no .
sfb341 and grant no .
sche/513/2 - 1 and from the nsf - office of science and technology centers under contract no .
dmr91 - 20000 science and technology center for superconductivity .
in this appendix some intermediate steps in the derivation of the coulomb gas model ( [ h.rs ] ) are given .
the starting point is the effective hamiltonian ( [ h.vil ] ) .
the first step is to replace the scalar field @xmath47 by the charge field @xmath56 with the use of the poisson summation formula : @xmath317 in the corresponding partition function all @xmath318 $ ] are integrated over and all @xmath319 and @xmath320 are summed over .
the integrations can be extended to @xmath321 if the summation over the @xmath46 is restricted to a summation over a subset @xmath322 .
the selection of the subset fixes a gauge for @xmath46 , which is the vector potential of the vortex density , @xmath323 .
the summation over the subset of @xmath322 then corresponds to a summation over the vortex density .
the integration over the angles is most conveniently performed in fourier space and results in the hamiltonian ( [ h.rs ] ) . in the following
we examine closer the vortex - charge interaction that reads on at an intermediate level @xmath324 note that it is not possible to choose a coulomb gauge with @xmath325 due to the integer nature of @xmath326 .
the statistical weights @xmath327 are invariant under a change of gauge @xmath328 due to the integer nature of the charge field @xmath329 and the gauge field @xmath330 . to transform expression ( [ h.2 ] ) back into real space ,
let us look for simplicity at the ground state of hamiltonian ( [ h.1 ] ) in the absence of fields and disorder , @xmath331 but this is just the phase field generated by vortices , which can be written in real space ( neglecting discretization effects ) @xmath332 where @xmath333 measures the angle enclosed between the vector @xmath334 and an arbitrary reference direction in the @xmath68 plane . in direct analogy
the coupling ( [ h.2 ] ) becomes ( [ h.vc ] ) in real space .
@xmath335 and @xmath336 do not depend on the choice of this reference direction for vortex configurations satisfying the neutrality ( [ neutr ] ) .
only these configurations have finite energy and contribute to the statistics .
here we show how the angular interaction between vortices and charges induces a screening ( [ kc.screen ] ) of the interaction among charges when vortices are integrated out from the partition sum . following kosterlitz@xcite and
young@xcite we integrate out every pair of vector vortices with radius less than @xmath337 .
the contribution of a vortex pair with @xmath338 to the hamiltonian @xmath336 is @xmath339.\ ] ] since we assume a diluted coulomb gas , so that typically @xmath340 , one can expand the parts of the partition sum including this pair : @xmath341 [ \omega({\bf r}_1-{\bf r}')-\omega({\bf r}_2-{\bf r}')].\end{aligned}\ ] ] in contrast to the screening effects within the vortices , the first nontrivial term in this expansion , which is even in the charge density and produces nonvanishing screening , is of second order in the hamiltonian . in the dilute limit
we can expand @xmath342 then it is straightforward to perform the integration over the annulus @xmath343 . after that step one can translate the vortex pair over the whole space excluding the spheres occupied by the other vector pairs and obtains a screening of charges by vortices very similar to the screening of vortices by vortices.@xcite
for a convenient comparison of the present work with co we include the following substitution list for symbols ( restricted to the replica limit @xmath130 ) .
@xmath344 @xmath185 becomes the usual vortex fugacity and @xmath346 represents negligible higher orders .
the agreement with the corrected flow equations eq .
( 5.17 ) of ref . is given if one omits the exponential factors , which we understand as attempt to integrate the initial values for the fugacities .
fisher , t.a .
tokuyasu , and a.p .
young , , 2931 ( 1991 ) ; m. cieplak , j.r .
banavar , and a. khurana , j. phys .
a * 24 * , l145 ( 1991 ) ; h.s .
bokil and a.p .
young , , 3021 ( 1995 ) ; r.a .
hyman , m. wallin , m.p.a .
fisher , s.m .
girvin , and a.p .
young , , 15304 ( 1995 ) . | the combined effect of disorder and symmetry - breaking fields on the two - dimensional @xmath0 model is examined .
the study includes disorder in the interaction among spins in the form of random phase shifts as well as disorder in the local orientation of the field .
the phase diagrams are determined and the properties of the various phases and phase transitions are calculated .
we use a renormalization group approach in the coulomb gas representation of the model .
our results differ from those obtained for special cases in previous works .
in particular , we find a changed topology of the phase diagram that is composed of phases with long - range order , quasi - long - range order , and short - range order .
the discrepancies can be ascribed to a breakdown of the fugacity expansion in the coulomb gas representation .
implications for physical systems such as planar josephson junctions and the faceting of crystal surfaces are discussed . 2 |
galactic black hole ( bh ) sources are interesting objects to study as these sources are observed only in binaries and the process of accretion gets very complex as the disk evolves with time , especially when the sources undergo outbursts and jet ejections take place .
some of the bh binaries show persistent emission ( e.g. cyg x@xmath11 ) along with aperiodic x - ray variability ( e.g. grs 1915@xmath0105 ) , over more than a period of decade . some bh sources show outbursting behaviour for shorter duration varying from few days to months , and
are called as outbursting bh sources or transients ( e.g. gx 339@xmath14 , xte j1118@xmath0480 , xte j1748@xmath1288 , h 1742@xmath1322 , gro j1655@xmath140 ) . the outbursting bh sources show different types of intensity variations ( i.e. , outburst profile ) over different period of time @xcite .
some of the sources after being quiescent for a long time , show a sudden increase in the intensity level , and attain a maximum intensity within few days and then decay back slowly to quiescence ( h 1743@xmath1322 , a 0620@xmath100 , 4u 1543@xmath147 ) .
their light curves have a ` fast rise and exponential decay ( fred ) ' profile .
some sources have a slow rise to the peak and decay slowly to the quiescence ( gx 339@xmath14 , xte j1752@xmath1223 ) and their light curve profile is termed as ` slow rise slow decay ( srsd ) ' . during this whole phenomena ,
outbursting sources show different spectral and timing variabilities and exhibit different spectral states like _ hard , hard - intermediate , soft - intermediate , soft state _
( @xcite , @xcite ) and in some cases a _ very high state _ also ( @xcite , @xcite ) , in their hardness - intensity diagram ( hid ) .
it is also observed that in outbursting bh sources , strong jets are emitted which are seen in radio observations @xcite , during the transition from hard - intermediate to soft - intermediate state in the rising phase ( e.g. h 1743@xmath1322 , @xcite ) and in the declining phase when the source transits from soft - intermediate to hard state ( e.g. xte j1752@xmath1223 , @xcite ) .
quasi - simultaneous multiwavelength observations of sources like gx 339@xmath14 @xcite , xte j1748@xmath1288 @xcite , h 1743@xmath1322 @xcite , strongly suggests that the radio flares emitted are associated with the disk emission . it has been reported from the study of grs 1915@xmath0105 that , during the jet ejections , qpos are not observed as well as the comptonized component gets suppressed ( @xcite ) .
this implies that the ` hot ' comptonized corona gets disrupted and evacuated , and the source spectra softens , suggesting that the x - ray emission is mostly from the disk ( @xcite , @xcite , @xcite , @xcite , @xcite ) .
the x - ray transient source xte j1859@xmath0226 was first discovered @xcite with all sky monitor ( asm ) onboard rossi x - ray timing explorer(rxte ) @xcite on oct 9 , 1999 .
subsequently , the source was monitored in x - rays with rxte / pca ( proportional + counter array ) and cgro - batse ( compton gamma ray observatory - burst and transient source experiment ) for several months @xcite .
the outburst showed the typical fred profile .
spectral and temporal characteristics confirmed the source as a black hole candidate @xcite .
several observations in optical and radio wavebands confirmed the presence of the counterpart of the source @xcite .
spectroscopic studies of the counterpart showed weak emission lines arising from balmer series of hydrogen and he ii , which is typical for spectra of lmxbs @xcite .
the mass function was estimated to be ( 4.5 + /-
0.6 ) m@xmath2 , and an assumed inclination angle of 70@xmath3 gave a lower mass limit of 5.42 m@xmath4 @xcite . during the 1999 outburst ,
the source xte j1859@xmath0226 was continuously and extensively monitored in x - rays and in radio , which revealed the x - ray / radio correlations @xcite .
the source is observed to exhibit multiple flaring events of five in number . from the study of the spectral evolution in radio , it was found that the jet generation is implied by the production of hard x - rays and also that a correlation exists between soft x - ray and radio ejection during the first flare , while a correlation existed between hard x - ray and radio observation during the other ejections .
@xcite studied the temporal properties of the source and found that , the qpos observed can be classified into three types viz .
type a , b and c. this classification scheme based on the qpo characteristics and phase lag , has been considered as one of the basic formalities to classify the qpos in black hole sources ( see also @xcite ) . the phase lag difference between different types of qpos , suggests that the shape of oscillation is different in different energies .
but the evolution of low frequency qpos ( c - type ) as well as their origin during the initial rising phase of the outburst is still not clear . in the context of disk - jet symbiosis of black hole sources ,
@xcite have provided a unified picture along with the estimation of jet power as a function of x - ray luminosity .
their work also highlighted the occurrence of five radio flares in xte j1859@xmath0226 and suggested that all the flares occur when the source is in a similar spectral state .
they found that the rms of the power density spectra ( pds ) reduces during the occurrence of a flare .
@xcite studied the spectral properties of the source during its rising phase and found that the non - thermal flux ( 2 - 50 kev ) remains constant during the first flare of the source .
@xcite studied the evolution of the source during the rising phase of the outburst and found that there is a signature of partial absence of the qpo ( or comptonized corona ) , when the first radio flare occurred .
@xcite studied the spectral behaviour of this source in the context of global study of disk dominated states of several bh sources . considering a simple phenomenological model ( diskbb+powerlaw ) for bh spectral study
, @xcite showed that a large fraction of disk dominated observations of xte j1859@xmath0226 fall below the ` standard ' luminosity - temperature ( l - t ) relation ( see @xcite ) .
@xcite has studied the spectral characteristics of the source based on bepposax and rxte observations , in the context of a bulk motion comptonization model ( bmc of xspec ) , and have observed an evolving high energy cut - off component during the hard to soft transition .
they have also performed a correlated study between the fraction of comptonization and the rms variability .
but the ` complex ' evolution of fold energy since the beginning of the outburst along with the evolution of temporal features ( qpos , rms etc . ) , especially during a radio flare are not explained in detail .
so , in order to have a coherent study of the evolution of temporal and spectral properties of the source ( during the entire outburst ) as well as the implications on disk dynamics during the multiple ejections , we re - analysed the rxte pca / hexte temporal and spectral data of the source xte j1859@xmath0226 in the energy band of 2 - 150 kev .
several models have been proposed to understand the ` spectro - temporal ' evolution of an outbursting black hole binary . the accretion - ejection instability model @xcite and global disk oscillation model @xcite attempts to explain the origin of qpos .
recently , @xcite based on propagating mass accretion rate fluctuations in hotter inner disk , and @xcite based on oscillations from a transition layer in between the disk and hot comptonized flow have proposed alternative models to explain the origin of qpos
. there have been other attempts also to understand the hard x - ray spectral state variations @xcite as well as state transitions across the hid @xcite of the outbursting bh sources .
but none of these models , addressed as a whole the issues of evolution of low frequency qpos as well as soft and hard x - ray spectral components , and spectral state changes across the hid during the entire outburst . on the other hand , quite independently , several alternative models ( see 4 for discussion ) and more recently the phenomenological model @xcite have been put forward only for understanding the disk - jet symbiosis in bh sources .
so , we attempted to understand the evolution of temporal and spectral properties associated with the different branches of the ` q - diagram ' of the outburst of the source xte j1859@xmath0226 within a single framework of the two component advective flow ( tcaf ) model @xcite . basically , this model consists of two different types of accreting flow @xcite : sub - keplerian ( i.e. , freely falling and less viscous flow ) and keplerian ( i.e. , moves in circular orbit and high viscous flow ) matter @xcite .
we also delve deep into the nature of the accretion dynamics during the jet ejections of this source based on the detailed temporal and spectral x - ray properties . in general ,
outbursting black hole sources show signature of evolution of qpo frequencies @xcite , which could be explained based on the propagating oscillatory shocks solution @xcite of tcaf model . in this work
, we provided the solution of evolution of qpos during the rising phase and possible explanation of the ` reverse ' nature of the evolution of the hard spectral component ( i.e. , the fold energy ) , which can be interpreted with the shock acceleration mechanism @xcite in the sub - keplerian flow .
our findings show that the qpos are not observed during the radio flares , which we have explained with the possible scenario of ` evacuation / disruption ' of the comptonized corona .
we attempt to estimate the mass of the source by modeling the qpo evolution and also from the correlation between qpo frequency and photon index @xcite .
the preliminary results of our findings were presented in @xcite .
this paper has been organized in the following manner : in the next section , we discuss the observations and the procedures applied for data analysis using the standard packages for rxte pca / hexte . in 3 , we presented the results obtained from the temporal and spectral analysis for the whole outburst , estimation of mass in 3.2 , followed by specific spectral and temporal characteristics during each radio flare in 3.3 . in 4 , we discuss the possible physical scenario to explain the evolution of x - ray properties observed during the entire outburst of xte j1859@xmath0226 and the disk - jet connection , and summarize our conclusions .
we analysed the public archival data obtained from the heasarc database for the rxte satellite to study the characteristics of the source xte j1859@xmath0226 over the entire duration of the single outburst which occurred during the entire rxte era .
we analysed the pca ( 2 - 25 kev ) and hexte ( 20 - 150 kev ) data spanning 166 days since october 9 , 1999 ( mjd 51460.76 ) to march 23 , 2000 ( mjd 51626.6 ) .
we excluded the observations of july 2000 when the source was in quiescence , as the counts were very less .
the standard ftools package of heasoft v 6.11 was used for data reduction . for spectral and temporal analysis purposes
, we used the packages xspec v 12.7 and xronos v 5.22 respectively .
for the timing analysis , we used the science data of pca in the binned mode ( b_8ms_16a_0_35_h_4p , fs37 * ) of maximum time resolution of 8 ms which spans over 0 - 35 channels , and event mode ( e_16us_16b_36_1s , fs3b * ) of 16 @xmath5s time resolution for 36 - 249 channels .
we used the ftools task _ xtefilt _ to create filter file , and _ maketime _ to generate a good time interval ( _ gti _ ) file , with conditions of _ elevation angle @xmath6 10@xmath7 , offset of @xmath8 0.02 and time since saa passage @xmath8 30 min_. since it was not possible to select the data only for a single pcu from the binned mode data , we decided to extract light curves for all pcus .
light curves were generated in various energy bands of 2 - 6 kev , 6 - 13 kev , 13 - 25 kev and 2 - 25 kev in order to perform an energy dependent study of the pds . since the binned mode data consisted of only 0 - 35 channels ( 2 - 13 kev ) , we generated the light curve of minimum binning time , for the channel ranges 0 - 15 ( 2 - 6 kev ) and 15 - 35 ( 6 - 13 kev ) , using _ saextrct_. the light curve from the event mode data was created using _
seextrct _ for the minimum available binning time , and the data belonging to channel range of 36 - 67 ( 13 - 25 kev ) was extracted from the total available channel range of 36 - 249 .
a combined light curve for 2 - 25 kev was obtained , by summing up the light curves from 0 - 35 channels ( 2 - 13 kev ) and 35 - 67 channels ( 13 - 25 kev ) , using _
lcmath_. since the contribution of background counts were very less ( 8.6 counts over 2674 source counts at the peak of the outburst and 6.6 counts over 61.9 source counts at the minimum ) , we did not subtract the background counts while generating the light curve .
for the spectral analysis , we decided to extract data for pcu2 only , since throughout the entire outburst only pcu2 remained ` on ' and the efficiency of the detector also holds good .
the source spectra were extracted from the standard2 data product ( fs4a * ) , which has a time resolution of 16 sec .
background spectra were obtained with the help of _ runpcabackest _ , applying the background model for bright source and saa passage history file obtained from the pca background web - page of rxte for the corresponding epoch during which the source has undergone outburst .
spectral response was created using the _ pcarsp _ tool .
we also generated the background subtracted light curve for pcu2 counts for the energy range 2 - 6 kev , 6 - 20 kev and 2 - 20 kev , in order to plot the hid .
we extracted high energy ( 20 - 150 kev ) spectral data mostly from cluster a ( fh52 * ) of hexte , since specific observation by cluster b ( fh58 * ) had lesser counts , except for initial two days ( mjd 51462.76 and mjd 51463.83 ) , where there were no cluster a observation . using standard procedures
, we generated the deadtime corrected source and background spectra , and spectral response was generated using _
hxtrsp_. radio observations were carried out in the bands of 1.43 ghz , 1.66 ghz , 2.25 ghz , 3.9 ghz by various observatories ( vla , merlin , gbi , ratan ) during the outburst of the source , and the combined radio light curve was presented in @xcite .
we re - produced the radio light curve presented in @xcite using the software ` dexter . '
figure [ rlc - pca ] shows the x - ray and the radio light curve for the source xte j1859@xmath0226 during the entire outburst .
@xcite ( bottom panel of figure [ rlc - pca ] ) identified five radio flares during the entire outburst .
the five flares were observed to have their _ peak _ radio flux on mjd 51467.90 ( f1 ) , 51475.00 ( f2 ) , 51479.94 ( f3 ) , 51484.62 ( f4 ) , 51487.63 ( f5 ) . to obtain the pds
from the lightcurves generated , we used the ` powspec v 1.0 ' of the xronos package and the customized software based on idl , _ general high energy aperiodic timing software _
( ghats ) v 1.0.1 . while using ` powspec ' , a normalisation factor of ` -2 ' was applied in order to subtract the expected white noise level from the data , to obtain the squared rms fractional variability from the integral of the pds .
further analysis shows that the factor of ` -2 ' deviates about 0.5% in almost all states except the soft - intermediate states where it deviates by 3% to 5% .
the ghats package considers dead time effect and hence the correct value of normalisation factor ( as in @xcite ) while generating the pds .
the available minimum time binning factor of 0.0078 sec , which corresponds to a nyquist frequency of 64 hz , was chosen over 8192 segments .
the power obtained in the pds has units of rms@xmath9/hz .
we modeled the power spectra from 0.1 to 64 hz , by different components of _ lorentzians _
( @xcite ) for the qpos and broad - band noise , and a powerlaw for the red noise , wherever required .
the resultant centroid frequency is considered as the qpo frequency .
we also estimated the q - factor ( centroid frequency / width of the lorentzian ) , amplitude in rms and significance of the qpo observed .
the standard qdp / plt command of _ statistic _ was used to obtain the integrated value ( rms ) for the qpo and for the overall pds , in the range of 0.1 to 64 hz .
the error values on each parameter of the components were estimated at 90% confidence interval , using the _ fit err _ command .
we also studied the evolution of pds in different branches of hid of the outburst .
the phase lag between soft ( 2 - 6 kev ) and hard ( 6 - 25 kev ) photons during the flares was estimated using ghats .
the lightcurves and hence the fast fourier transforms for the 2 - 6 kev and 6 - 25 kev energy bands are generated .
the cross spectrum is computed , taking into account the subtraction of poissonian noise ( as per @xcite ) .
we followed the standard procedures of the ghats package to estimate the lag ( see also @xcite & figure [ lag - f3 ] of this work ) .
we performed a simultaneous fit of the pca and hexte data for the energy range of 3 - 150 kev , including a constant factor close to unity for normalising both the spectra . in order to get an idea of the normalising factor we estimated the factor by simultaneously fitting the crab spectra ( since it does not vary with time ) observed by rxte during the different spectral states of xte j1859@xmath0226 ( observation ids 40093 - 01 - 12 - 00 and 40093 - 01 - 15 - 00 ) .
the constant was observed to be @xmath10 1
. a simultaneous fit to the pca and hexte spectra of xte j1859@xmath0226 showed that when the constant factor was fixed at 1 for pca , a value varying between 0.9 and 1.1 ( over the entire data set ) was obtained for the constant factor corresponding to hexte spectra , which was used during the fitting .
we modeled the energy spectra using the standard accretion disk models ( _ diskbb _ and _ powerlaw _ ) for bh spectra , modified by the _ phabs _ model to account for the interstellar absorption . from an initial fit to the data sets during the entire outburst , we found that the nh parameter has an average value of 0.2 @xmath11 10@xmath12 atoms @xmath13 and fixed this value for further analysis .
this value of nh , agrees well with the average value of 0.216 @xmath11 10@xmath12 atoms @xmath13 , estimated by the leiden / argentine / bonn ( lab ) survey , and the dickey & lockman ( dl ) survey value of 0.221 @xmath11 10@xmath12 atoms @xmath13 ( ) .
@xcite quoted a slightly higher value of 1.1 @xmath11 10@xmath12 atoms @xmath13 .
we find that a change in nh from 0.2 to 1.1 , does not significantly change the overall results . for an initial fit to the observation 40124 - 01 - 11 - 00 with the model _
, we obtained @xmath14 ( = @xmath15/dof ) of 145.43/82 .
this resulted in residuals around 6.4 kev along with a smeared edge of @xmath10 8 kev @xcite and the model considered was + _ phabs(diskbb+gauss+smedge*powerlaw)const _ , which resulted in @xmath14 of 64.55/77 .
the width of the gaussian was fixed at 0.7 kev ( varies between 0.62 kev and 0.88 kev ) which was obtained as an average to all the fits ( except for the hard states ) .
systematic error of 0.5% was included in all the fits , for considering the uncertainities in the data .
we obtained the parameters of disk temperature , disk normalization , photon index with the powerlaw norm and the line energy of the gaussian line .
the photon index of the hexte component was tied to that of the pca component .
since the combined spectra showed a cut - off feature in the residuals at higher energies , we decided to include a _
highecut _ model for considering this .
the fits improved with @xmath14 = 57.57/75 for the model of + _ phabs*(diskbb+gauss+smedge*powerlaw*highecut)_. the f - test probability for inclusion of _ highecut _ model for this observation is 0.014 , but for the observations during the rising phase of the outburst the f - test probability is of the order of 10@xmath16 to 10@xmath17 implying that inclusion of _ highecut _ model is statistically significant .
we have considered the _ highecut _ model in order to understand the evolution of the cut - off and fold energy during the spectral states , although the f - test probability for inclusion of _ highecut _ during the soft - intermediate state is high .
figure [ ufspec - highecut ] shows the unfolded spectrum for the fits using _ highecut _ model for one of the observations .
hence , it was decided to consider + _
phabs*(diskbb+gauss+smedge*powerlaw*highecut ) _ as the final model , and similar method of analysis was followed for all the data sets during the entire outburst .
we also attempted to model the spectrum with _ comptt _
@xcite to take into account the parameters related to the non - thermal component of the spectra ( comptonized corona ) , and obtained @xmath14 = 72.59/78 for the same observation i d .
although the fit for the model + _
phabs*(diskbb+gauss+smedge*comptt ) _ was statistically good , the variations seen in the parameters were not able to explain the nature of the spectral evolution .
we observed that as the source rise towards the peak , the optical depth varied from 1.5 to 0.01 , and the seed photon temperature reduced from 2.5 to 0.49 which seems to be unrealistic , although the electron temperature increased from 29 kev to 239 kev .
recently @xcite have performed the spectral analysis of the source in the context of a complex bulk motion comptonization model . in order to have a comparative study
, we also applied the model of + _ wabs*(diskbb+gauss+bmc*highecut)const _ to the same observation i d , and obtained @xmath15/dof = 93.83/77 . a comparison with the analysis performed by @xcite shows that the fit will probably improve if the bepposax data is also included . in spite of the fact that we did not consider the bepposax data
, we studied this model for the different spectral states , and found that the variation of the parameters characterising the disk properties were similar to those shown in figure 4 of @xcite .
we also found that the value of parameters related to the disk and the photon index obtained using the model we have applied and that used by @xcite are within 1@xmath18 error limit .
since mjd 51486.87 ( @xmath10 @xmath19 day ) to the end of the outburst , the hexte spectrum was found to be mostly background dominated ( few cases spectra up to @xmath10 50 kev ) .
hence , we decided to fit only the 3 - 25 kev pca spectrum using _ diskbb _ and _ powerlaw_. better fits were obtained with the model of _
phabs(diskbb+gauss+smedge*powerlaw ) _ with @xmath15/dof of 28.62/39 ( for observation i d : 40124 - 01 - 40 - 01 ) .
it is to be noted that since we are considering rxte data @xmath20 3 kev only , the value of inner disk temperature obtained from the fits will not be exact .
@xcite pointed out that the diskbb model parameters should not be considered as ideal , since the parameter of disk radius varies due to the change in spectral hardening factor ( see also @xcite ) .
the convolution model , _ cflux _ was included to find the flux contribution of the individual components , in the energy range of 3 - 150 kev .
the error values for each parameter were obtained using the _ err _ command at a confidence interval of 90% .
we observe that the uncertainty in flux estimated using _
cflux _ , is not of large variation , except a few cases of the soft - intermediate state where a 5% variation is observed .
the variation between the values obtained by _
cflux _ and _ flux _ commands are found to be a maximum of 4% .
226 during the 1999 outburst showing variation of background subtracted count rate with the color , which has been observed with rxte / pca ( i.e. , pcu2 ) .
the points marked a @xmath21 h depict the start / state transitions / end of the outburst .
the five flares ( f1 @xmath21 f5 ) observed along with the weak signature of secondary outburst phase ( marked as rectangular box ) are also shown .
see text for details.,width=332 ]
in this section , we present the evolution of temporal and spectral properties ( in broad spectral band of 3 - 150 kev ) associated with different branches of hid of the source xte j1859@xmath0226 to understand the accretion dynamics from a theoretical point of view .
we estimate mass of the source by modeling the qpo evolution and also by correlation studies , and describe the x - ray features observed during the radio flares and thereby investigate the disk - jet symbiosis . in this sub - section , we present the characteristics of the hid along with variation of x - ray and radio flux .
figure [ rlc - pca ] shows the intensity variations of the emission from the source , both in x - rays and radio .
we observe that as the source rises from quiescence , the background subtracted x - ray count rate ( 2 - 20 kev ) rises from @xmath10 1 count / sec ( mjd 51460.76 = @xmath22 day ) to a peak of 2714 counts / sec ( mjd 51467.58 ) .
although for the first observation , the source count rate is very less with large error bars , this observation is important to follow the evolution of the source since the beginning of the outburst .
similar values were quoted by previous studies carried out by @xcite and @xcite .
the radio flux peaks ( mjd 51467.9 ) at a value of 114 mjy resulting in peak flare f1 , implying the release of matter in the form of jets @xcite .
multiple ejections have been observed resulting in flare f2 on mjd 51475 at 19 mjy when x - ray intensity was 2045 counts / sec ( on mjd 51474.28 ) , flare f3 on mjd 51479.9 at 31 mjy during which the x - ray intensity was 1637 counts / sec , flare f4 at a peak flux of 5.2 mjy on mjd 51484.6 where the x - ray intensity is around 1391 counts / sec .
the source count rate starts decreasing from 1444 counts / sec ( after mjd 51483.94 ) to @xmath10 4.9 counts / sec ( mjd 51626.6 = @xmath23 day ) and attains quiescence . while decaying , the source is observed to show a weak secondary emission ( see 3.1.5 ) for around 24 days with peak emission on @xmath24 day ( mjd 51530.13 ) of the outburst .
a fifth flare ( f5 ) is reported by @xcite based on asm lightcurve to occur on mjd 51487.6 , but there were no radio observations on that day .
figure [ q - dia ] shows the hid ( see also @xcite ) plotted using the background subtracted pcu2 count rate .
the points where the five flares occurred are also marked in the figure ( see also @xcite ) .
the possible transitions which occurred , are labelled from a to h. in general , outbursting sources undergo state transitions from _
hard @xmath21 hard - intermediate @xmath21 soft - intermediate @xmath21 soft @xmath21 soft - intermediate @xmath21hard- + intermediate @xmath21 hard _
@xcite and finally into a quiescent state . 226 .
absence of qpos during radio flares ( ejections ) are shown in the plot .
peak timings ( in unit of days ) of radio flares are also marked as f1 , f2 , f3 , f4 & f5 ( see table [ tabflares ] for details).,width=359,height=264 ] in figure [ qpo - evo ] , we show the variation of qpo frequency , qpo rms , the q - factor and significance of the qpos observed , over the entire outburst of the source xte j1859@xmath0226 .
the qpo frequencies increase monotonically from 0.46 hz to 6.09 hz in the rising phase of the outburst , which is a quite natural phenomena as observed in other outbursting bh sources @xcite . in the later phase of the outburst , the presence / absence of qpo frequencies
are directly linked with jet emission ( see figure [ qpo - evo ] and 3.3 ) as well as with the ` secondary ' emission of the outburst ( see figure [ rlc - pca ] and 3.1.5 ) .
evolution of the parameters of the spectral fit , obtained from the analysis are shown in figures [ spec - evo_f1234 ] , [ fold - ene ] and [ spec - evo_f5 - 6 ] . figure [ flux - evo_f123 ]
depicts the variation of disk flux and powerlaw flux in the 3 - 20 kev range and also for the 20 - 150 kev powerlaw flux ; and figure [ flux - evo_ratio ] shows the fraction of thermal flux contributing to the total 3 - 20 kev flux , the contribution of thermal flux over non - thermal flux ( 3 - 20 kev ) and also that of the flux ratio in 3 - 20 kev to 20 - 150 kev energy range ( up to mjd 51486.82 ) . during the first pca observation ( point a in figure [ q - dia ] ) ,
the pds was dominated by broad - band noise .
the next observation ( on mjd 51461.54 ) has a red noise along with a qpo of 0.46 hz ( q - factor = 8.31 , significance = 2.6 , rms amplitude = 19.3% ) which was not reported by @xcite , and is marked as the first point in the top panel of figure [ qpo - evo ] , figure [ qpo - pos ] and figure [ scal ] .
figure [ qpo - evo ] implies that the qpo frequency starts rising from 0.46 hz to 1.19 hz ( type c qpos ; see @xcite for details ) with the power spectra being mostly dominated by red noise and the rms of pds varies from 30% to 24% .
this variation in total rms is similar to that observed in other bh sources like gx 339@xmath14 ( see @xcite ) . in this branch ,
the photon indices of the energy spectra rise from 1.6 to 2.0 , the disk temperature ( t@xmath25 ) varies in - between 1.01 kev to 0.83 kev ( figure [ spec - evo_f1234 ] ) and hard x - ray flux dominates over the disk flux ( figure [ flux - evo_f123 ] ) ; and the hard x - ray spectral component ( i.e. , fold energy ) increased from 53 kev to 114 kev . in this branch
, the power spectrum has a flat top noise along with qpos and a powerlaw form at frequencies @xmath6 10 hz , with the total rms varying from 24% to 16% and the qpo frequency increasing from 3.05 hz to 5.97 hz ( type c ; see @xcite ) .
the photon index increases from 2.0 to 2.4 and the disk temperature slowly increases from 0.75 kev to 0.85 kev .
the disk flux is observed to be increasing as compared to the non - thermal powerlaw flux ( figure [ flux - evo_f123 ] ) , as well as the contribution of low energy to high energy powerlaw flux increases ( figure [ flux - evo_ratio ] ) .
figure [ fold - ene ] shows that during branch bc , the fold energy increased from 108.3 kev to 201.2 kev .
thus we observe an increase in fold energy from 53 kev to 201.2 kev , when the source moves through the branches ab and bc during the rising phase .
@xcite reported about an increase in the cut - off component during the transition from the hard to soft state . during the initial phase of the branch cd ( from mjd 51468.4 to mjd 51473.82 ) , the steepness of the red noise component starts decreasing and the power spectra has a flat top noise along with qpos ( both a and b types ) .
qpos of type c and also type b _ cathedral or twin qpos _ ( @xcite ) are observed during this branch .
qpos observed after mjd 51484.276 have not been classified into any types ( @xcite ) .
the total rms of the pds is observed to increase from 7% to 14% , while during the end phase of the branch ( after mjd 51484.87 ) , the pds has only broadband noise of amplitude 6% to 3% rms , with or without qpos along with a powerlaw form from 10 hz to 64 hz .
the fold energy is observed to vary randomly in between 64 kev and 180 kev , around an average of 130 kev within 1@xmath18 error bars ( except few data points , which are within 1@xmath18 to 2@xmath18 ) .
the disk flux is seen to vary randomly , whereas the power - law flux decreases gradually as shown in figure [ flux - evo_f123 ] . during the later phase of this state , the disk temperature slightly decreases from 0.85 kev to 0.7 kev and photon index decreases from 2.4 to 2.0 ( figure [ spec - evo_f5 - 6 ] ) .
although @xcite has reported a photon index of 1.8 during the period of mjd 51490 to 51501 , we observed a photon index varying between 1.9 and 2.4 with disk flux contribution of @xmath665% .
these parameters imply the characteristics of a soft - intermediate state during this period and not a hard state as suspected by @xcite . during this state ,
multiple ejection events have occurred and observed in radio as flares @xcite .
@xcite has reported on the features of hid and temporal characteristics during a flare .
we present detailed observational results of the spectral and temporal properties , during each of these flares in 3.3 . ) and photon index ( @xmath26 ) of observations before flare 5 ( f5 ) of the source xte j1859@xmath0226._,width=377 ] ) during the rising phase ( branches ab , bc ; top panel ) and soft - intermediate state ( branch cd ; bottom panel ) of the hid for the source xte j1859@xmath0226.,width=377 ] , photon index @xmath26 _ for observations from flare 5 ( f5 ) till the end of the outburst of xte j1859@xmath0226 .
note , spectral data is fitted up to 15 kev ( around @xmath27 day ) during the unusual variation of photon index.,width=377 ] during the transition from branch cd , we find that a few of the observations ( mjd 51504.31 to mjd 51510.29 in figure [ rlc - pca ] ) show hardness ratio around 0.1 ( figure [ q - dia ] ) and total rms @xmath10 1% to 2% , which suggest a short presence of soft state ( see @xcite ) . similar value of hardness ratio was quoted in figure 19 of @xcite , figure 1 of @xcite and figure a.1 of @xcite , for this source . during these observations the thermal flux dominates the spectra ( see top panel of figure [ flux - evo_ratio ] ) and qpos are also not observed .
but the other spectral features during this phase seems not to imply the characteristics of the ` canonical ' soft state as suggested by @xcite .
we observe that as compared to branch cd ( soft - intermediate state ) , the disk temperature reduces from 0.9 kev to 0.7 kev _
( to be noted that it is not the exact value , due to the limitations of rxte and diskbb model @xcite ) _ and the photon index remains @xmath10 2.2 ( spectra are harder compare to soft - intermediate state ) .
these variations are very different from the observations of other bh sources where the disk temperature remains high above 1 kev , photon index is @xmath62.5 during the soft state ( see 4.3.5 and figures 4.8 and 4.9 in @xcite ; @xcite ) .
all these results deviate from the ` canonical ' description of soft spectral state observed in outbursting bh sources @xcite . in branch ef ,
no significant qpos are observed ( @xcite has not classified the qpos also ) , and the pds has broadband noise of rms value @xmath10 3% .
the source is in the declining phase and we observe that the disk and power - law flux start gradually decreasing ( figure [ flux - evo_f123 ] ) without any signature of hard x - ray flux beyond 50 kev . * the outburst profile ( top panel of figure [ rlc - pca ] , see also figure [ q - dia ] ) indicates that , during this phase there is an enhancement in x - ray flux ( also observed by rxte - asm ) and it gradually decreases to the quiescence . from the spectral analysis
, we observed that there is an increase in the disk and powerlaw flux ( figure [ flux - evo_f123 ] ) along with increase in disk temperature for @xmath10 24 days i.e. , from @xmath28 to @xmath29 day of the outburst ( figure [ spec - evo_f5 - 6 ] and figure [ rlc - pca ] ) .
although the value of inner disk temperature ( t@xmath25 ) is less , we note that the variation in uncertainty of t@xmath25 is less than the value of t@xmath25 itself ( see also @xcite ) .
the increase in disk and powerlaw flux can also be noted in figure 14 of @xcite , after @xmath10 mjd 51530 .
although such an increase in flux has been observed in other sources like gx 339@xmath14 @xcite , but that particular observation ( @xmath10 mjd 52710 ) was associated with the hard - intermediate state @xcite .
the temporal analysis for the source xte j1859@xmath0226 indicates weak signature of qpos in the pds , during this phase of branch ef for the days from @xmath30 to @xmath31 of the outburst as shown in figure [ qpo - evo ] ( see also @xcite ) .
table [ secondary ] which summarizes the details of temporal features during this phase , implies the presence of qpos .
qpos of significance @xmath20 2.7 ( with higher q - factor ) are observed during mjd 51527.004 , mjd 51530.133 and mjd 51546.036 .
other qpos seem to be weak ( significance @xmath10 2.5 ) in nature ( except the qpo observed on mjd 51523.941 with significance of @xmath10 2 ) , but having q - factor @xmath10 3 . since the source is in the declining phase of the outburst
, these qpos observed can be classified as type c*. both the spectral and temporal properties seem to suggest that probably this phase / feature could be associated with a ` secondary ' emission , within the same outburst .
as the source approaches the end phase of branch ef , there is a sudden increase in the photon index from 2.2 to 3.2 ( spectral data fitted up to 15 kev ) , but with large errors ( star points in figure [ spec - evo_f5 - 6 ] ) , which turns out to be a puzzling nature .
0.0 cm ccccccc + mjd & & day & qpo frequency & significance & qpo rms & q - factor + & & & ( hz ) & of qpo & ( % ) & + + 51523.941 & & 63.18 & 0 & 0 & 0 & 0 + 51524.940 & & 64.18 & 8.66@xmath32 & 2.04 & 6.87 & 3.02@xmath33 + 51525.353 & & 64.59 & 8.21@xmath34 & 2.48 & 6.03 & 4.02@xmath35 + 51527.004 & & 66.24 & 9.25@xmath36 & 3.75 & 8.02 & 9.24@xmath37 + 51527.736 & & 66.97 & 9.75@xmath38 & 2.56 & 8.03 & 3.23@xmath39 + 51530.133 & & 69.37 & 8.86@xmath40 & 2.66 & 6.32 & 3.11@xmath41 + 51546.036 & & 85.27 & 4.71@xmath42 & 2.90 & 3.82 & 4.58@xmath43 + 51547.766 & & 87.00 & 0 & 0 & 0 & 0 + it is observed that during both the soft - intermediate branches cd and ef , there are qpos of types a , b , and a few c and c * ( see also @xcite ) . in branch fg , we do not observe any qpos and the power spectra remain to have broadband noise with total amplitude ( rms ) varying from 7% to 11% .
the source is in the declining phase and the spectra is dominated by powerlaw component , although we find weak signature of thermal flux with reduced inner disk temperature . at the end of the outburst ,
the source finally moved towards the hard state , which has characteristics similar to branch ab . the amplitude ( rms ) of the power spectrum has increased to @xmath1025% without any signature of qpos .
the energy spectra are mostly powerlaw in nature with photon index around 1.4 except during the last three observations which show a photon index of @xmath44 2.1 although the source count rates are very less with large errors , and it was difficult to fit the spectra .
erg @xmath13 s@xmath45.,width=359 ] erg @xmath13 s@xmath45.,width=340 ]
* * qpo evolution * + the evolution of qpo frequencies during rising phase of xte j1859@xmath0226 can be understood based on the propagating oscillatory shock ( pos ) solution @xcite of the tcaf . according to this model ,
the oscillating frequencies observed as qpos are due to the movement of the shock surface ( i.e. , comptonizing region ) . during the rising phase , the shock moves towards the bh and during the declining phase it moves away from the bh .
the pos solution states that the qpo frequency is inversely proportional to the infall time scale , and the qpo frequency ( @xmath46 ) can be obtained with a knowledge of the instantaneous shock location or vice - versa , and from the value of the compression ratio ( r = @xmath47/@xmath48 , where @xmath47 and @xmath48 are the densities in post and pre shock flows ) . according to @xcite , @xcite , the frequency of the shock location is similar to the observed qpo frequency .
thus , the qpo frequency for an axisymmetric toroidal bounded system ( i.e. , shock surface ; see figure [ mtcaf ] ) can be written as + 226 .
the fwhm of the fitted lorentzian of the qpo has been considered as the error bars .
the curves show the fitted solution based on the pos model .
the spectral states ( hs : hard state , his : hard - intermediate state , sis : soft - intermediate state ) has also been marked .
the bottom panel shows the variation of the shock location ( i.e. , comptonizing region ; see also 4 for discussions on the same).,width=332 ] + @xmath49},\end{aligned}\ ] ] + where @xmath50 = c / r@xmath51 is the inverse of the light crossing time of the bh , and r@xmath51=2gm / c@xmath52 for a bh of mass m. since the shock will be drifting with time @xcite , the time dependent shock location can be expressed as @xmath53 + where r@xmath54 is the shock location at time t=0 and v@xmath55 is the corresponding shock velocity @xcite .
the ` + ' sign is used for an outgoing shock in the declining phase and the ` - ' sign is used for an inward shock during the rising phase .
+ in figure [ qpo - pos ] , we show the evolution of qpo frequency with time ( days ) during the rising phase of xte j1859@xmath0226 , which we modeled using above equations of the pos solution .
fitted result allows to calculate the shock location and hence to estimate the size of the comptonizing region .
similar attempts have been performed for gro j1655@xmath140 , xte j1550@xmath1564 @xcite , gx339@xmath14 @xcite and h 1743@xmath1322 @xcite .
+ the first day of pca observation is considered as t=0 sec . figure [ qpo - pos ] shows that the initial three observations of qpos follow one trend while the rest shows another .
hence the evolution follows two different solutions for different parameters of shock location , velocity etc .. the fwhms obtained from the fits to the lorentzian feature of the qpos have been considered as error bars .
the fit to the initial three qpo frequencies gives the value of compression ratio as r=3.8 , initial shock location r@xmath54 of 138.8 r@xmath51 ( t=0 sec ) and the shock velocity v@xmath54 = 8.5 ms@xmath45 .
the fit to the rest of the qpo frequencies during the rising phase gives r=3.2 , r@xmath54=56.5 r@xmath51 , v@xmath54=1.5 ms@xmath45 .
the overall fit yields a mass of bh of 7.12 @xmath56 0.54 m@xmath2 , with reduced @xmath57 of 0.69 .
+ in the final fitting , we excluded the last observation ( as shown in cyan color ) on mjd 51466.89 , which is just before the first radio flare ( f1 ) .
the variation of shock locations ( i.e , the size of comptonizing region ) is shown in the bottom panel of the figure .
the oscillation of shock stalled at 21.2 r@xmath51 and this gives the estimate of minimum size of the comptonizing region ( before the flare f1 ) . from figure [ qpo - pos ]
, it can be observed that there is a kink ( see also @xcite ; 1998 outburst of xte j1550@xmath1564 ) at the point where the two fitted functions meet . * * @xmath26 - qpo correlation * + based on the model developed by @xcite , @xcite found that a correlation exists between the qpo frequency and photon index for bh sources .
they found that the correlation can be explained by the function @xmath58 by scaling w.r.t a source which has similar evolution .
the above correlation is linear with a slope for lower frequencies , and nearer to a particular frequency @xmath59 it becomes constant .
the parameter a denotes the value of the index saturation level , b is slope of the low - frequency part of the data and d controls how fast the transition occurs .
+ in figure [ scal ] , we have shown the correlation of photon index and qpo frequency for the source xte j1859@xmath0226 .
+ and qpo frequency during the rising phase of the outburst ( before f2 ) of xte j1859@xmath0226 . the fwhm of the fitted lorentzian of the qpo has been considered as the error bars.,width=340 ] + similar studies was performed by @xcite and the mass of the source xte j1859@xmath0226 was estimated as 7.7 @xmath56 1.3 m@xmath2 . in our analysis
, we have considered the observation results of mjd 51461.54 , which was not reported in @xcite , and those during the rising phase of the outburst .
we performed the scaling w.r.t gro j1655@xmath140 , since the evolution of photon index of xte j1859@xmath0226 saturates @xmath10 2.4 , which is closer to that of gro j1655@xmath140 at 2.3 , as mentioned in @xcite .
gro j1655@xmath140 is also a well studied source with a dynamical mass estimate of 6.3 @xmath56 0.5 m@xmath2 .
the value of d was fixed at 0.6 hz .
the fit was perfomed by applying craig markwardt s idl routines of _ mpfit _
@xcite , considering the errors in @xmath26 and qpo frequencies .
the resultant fit parameters obtained were a1859 = 2.427 @xmath56 0.008 , b1859 = 0.168 @xmath56 0.008 hz@xmath45 and @xmath59 = 4.84 @xmath56 0.18 hz . thus from the relation + m1859 = b1859
@xmath11 ( m1655/b1655 ) , + we obtained the mass of the source xte j1859@xmath0226 as 7.96 @xmath56 0.88 m@xmath2 , where b1655 = 0.133 @xmath56 0.008 hz@xmath45 . in this section , we present the results based on the variations of temporal and spectral features during radio flares i.e. , when the jet ejections have taken place to understand the disk - jet symbiosis . in table
[ tabflares ] , we summarize the properties of the source during the flares . as mentioned in 2 , we have performed an energy dependent study of the pds ( in the bands 2 - 6 kev , 6 - 13 kev and 13 - 25 kev ) of all the observations , in order to understand the nature of the x - ray features during the flares .
0.0 cm ccccccccccccc + x - ray / radio & x - ray flux & mjd & date & time & radio flux & & @xmath26 & & t@xmath25 + + + ( id / flare ) & ( cts / sec ) & & & & ( mjy ) & frequency(hz ) & rms(% ) & sig . &
type & & & ( kev ) + + 40124 - 01 - 11 - 00@xmath60 & 1505 & 51466.896 & 1999 - 10 - 15&21:30:08 & & 5.97@xmath61 & 15.4 & 6.83 & c & 2.35@xmath62 & & 0.88@xmath63 + + 40124 - 01 - 12 - 00 & 2714 & 51467.581 & 1999 - 10 - 16&13:57:20 & & 0 & 0 & 0 & - & 2.51@xmath64 & & 1.05@xmath61 + & & & & ( 7 hrs)@xmath65 & + * f1 & & 51467.904 & 1999 - 10 - 16 & 21:41:45.6 & 114 & + + 40124 - 01 - 13 - 00@xmath66 & 2597 & 51467.961 & 1999 - 10 - 16&23:04:16 & & 6.1@xmath61 & 3.8 & 7.24 & b & 2.42@xmath67 & & 1.07@xmath61 + + + 40124 - 01 - 23 - 01@xmath60 & 1558 & 51473.890 & 1999 - 10 - 22&21:22:08 & & 7.6@xmath68 & 4.3 & 5.82 & c * & 2.22@xmath69 & & 0.87@xmath70 + + 40124 - 01 - 15 - 02 & 1726 & 51474.087 & 1999 - 10 - 23&02:06:40 & & 7.3@xmath71 @xmath72 & 5.4 & 5.75 & c * & 2.25@xmath61 & & 0.97@xmath73 + + 40124 - 01 - 15 - 03 & 2045 & 51474.287 & 1999 - 10 - 23&06:54:40 & & 0 & 0 & 0 & - & 2.32@xmath73 & & 1.00@xmath73 + + 40124 - 01 - 24 - 00@xmath66 & 1984 & 51474.429 & 1999 - 10 - 23&10:18:08 & & 5.87@xmath74 & 2.7 & 7.36 & b & 2.44@xmath67 & & 1.04@xmath75 + + 40124 - 01 - 25 - 00 & 1248 & 51474.820 & 1999 - 10 - 23&19:41:20 & & 6.14@xmath76 & 7.1 & 8.27 & c & 2.09@xmath77 & & 0.87@xmath78 + & & & & ( 6 hrs)@xmath65 & + * f2 & & 51475.0 & 1999 - 10 - 24 & 00:00:00 & 19 & + + + 40124 - 01 - 31 - 00@xmath60 & 1262 & 51478.777 & 1999 - 10 - 27&18:39:28 & & 7.24@xmath79 & 6.4 & 5.76 & c * & 2.16@xmath67 & & 0.94@xmath63 + + + 40124 - 01 - 32 - 00 & 1637 & 51479.635 & 1999 - 10 - 28&15:15:28 & & 0 & 0 & 0 & - & 2.24@xmath67 & & 0.96@xmath62 + & & & & ( 7 hrs)@xmath65 & + * f3 & & 51479.940 & 1999 - 10 - 28 & 22:45:07.2 & 31 & + + 40124 - 01 - 33 - 01 & 1625 & 51480.044 & 1999 - 10 - 29&01:04:32 & & 0 & 0 & 0 & - & 2.28@xmath61 & & 0.96@xmath64 + + 40124 - 01 - 36 - 00@xmath66 & 1474 & 51483.106 & 1999 - 11 - 01&02:33:20 & & 4.87@xmath76 & 4.7 & 16.6 & b & 2.32@xmath64 & & 0.98@xmath61 + + + 40124 - 01 - 37 - 02@xmath60 & 1391 & 51484.275 & 1999 - 11 - 02&06:37:20 & & 4.43@xmath61 & 3.3 & 10.6 & b & 2.25@xmath64 & & 0.94@xmath64 + & & & & ( 8 hrs)@xmath65 & + * f4 & & 51484.625 & 1999 - 11 - 02 & 15:00:35.8 & 5.2 & + + 40124 - 01 - 38 - 00@xmath60 & 1266 & 51484.872 & 1999 - 11 - 02&20:56:16 & & 0 & 0 & 0 & - & 2.21@xmath77 & & 0.94@xmath61 + + 40124 - 01 - 39 - 00@xmath66 & 952 & 51485.874 & 1999 - 11 - 03&20:59:44 & & 6.97@xmath80 & 3 & 5.56 & c * @xmath81 & 2.14@xmath77 & & 0.87@xmath64 + + + 40124 - 01 - 41 - 00@xmath60 & 870 & 51487.009 & 1999 - 11 - 05&00:13:36 & & 4.77@xmath82 & 2.1 & 7.59 & c * @xmath81 & 2.34@xmath83 & & 0.85@xmath62 + & & & & ( 15 hrs)@xmath65 & + * f5@xmath84 & & 51487.63 & 1999 - 11 - 05 & 15:15:23.1 & - & + + 40124 - 01 - 42 - 00 & 857 & 51488.41 & 1999 - 11 - 06 & 09:49:36 & & 7.39@xmath85 @xmath72 & 3.1 & 1.3 & c * @xmath81 & 2.33@xmath86 & & 0.84@xmath61 + + & & 51488.48 & 1999 - 11 - 06 & 11:31:28 & & 7.34@xmath87 & 3.1 & 4.5 & c * @xmath81 & & & + + 40124 - 01 - 43 - 00 & 765 & 51489.47 & 1999 - 11 - 07 & 11:27:28 & & 0 & 0 & 0 & - & 2.05@xmath88 & & 0.85@xmath61 + + 40124 - 01 - 49 - 00 & 502 & 51496.462 & 1999 - 11 - 14 & 11:06:24 & & 0 & 0 & 0 & - & 2.10@xmath61 & & 0.79@xmath62 + + + 40124 - 01 - 49 - 01@xmath89 & 571 & 51497.254 & 1999 - 11 - 15 & 06:06:40 & & 5.02@xmath90 & 2.7 & 6.32 & c * @xmath81 & 2.18@xmath91 & & 0.79@xmath61 + + 40124 - 01 - 50 - 01 & 502 & 51498.390 & 1999 - 11 - 16 & 09:22:24 & & 0 & 0 & 0 & - & 1.96@xmath92 & & 0.77@xmath62 + & & & & ( 16 hrs)@xmath65 & + * f6 ( ? ) & & 51499.100 & 1999 - 11 - 17 & 02:24:00.0 & 7.9 & + + 40124 - 01 - 51 - 00 & 406 & 51501.384 & 1999 - 11 - 19 & 09:13:36 & & 0 & 0 & 0 & - & 1.90@xmath77 & & 0.76@xmath64 + * * * * * * + absence of qpo during f1 ( in 2 - 5 kev and 13 - 25 kev band ) , whereas qpos were absent in 2 - 25 kev during all other flares . * a * - minimum duration of x - ray observation before flare ; * b * - observation of last qpo before flare ; * c * - observation when qpo reappeared ; * d * - no radio observations , flare day based on asm light curve @xcite , radio flare might have occurred after this ( see figure [ rlc - pca ] ) ; * e * - not classified in @xcite ; @xmath93 - weak indication
the x - ray observation @xmath10 24 hrs before ( mjd 51466.896 ) the radio flare peak on mjd 51467.9 , shows the presence of a type c qpo of 5.97 hz . for the next observation ( mjd 51467.581 ) which is 7 hrs before the flare peak we performed energy dependent study in narrower bands ( 2 - 4 kev , 2 - 5 kev , 2 - 6 kev etc . ) and found that the power spectra for 2 - 5 kev and 13 - 25 kev bands does not show any signature of qpo . a broad 7.79 hz type a qpo
is observed , only above 5 kev as evident in the 6 - 13 kev pds with rms amplitude of 3% , which is typical for type a qpos .
so , in the overall range of 2 - 25 kev , a broad qpo of 7.79 hz is observed ( see also @xcite ) .
the next observation which is 2 hrs after the flare peak , shows a type b qpo of 6.1 hz in all the energy bands .
we also observe that the total rms of the pds reduces ( see also @xcite ) to a value of 2.7% when the qpo is not observed .
since the qpo was not observed in the soft band of 2 - 5 kev , we have shown the variation in the pds over 2 - 5 kev in figure [ f1 - 2 - 5-kev ] , for observations before and after the flare peak .
similar variations in the power spectra of 13 - 25 kev band also observed ( not shown in figure ) .
it is seen that , before the flare occurred i.e. , when the qpo was observed , the hard photons slightly lag behind the soft photons .
7 hrs before the flare peak , the 2 - 5 kev soft photons are lagging compared to 5 - 25 kev hard photons . when the qpo is observed again in the pds , the hard photons
are observed to lag the soft photons .
we also note that before the flare peak , the photon index increased to 2.51 , disk temperature increased from 0.88 kev to 1.06 kev ( table [ tabflares ] and figure [ spec - evo_f1234 ] ) and the fold energy of the electrons increased from 151 kev to 201 kev .
the disk flux increased from 2.14 @xmath11 10@xmath94 erg @xmath13 s@xmath45 to 8.03 @xmath11 10@xmath94 erg @xmath13 s@xmath45 ( figure [ flux - evo_f123 ] ) and dominated over the powerlaw flux ( 3 - 20 kev ) by a factor of 0.55 .
the 3 - 20 kev flux is found to be more than the 20 - 150 kev flux , by a factor of 5 ( figure [ flux - evo_ratio ] ) , suggesting the soft nature of the spectrum .
this implies that the emission is disk dominated .
after the flare , the photon index decreased to 2.42 and disk flux reduced to 7.83 @xmath11 10@xmath94 erg @xmath13 s@xmath45 .
this implies the hard nature of the spectrum although the disk temperature was around 1.07 kev and the fold energy had reduced to 156 kev .
thus the broad qpo seen ( just 7 hrs before the flare peak ) in the 6 - 13 kev band while not in 2 - 5 kev and 13 - 25 kev , implies that the qpo is ` partially ' observed ( see also @xcite ) .
we could not observe a ` complete absence of the qpo ' ( i.e. , in 2 - 25 kev band ) due to lack of continuous x - ray observations during the flare and just before / after .
the spectrum also softens as compared to the spectrum of the observation before the flare .
17 hrs before the radio flare peak of f2 .
the power spectra are scaled by factor of 1000 , 10 , 1 , 0.01 from top to bottom.,width=340 ] radio observation shows that the flare f2 , peaked on mjd 51475.0 with a flux of 19 mjy .
25 hrs earlier ( mjd 51473.89 ) to the peak of f2 , we observe type c * qpos in the pds .
the observations @xmath1021 hrs before ( mjd 51474.087 ) peak of f2 shows a weak signature of qpo with an increase in frequency ( 7.3 hz with q - factor of 1.2 , significance of 5.75 ) .
4 hrs later ( mjd 51474.287 ) the pds of f2 shows complete absence of qpos and the total rms reduces to 3.02% ( see also @xcite ) .
_ cathedral b - type _ qpos are observed at frequencies 3 hz & 6.1 hz in the next x - ray observation which is 14 hrs before ( mjd 51474.429 ) peak of f2 .
the evolution of power spectra is shown in figure [ pds - f2 ] , which clearly indicates the ` absence ' of qpo during the flare , but qpo is observed again before the flare peak ( see 4 for discussion ) .
we find that when the qpo is not significant in the power spectra , the soft photons are lagging .
17 hrs before peak of flare f2 where there is a complete absence of qpo , no lag of soft to hard photons is seen . when the qpo is observed again , @xmath1014 hrs before peak of f2 , the hard photons
are observed to lag behind the soft photons . the corresponding softening of the spectra is also observed in the variation of the spectral parameters .
when the qpo was not present ( over mjd 51473.89 to mjd 51474.287 ) , we noted that the disk temperature increases from 0.87 kev to 1 kev , and the photon index also increased from 2.22 to 2.32 .
the disk flux increased from 2 @xmath11 10@xmath94 erg @xmath13 s@xmath45 to 7 @xmath11 10@xmath94 erg @xmath13 s@xmath45 ( figure [ flux - evo_f123 ] ) , but the fold energy increased from 83 kev to 138 kev .
the ratio of disk flux over powerlaw flux increased from 0.43 to 0.65 with a significant decrease in hard x - ray flux ( figure [ flux - evo_ratio ] ) . when the qpo was observed again ,
the photon index further increased to 2.44 , with the disk temperature @xmath10 1 kev only .
the disk flux reduced to @xmath10 4.28 @xmath11 10@xmath94 erg @xmath13 s@xmath45 , whereas the fold energy increased to 154 kev .
this suggest that , the nature of variation in spectral features along with the temporal properties ( i.e. , qpo features ) during f2 , is different from that of flare f1 and hence disk - jet symbiosis could be further complex ( see 4 for discussions ) . on mjd 51478.777 , a 7.2 hz type c * qpo is observed .
radio observations indicate that multiple flares ( see figure [ rlc - pca ] ) have occurred on mjd 51479.94 within @xmath10 40 hrs of gap with flux values of 31 mjy & 24 mjy during the period when f3 occurred .
the x - ray observation which is around 7 hrs before f3 , does not show any signature of qpo and the pds is dominated by broad - band noise component ( figure [ pds - f3 ] ) .
the subsequent x - ray observations does not show qpos and the total rms decreased ( see also @xcite ) to 2.4% .
qpos are observed almost after 3 days at a frequency of 4.8 hz ( type - b ) .
we find an indication of soft photon lagging just before the flare peak , whereas when the qpo is not at all observed ( @xmath10 3 days ) , no lag is observed .
the hard photons are observed to be lagging , when the qpo is observed again after 3 days ( see figure [ lag - f3 ] ) . during f3 ,
disk temperature and photon index increases slightly , whereas the soft flux increased from 2.2 @xmath11 10@xmath94 erg @xmath13 s@xmath45 to 5.2 @xmath11 10@xmath94 erg @xmath13 s@xmath45 ( figure [ flux - evo_f123 ] ) , which implies the softening of the spectra .
the fold energy decreased from 126 kev to 97 kev .
we also note that during the observations where there were no qpos , disk flux dominates over the powerlaw flux by a factor of 0.6 to 0.9 with a significant decrease in hard x - ray flux ( see bottom panel of figure [ flux - evo_ratio ] ) . at the same time , the powerlaw index and fold energy varies , but the disk temperature remains almost constant ( figure [ spec - evo_f1234 ] ) .
7 hrs before the radio flare peak of f3 .
the power spectra are scaled by factor of 100 , 10 , 0.01 from top to bottom.,width=340 ] the qpos observed before and after the ejections during the flares f1 , f2 and f3 are of type c / c * and type b respectively ( see table [ tabflares ] ) . around 8 hrs before ( i.e. on mjd 51484.27 ) the peak of next flare f4 , a type b qpo is observed .
the subsequent observations ( on mjd 51484.87 and mjd 51485.07 ) does not show any signature of qpos ( figure [ qpo - evo ] ) , except broadband noise in the pds ( power spectral evolution not shown here ) with decreased total rms of 2.5% which is consistent with @xcite .
it is also clear from figure [ rlc - pca ] that a flare has occurred with a flux of 5.2 mjy ( mjd 51484.625 ) .
@xcite reported the time of flare based on x - ray observation from asm .
6 hrs after the peak of f4 ( mjd 51484.87 ) , qpos are completely absent in the power spectra and are observed only after 30 hrs ( on mjd 51485.874 ) at a frequency of 6.97 hz ( type c * with less rms and less q - factor ; see @xcite for details of type c * qpos ) . during the flare f4 , soft photons
are observed to be lagging , while in flares f2 and f3 , no lag of soft to hard photons was observed .
it is also seen from figure [ spec - evo_f1234 ] and table [ tabflares ] that during the flare peak , the spectral index varied around 2.2 .
the disk temperature remained around 0.9 kev , whereas the fold energy varied around 150 kev ( figure [ spec - evo_f1234 ] ) .
the ratio of low energy to high energy flux increases with a factor of @xmath10 6 ( lower panel of figure [ flux - evo_ratio ] ) , implying the nature of spectral softening . the disk flux dominated over the non - thermal flux by a factor of 0.57 to 0.71 , as seen in figure [ flux - evo_ratio ] .
when the qpo was observed again , the photon index decreased to 2.14 , with inner disk temperature of @xmath10 0.87 kev , and decreased fold energy of 96 kev .
we observe a 4.7 hz qpo in the pds of mjd 51487.01 .
although the q - factor is less , these qpos can be considered as of type c * class based on the qpo frequency and amplitude .
the next x - ray observation ( mjd 51488.41 ) indicates the presence of a weak qpo at 7.39 hz ( type c * ) in the pds of mjd 51488.41 and another type c * qpo at 7.34 hz on mjd 51488.48 . during the observation on mjd
51489.47 qpos are not observed in the power spectra ( see figure [ pds - f5 ] ) .
@xcite has reported a ejection / jet ( radio flare ) , to have occurred on mjd 51487.63 , based on the x - ray observations by asm ( see also @xcite ) .
but there were no radio observations ( see bottom panel of figure [ rlc - pca ] ) during this time .
the weak presence of qpo on mjd 51488.41 and mjd 51488.48 , was followed by a complete absence of qpo in the power spectra of mjd 51489.47 along with softening of the spectra .
this indicates that the fifth radio flare would have probably occurred after the time reported by @xcite .
this is also implied by the detection of a 2.4 mjy radio flux on mjd 51490.4 ( see figure [ rlc - pca ] ) .
qpos ( type c * ) are observed only after @xmath1010 days ( mjd 51497.25 ) .
thus during f5 we observe that the qpo observed before and after the flare is of type c * , whereas during flare f4 the evolution of qpo is from type b to type c*. these features are not similar to that observed during the flares f1 , f2 and f3 .
the variation in spectral parameters does not show any significant change during the flare f5 .
the photon index is seen to vary in between 2.38 and 2.33 , while the disk temperature remains almost constant around 0.79 .
although the disk flux decreased from 4.24 @xmath11 10@xmath94 erg @xmath13 s@xmath45 to @xmath10 3.48 @xmath11 10@xmath94 erg @xmath13 s@xmath45 , the ratio of disk to power - law flux is observed to increase by a factor of 3 to 6.7 ( top panel of figure [ flux - evo_ratio ] ) along with significant decrease of hard x - rays photons ( beyond 50 kev ) , implying the softening nature of the spectrum . when the qpo is observed again ,
the photon index has reduced to 2.18 with the disk temperature of @xmath100.79 kev .
the disk flux value is observed to dominate over the non - thermal flux , but with a lesser factor of 1.9 .
we observe a signature of qpo ( type c * with lesser q - factor ) at 5.02 hz ( on mjd 51497.25 ) , after 10 days of flare f5 .
the next observation ( on mjd 51498.39 ) does not show any qpos ( figure [ qpo - evo ] ) in the pds , except a broad band noise .
the total rms of the pds reduces from 4.02% to 1.6% .
radio observations also show that the flux decreases to @xmath10 1 mjy and later the source becomes non - detectable .
we find that , there is a radio brightening ( see @xcite ) on mjd 51499.10 ( after 16 hrs of x - ray obs . ) of flux of 7.9 mjy ( figure [ rlc - pca ] ) , implying the possible signature of occurrence of a flare ( f6 ? ) that could have triggered on or before this time .
since the evolution of the different types of qpos during the ejections seems to be complex , we attempt to understand if any correlation exists between the qpo frequencies of different qpo types , with the thermal and non - thermal flux .
the same has been presented in the following sub - section .
erg @xmath13 s@xmath45.,width=359 ] in figure [ flux - qpo ] , we show the variation of disk and powerlaw flux as a function of qpo frequency for the observations in hard state ( ab ) , hard - intermediate state ( bc ) and up to mjd 51484.27 during flare f4 in the soft - intermediate ( branch cd ) of hid ( see figure [ rlc - pca ] & [ q - dia ] ) , so as to consider the types of qpos which have been already classified in @xcite .
we observed that during the rising phase ( hard and hard - intermediate states ) the qpo frequency of type - c increases as both the disk and powerlaw flux increases , implying a positive correlation . during the soft - intermediate state ( branch cd )
the type b qpos are observed to be correlated with the powerlaw ( non - thermal ) flux but not with the disk flux .
a few type c qpos detected during the soft - intermediate state ( branch cd ) are observed not to show any correlation with the powerlaw flux .
figure [ flux - qpo ] also shows that a few of the type c * qpos have a correlation with the disk flux but are not at all correlated with the powerlaw flux as indicated by the random variation .
only very few type a qpos are observed , and hence it is not possible to find any correlation with disk or powerlaw flux .
the evolution of temporal and spectral properties of the source xte j1859@xmath0226 during the 1999 outburst suggest that the source does evolve via different spectral states throughout the outburst .
detailed analysis during the multiple ejections ( i.e. , radio flares ) reveals that the ejections are possibly associated with the comptonized corona , and hence the sudden ` appearance ' of such radio flares or ` jets ' could be because of some instability happening in the comptonized corona .
although several attempts have been made to understand the evolution of ` spectro - temporal ' characteristics as well as the ejection events observed in outbursting bh sources , till date , there is no comprehensive picture for the 1999 outburst of xte j1859@xmath0226 . in this paper
, we attempted to understand the possible accretion dynamics associated with the evolution of various x - ray features of xte j1859@xmath0226 , in the context of two component advective flow ( tcaf ) model ( @xcite ) . according to tcaf , the compton cloud ( formed in shocked accretion phase ) is nothing but the sub - keplerian matter dominated centrifugal pressure supported boundary layer ( i.e. , cenbol ) , which intercepts soft photons , mostly originated from the keplerian disk , and the cenbol up - scatters them via inverse comptonization to produce high energy photons .
the cenbol forms due to shock transition in the flow and its surface may oscillate due to the oscillation of shock , which could be responsible for the generation of qpos ( @xcite , see references therein ) . during the rising phase , we observed a monotonic increase of qpo frequency ( c type ) from 0.46 hz to 5.84 hz in both hard and hard - intermediate states .
this happens because , the shock location propagates towards the bh and the size of the shock formed region i.e. , the cenbol surface or region of comptonization , ( which is oscillating ) reduces , and hence the qpo frequency increases ( @xcite ) .
we estimated the shock location ( size of cenbol ) based on the pos solution ( see 3.3 ) . during this period , the instantaneous shock location ( see figure [ qpo - pos ] ) and
hence the ` compton cloud / cenbol ' varies from 114.7 r@xmath95 to 28.7 r@xmath95 . using the same solution
, we also estimated the size of the cenbol region , before and after the flare . during the soft - intermediate state
, the qpo frequency ( all types of qpos ) varies in between 6 hz to 8 hz , and this implies that the size of the corona / cenbol remains almost constant ( i.e. , shock location varies between 21.1 r@xmath95 to 17.5 r@xmath95 ) .
we also found that qpos in the soft - intermediate state are sporadic in nature i.e. off ( qpo not observed ) and on ( qpo observed again ) , which could be linked with jet ejection ( see figure [ qpo - evo ] and below for discussion ) . during the declining phase of soft - intermediate ( except during the ` secondary ' emission of outburst ) , hard - intermediate and hard states , qpos are hardly observable , whereas for other bh sources like gx 339@xmath14 , h 1743@xmath1322 , xte j1550@xmath1564 qpos are observed during all the spectral states ( except soft states ) @xcite .
@xcite evoked the idea of the diffusive propagation of perturbation in the disk like configuration of keplerian ( as extended disk ) and sub - keplerian ( inner part of the disk as ` compton ' cloud ) matter distribution @xcite , to understand the different noise component in the power spectra .
so , it could be possible to explain the evolution of total rms as well as modeling the broadband power spectra of different branches of hid of xte j1859@xmath0226 , in the context of tcaf model , although detailed modeling of evolution of the power spectra of the source is beyond the scope of the present work .
the evolution of the hid ( figure [ q - dia ] ) shows that during the rising phase , the source was initially in the hard state ( region ab ) , and the variation of the spectral parameters , suggest that the spectra is mostly dominated by the sub - keplerian component ( i.e. , larger ` compton ' cloud ) with negligible contribution from the keplerian disk . as the source moves towards the hard - intermediate state
( implied by figures [ q - dia ] and [ spec - evo_f1234 ] ) , we observe that thermal flux increases relatively more than the non - thermal flux ( figure [ flux - evo_f123 ] ) and both are correlated with the increase in frequency of type c qpos ( see figure [ flux - qpo ] ) .
this implies that the keplerian flow starts increasing along with the sub - keplerian flow , which is still dominating over the keplerian in the hard - intermediate state ( bc of hid ) . as a result ,
the energy spectra become softer and the size of the cenbol also starts decreasing ( see figure [ qpo - pos ] ) .
this is evident from the increase of qpo frequencies during the hard - intermediate state .
it is observed that as the source move from hard to hard - intermediate states , the fold energy also ( see figure [ fold - ene ] ) increases ( increase in cut - off energy during hard to soft transition was observed by @xcite ) .
the increase in fold energy seems to be quite ` unnatural ' compared to other bh sources , like gx 339@xmath14 and xte j1550@xmath1564 , where figure 6 of @xcite and figures 1 and 2 of @xcite respectively , pointed out the monotonic decrease of cutoff energy during the rising phase .
it seems that the shock acceleration mechanism , which converts fraction of thermal electron to the non - thermal electron within the compton cloud , becomes more important as the source evolves in this phase , and hence the fold energy increases although the source spectra becomes softer and softer .
similar kind of behaviour is observed in the hard and soft state spectrum of cyg x@xmath11 , which has been modeled using shock acceleration mechanism in tcaf @xcite .
it has to be noted that the kink observed around @xmath96 day during the rising phase ( see figure [ qpo - pos ] ) , when the qpo frequency changes from 3.05 hz to 3.64 hz , is peculiar in sense that at the same time there is an abrupt change in the photon index from 2.05 to 2.2 along with an increase in disk flux from 0.42 @xmath11 10@xmath94 erg @xmath13 s@xmath45 to 0.7 @xmath11 10@xmath94 erg @xmath13 s@xmath45
. this could be possible due to sudden change in the accretion flow dynamics during this phase . during the soft - intermediate state ( cd &
ef ) the evolution pattern of hid is complex .
we find that the photon index varies in between 2.5 to 2.1 and disk temperature varies around 0.9 kev , whereas variation of diskbb and power - law flux is not at all correlated ( see figure [ flux - evo_f123 ] ) .
this unnatural variation of flux occurs in short time - scales ( few hrs to day scales ) .
also , the fold energy varies randomly about an average value of 130 kev within 1@xmath18 error limits , whereas the change in fold energy during rising phase is well above the 2@xmath18 error limit .
this could be due to multiple ejections that have taken place , and they are directly associated with the disk dynamics .
we observe all types of qpos during this phase of which the b - type qpo correlates with power - law flux whereas its variation is random with disk flux and hence no correlation exists .
c - type qpos are correlated with the disk flux but does not show any specific correlation with the power - law flux .
so , it is possible that sub - keplerian flow plays a major role ( to generate various types of qpos , relation between various types of qpos with flux ) compared to the keplerian flow , as the ` hot ' compton cloud evolves faster ( i.e. , less viscous ) than the keplerian disk .
we could observe a short presence of a ` soft state ' during branch de for a few observations ( see figure [ q - dia ] ) where the hardness ratio is of @xmath100.1 and total rms of pds @xmath10 1% to 2% without any signature of qpo . during this phase ,
the keplerian flow is supposed to dominate over the sub - keplerian flow to produce ` softest ' spectrum .
as observed , the energy spectra is dominated by thermal emission ( see figure [ flux - evo_ratio ] ) but harder in nature ( disk temperature also less ) compared to the soft - intermediate state ( see figure [ spec - evo_f1234 ] and [ spec - evo_f5 - 6 ] ) .
hence , the observed state could not be a ` canonical ' soft state .
this could be possible due to enhancement of sub - keplerian flow during this time that makes the spectra harder along with the strong presence of keplerian flow ( i.e. , thermal emission ) .
it can be understood in the context of tcaf model that , the multiple ejections during the soft - intermediate state restrict the disk to become keplerian dominated flow since the ejections are coupled with the comptonized corona , which is mostly sub - keplerian in nature . while the source continued in the soft - intermediate state ( branch ef ) , there is an indication of weak increase in x - ray intensity ( ` secondary ' emission ) of around 24 days with peak emission on @xmath24 day of the outburst ( see figure [ rlc - pca ] ) .
the monotonic increase in qpo frequencies ( see figure [ qpo - evo ] ) and increase in other spectral parameters ( figure [ spec - evo_f5 - 6 ] , [ flux - evo_f123 ] and table [ secondary ] ; see also @xcite ) , indicate that there could have been sudden increase in keplerian as well as in sub - keplerian flow during the declining phase of the outburst , which results as a weak ` secondary ' emission .
the weak presence of soft state and the weak ` secondary ' emission , makes the hid different from the ` standard ' evolution pattern of other outbursting sources ( @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) . at the end of the outburst ,
the source transits from soft - intermediate to hard via hard - intermediate state before reaching the quiescence phase .
the photon index starts decreasing , with a very weak signature of thermal emission ( figure [ spec - evo_f5 - 6 ] and [ flux - evo_f123 ] ) , which implies that the emission component is again dominated by sub - keplerian matter over keplerian matter .
the modeling of the hid ( i.e. , q - track , see also @xcite ) which follows the evolution of x - ray features of the outburst , and of the broadband spectrum ( 3 - 150 kev ) will be carried out using tcaf model and will be presented elsewhere ( @xcite ) .
results of the modeling ( see 3.2 ) of qpo evolution based on pos solution and @xmath26-qpo correlation suggest that the mass of the source is @xmath10 6.58 to 8.84 m@xmath2 .
this matches well with the previous estimate of the mass as 7.7 @xmath56 1.3 m@xmath2 by @xcite
. the sudden occurrence of a radio flare can be due to some disturbance in the disk system , which results in ejection of the material as jets .
several attempts have been made until now , to explain the phenomena of jet ejection in bh binaries ( see also @xcite ) .
study of the jet emissions , using mhd simulations were carried out by @xcite , @xcite and @xcite .
these models suggest that the corona is like a windy hot material which blows away as jets from the inner regions of the accretion flow . in this work
, we attempted to explain the ejection mechanism based on dynamics of magnetic flux tube inside the tcaf ( see figure [ mtcaf ] ) as described by @xcite .
when the ejection takes place due to the collapse of magnetic flux tube , the cenbol gets disrupted and the matter is ejected as a jet . as a result , oscillation of the cenbol ( i.e. , shocked surface ) ceases and hence , the qpos will not be observed in the pds .
the energy spectra shall be dominated by thermal emission from the disk , as the matter from the cenbol ( ` hot ' electron ) is ejected .
the fold energy is observed to increase just before the flare and decrease just after the flare .
this could be the possible indication of ` disruption ' of the corona / cenbol .
the subsequent observation of qpo and spectral hardening in a time scale of few hours to days suggest that the matter flows in sub - keplerian disk forming the compton cloud / cenbol responsible for qpo generation .
as the observations suggest that the companion is of k - type @xcite and hence magnetically active , it is possible that the strong magnetic field anchored with the matter is continuously fed into the disk system during the rising phase of the outburst .
hence , during the transition from hard to soft - intermediate state , the field gets sheared and stronger ( in the form of toroidal flux tube ) and disrupts the comptonized corona ( in different time scales ) , resulting in multiple ejections at different time - scales . for the flare f1
, we observe a type c qpo of 5.97 hz which corresponds to a cenbol size of 21.2 r@xmath95 . during the partial absence of the corona ( i.e.
, qpo is not observed in 2 - 5 kev and 13 - 25 kev bands ) @xmath10 7 hrs before the flare peak , we observe a 7.79 hz qpo at a shock location of 17.8 r@xmath95 . since the sub - keplerian flow moves in shorter time scale it takes less duration of time to form the cenbol again , which starts oscillating .
hence qpo of 6.1 hz ( type b ) is observed 2 hrs after the flare peak at a shock location of 20.9 r@xmath95 .
thus , a duration of 9 hrs ( maximum ) was required for the cenbol to get disrupted and form again . during flare f2 , a 7.7 hz type c
* qpo is observed at 17.9 r@xmath95 around 21 hrs before the flare peak . but within 7 hrs
, the sub - keplerian component ( as the flow is less viscous ) has formed the corona and the oscillations of 5.87 hz ( type b ) are re - stored at 21.4 r@xmath95 . a closer look into the pds evolution ( figure [ pds - f2 ] ) shows that the qpo frequency increases with diminishing power , before the flare occurs and the qpo is not observed ( see 3.3.4 ) . the qpo was again observed before the radio flare peak , which seems to be inconsistent as compared to other flares , where we observe a complete absence of qpo before the radio peak .
we observe similar characteristics in other sources like xte j1550@xmath1564 , where the time gap between qpo being not observed and the flare peak was more than 24 hrs ( @xcite ) .
so , it is possible that the process of evacuation of the cenbol is rather slow during f2 , and hence the ejected material took long time to peak as a radio flare . by that time , the corona is formed again to reproduce the qpos , which are interestingly of _
cathedral type_. as we mentioned , the qpo and jet formation are coupled with the comptonized corona , which is sub - keplerian in nature .
hence , the disk - jet coupling mechanism in f2 , which seems to be ` complex ' in nature ( ` spectro - temporal ' properties are totally different from other flares ) could be because the dynamics ( i.e. , the amplification of magnetic field , motion of tube etc . ) of flux tube within the cenbol are different .
it requires further investigation to study the coupling between flux tubes and sub - keplerian flow and hence to understand the ` complex ' disk - jet symbiosis .
but , for both the flares f1 and f2 , the disruption of the inner part of disk and its re - filling time required for oscillation of cenbol , is around 7 to 9 hrs .
so , it is evident that within the hours time - scale only the sub - keplerian matter can easily move in , whereas the keplerian flow does not since it moves in viscous time - scale . before the peak of flare f3 a 7.24 hz type c * qpo
is observed corresponding to the cenbol size of 18.6 r@xmath95 . when the qpo is observed again at 4.87 hz ( type b )
the shock location is 24.3 r@xmath95 .
we find that , the disruption and refilling time of the inner disk ( formation of oscillating cenbol ) is @xmath10 3 days .
it is possible that the magnetic field would have been strong enough ( i.e. , continuous collapse of flux tube ) to disrupt the disk repeatedly , resulting in multiple signature of radio activity ( see bottom panel of figure [ rlc - pca ] ) . during this long disruption process ,
no lags in soft to hard photons are observed , disk temperature remains almost constant with increased disk flux and the fold energy decreases .
these indicate that the ` hot ' compton cloud is ejected in the form of jets . as the source is also in the declining phase ,
the matter supply in both the form of keplerian and sub - keplerian flow has reduced and hence , took longer duration to form the inner - part of the disk for the oscillation of the cenbol to start again . during f4 , before the flare peak the qpo was observed at 4.43 hz ( type b ) and the cenbol size is of 25.9 r@xmath95 . when the qpo is observed again
its frequency is 6.97 hz ( type c * ) at 19.1 r@xmath95 . before f5
, we observed a significant qpo of 4.7 hz ( type c * ) at 24.9 r@xmath95 followed by weak indication of 7.3 hz qpos ( type c * ) at 18.2 r@xmath95 , and after the flare the qpo is again observed with frequency of 5.02 hz ( type c * ) at 23.8 r@xmath95 .
the flares f4 and f5 occurred when the source was in the declining phase of the outburst , where the sub - keplerian flow dominates . while in decay the rate of incoming matter ( both the keplerian and sub - keplerian ) from the companion reduces , and hence the strength of the magnetic field also reduces .
so , we notice that the radio flares observed have less peak flux as compared to previous flares .
although there are no radio observations during the flare f5 ( see @xcite ) , the results we obtained ( 3.3.5 ) suggest that the flare would have probably occurred after the day reported by @xcite . as a result , during both the flares
, the disruption and refilling time of inner part of the disk ( formation of oscillating cenbol ) is longer of around 30 hrs for f4 ( with a possibility of secondary emission after @xmath10 16.8 hrs with radio flux of @xmath10 8.9 mjy as shown in figure [ rlc - pca ] ) and 10 days for f5 respectively .
we found a possibility of occurrence of another radio flare ( i.e. , f6 ) on or before mjd 51499.10 with radio flux of @xmath10 7.9 mjy .
there is an indication of qpo not being observed after the peak radio flux . as there is lack of x - ray and radio observations during this time ,
it is not possible to pinpoint the exact time of flare ejection . above
discussions on all the flares suggest that the radio flare take different time to peak and qpos are not observed during the flares .
the re - filling time scale ( @xmath10 few hrs to days based on available data ) to form oscillating cenbol ( corona ) resulting in qpos , also suggests that the accretion flow could be sub - keplerian in nature .
similar study of the disk - jet connection based on the magnetized tcaf model has been earlier presented for grs 1915@xmath0105 by @xcite .
we observe that the type of qpo is independent of the ejection event ( see 3.3 and column 10 of table [ tabflares ] ) .
the results mentioned in 3.3 clearly show that the qpos observed before and after the flare , need not be of specific type .
but , the prediction of jet ejections ( without any radio detection ) , based on qpos being observed or not in the power spectra , requires further studies in the context of tcaf model which could be possible after detailed investigation of x - ray observation of other outbursting sources .
we have also observed similar characteristics ( but with no definite relation between the types of qpos before and after the ejection ) in other bh sources . in the case of xte j1550@xmath1564 , gro j1655@xmath140 and xte j1752@xmath1223
, we found that the qpo observed before and after the flare are of type c. for h 1743@xmath1322 , we observed a type c qpo before the ejection and the pds showed a type b qpo after the ejection .
so , in the present scenario , it is not possible to generalise or establish a relation between types of qpo and flare ejection events .
although , there were attempts to establish the link of b type qpos with radio flare , along with the drop in rms in the power spectra ( see @xcite ) .
detailed study on qpo types and radio flares will be performed for several bh sources ( based on available x - ray and radio data ) ( @xcite ) . due to the complex behaviour of qpos during ejection events , we presented in 3.4 how different types of qpos are correlated with the flux variations
. the variation of type b / c qpos w.r.t disk and powerlaw flux is not similar to that observed for gx 339@xmath14 by @xcite , where both type b and c qpos are correlated with the disk and powerlaw flux ( see figures 5 and 8 in @xcite ) .
figure [ flux - qpo ] also shows that the type c * qpos have a weak correlation with the disk flux but does not have any correlation with the powerlaw flux , and a similar characteristic was observed for gx 339@xmath14 by @xcite .
type a qpos do not show any correlation with flux for gx 339@xmath14 , while xte j1859@xmath0226 has only very few type a qpos , which limits the results . @xcite and @xcite observed a positive correlation between qpo frequency and disk flux during the rising phase ( for gro j1655@xmath140 , xte j1550@xmath1564 , h 1743@xmath1322 ) , which is also observed for xte j1859@xmath0226 .
an opposite correlation was observed between qpo frequency and powerlaw flux for gro j1655@xmath140 , while a positive correlation was observed for xte j1550@xmath1564 and h 1743@xmath1322 . in the present work
, we have not explained the origin of different types of qpos before and after the ejections . according to tcaf model , the base of the jet and origin of qpos both
are directly coupled with the cenbol dynamics .
recently , we have done simulation to show that the mass loss from the disk and origin of qpos in bh are linked with the cenbol ( @xcite , see references therein ) .
so , it could be possible to address the types of qpos during the ejections , as the qpo generation in tcaf model depends on the sub - keplerian flow characteristics ( i.e. , viscosity , angular momentum , energy etc . of the flow ) .
this work which involves detailed simulation ( sph and tvd based ) is under progress and would be presented elsewhere .
there have been reports of flare events ( radio ejections ) during the outburst of sources like xte j1748@xmath1288 ( @xcite ) , xte j1752@xmath1223 ( @xcite ) and h 1743@xmath1322 ( @xcite ) .
the study of the outburst evolution and flaring events associated with x - ray properties observed for the bh sources xte j1748@xmath1288 , xte j1752@xmath1223 , h 1743@xmath1322 , xte j1550@xmath1564 , gro j1655@xmath140 , gx 339@xmath14 and maxi j1836@xmath1194 , in the context of tcaf model is under progress .
preliminary results on the presence and absence of qpos associated with flaring events for several outbursting black holes have been presented in @xcite .
based on the findings of ` spectro - temporal ' evolution of the source xte j1859@xmath0226 , we can summarize that : * the source evolved through various states in hid with a sequence of _ hard @xmath21 hard - intermediate @xmath21 soft - intermediate @xmath21 soft state ( not ` canonical ' ) @xmath21 soft - intermediate @xmath21 hard - intermediate @xmath21 hard states _ before reaching to the quiescence phase .
multiple radio ejections ( or the enhancement in sub - keplerian flow ) might have stopped the source to evolve from soft - intermediate to ` canonical ' soft state . * the qpo frequency increases monotonically during the rising phase of the outburst .
the evolution is modeled with pos solution to estimate the size of the ` compton ' corona .
similar kind of variation is also observed in other outbursting bh sources . *
results from the modeling of the qpo evolution and @xmath26-qpo correlation suggest that mass of the source could be between 6.58 m@xmath2 and 8.84 m@xmath2 . *
the temporal evolution of the pds shows that the qpos are not observed , during the ejection as observed in radio . * during f1 , no qpo is observed in the soft band of 2 - 5 kev ( also not in 13 - 25 kev band ) , but a broad qpo of less amplitude is observed in the 6 - 13 kev band .
this implies that there has been a ` partial ' absence of the qpo in 2 - 5 kev and 13 - 25 kev bands . * during f2 , f3 , f4 and f5 , we find that the qpos are not at all observed in the pds . *
the qpo frequency observed before and after the flare need not be of same type during all the flares . * during all the ejections , the x - ray spectrum gets soften and fold energy ( @xmath97 ) decreases , as the ` hot ' electron cloud ( corona ) gets disrupted and spectrum is dominated by disk emission . only in the case of f2 ,
@xmath97 increases , the evacuation process seems to be complex . * ` disruption ' of inner part of the disk ( corona / cenbol ) occurs as a resultant of dominant thermal emission over non - thermal flux resulting in qpos being not observed .
this could be due to the ` catastrophic collapse ' of toroidal flux tube of strong magnetic field in the hot region of sub - keplerian flow .
* during the declining phase , the source might have undergone a ` secondary ' outburst observed as a weak x - ray flaring activity for @xmath10 24 days . * due to lack of continuous x - ray observation during the flare time , it was not possible to tighten the duration of disruption and refilling time scale .
so , it is required to have more precise and simultaneous observation in _ radio - uv - x - rays _ to provide better picture of disk - jet dynamics in outbursting sources . in this context , continuous monitoring of outbursting sources with india s upcoming multi - wavelength satellite astrosat along with gmrt , vlbi etc
. will be a best opportunity for studying the complex accretion dynamics of bh sources .
we thank dr .
p. sreekumar ( iia , bangalore ) , dr .
s. seetha ( isro headquarters ) and dr .
a. agarwal ( isac , bangalore ) for various suggestions and support .
radhika d. , acknowledges the research fellowship provided by isro satellite centre . this research has made use of the data obtained through high energy astrophysics science archive research center on - line service , provided by nasa / goddard space flight center and of the general high - energy aperiodic timing software ( ghats ) package developed by dr .
tomaso belloni at inaf - osservatorio astronomico di brera .
we thank dr .
dipankar bhattacharya of inter university centre for astronomy & astrophysics , pune for providing an opportunity to participate in the ` advanced x - ray timing workshop ' and also to dr .
tomaso belloni for the sessions on usage of ghats .
we are thankful to the anonymous referees for their suggestions which helped to improve the manuscript .
belloni t. m. , hasinger g. , 1990 , a&a , 230 , 103 belloni t. m. , psaltis d. , van der klis m. , 2002 , apj , 572 , 392 belloni t. m. , homan j. , casella p. , et al . , 2005 , a&a , 440 , 207 belloni t. m. , parolin i. , del santo m. , et al . , 2006 , mnras , 367 , 1113 belloni t. m. , 2010 , ` the jet paradigm - from microquasars to quasars ' , lect .
notes phys 794 , edited by t. belloni , 794 , 53 blandford r. d. , znajek r. l. , 1977 , mnras , 179 , 433 bradt h. v. , rothschild r. e. , swank j. h. , 1993 , a&a supplement series , 97 , 355 brocksopp c. , fender r. p. , mccollough m. , et al . , 2002 , mnras , 331 , 765 brocksopp c. , miller - jones j. c. a. , fender r. p. , stappers b. w. , 2007 , mnras , 378 , 1111 brocksopp c. , corbel , s. , tzioumis t. , fender r. et al . , 2010 , atel , 2400 cadolle bel m. , rodriguez j. , davanzo p.
et al . , 2011 ,
a&a , 534 , 119 casella p. , belloni t. , homan j. , stella l. , 2004 , a&a , 426 , 587 casella p. , belloni t. , homan j. , stella l. , 2005 , apj , 629 , 403 chakrabarti s. k. , & dsilva , s. , 1994 , apj , 424 , 138 chakrabarti s.k .
, titarchuk l.g . , 1995 ,
apj , 455 , 623 chakrabarti s.k . & manickam , s.g . ,
2000 , apj , 531 , l41 chakrabarti , s. k. , nandi , a. , manickam , s. , et al . , 2002 , apj , 579 , l21 chakrabarti s. k. , acharya k. , molteni d. , 2004 , a&a , 421 , 1 chakrabarti s. k. , mandal s. , 2006 , apj , 642 , 49 chakrabarti s. k. , debnath d. , nandi a. & pal p. s. , 2008 , a&a , 489 , l41 chakrabarti s. k. , dutta b. g. , & pal p. s. , 2009 , mnras , 394 , 1463 corral - santana j. m. , casares j. , shahbaz t. , et al . , 2011 , mnras , 413 , l15 das s. , chattopadhyay i. , nandi a. , molteni d. , 2014 , mnras , 442 , 251 de villiers j. p. , hawley j. f. , krolik j. h. , hirose s. , 2005 , apj , 620 , 878 debnath d. , chakrabarti s. k. , nandi a. , 2008 , basi , 36 , 151 debnath d. , chakrabarti s. k. , nandi a. , 2013 , adspr , 52 , 2143 dunn r. j. h. , fender r. p. , kording e. g. , et al .
, 2011 , mnras , 403 , 61 dunn r. j. h. , fender r. p. , kording e. g. , et al . , 2011 , mnras , 411 , 337 ebisawa k. , ogawa m. , aoki t. , et al . , 1994 , pasj , 46 , 375 esin a. a. , mcclintock j. e. , narayan r. , 1997 , apj , 489 , 865 farinelli r. , amati l. , shaposhnikov n. , et al .
, 2013 , mnras , 428 , 3295 fender r. p. , belloni t. , gallo e. , 2004 , mnras , 355 , 1105 fender r. p. , homan j. , belloni t. , 2009 , mnras , 396 , 1307 feroci m. , matt g. , pooley g. , costa e. , et al . , 1999 , a&a , 351 , 985 garnavich p. m. , stanek k. z. , berlind p. , 1999
, iauc , 7276 gierlinski m. , done c. , 2004 , mnras , 347 , 885 giri k. , chakrabarti s. k. , 2013 , mnras , 430 , 2836 homan j. , belloni t. , 2005 ap&ss , 300 , 107 ingram a. , done c. , 2011 , mnras , 415 , 2323 lee , s. , ryu , d. , chattopadhyay , i. , 2011 , apj , 728 , 142 mandal s. , chakrabarti s. k. , 2010 , apj , 710 , l147 markwardt c. b. , 2001 , astrophysics & space science , 276 , 209 markwardt c. b. , 2009 , aspc , 411 , 251 mcclintock j. e. , remilard r. a. , 2006 , ` black hole binaries ' , compact stellar x - ray sources , edited by lewin w. h. g. and m. van der klis mcclintock j. e. , remilard r. a. , rupen m. p. , et al . , 2009 , apj , 698 , 1398 mccollough m. l. , wilson c. a. , 1999 , iauc , 7282 mckinney j. c. , gammie c. f. , 2004 , apj , 611 , 977 meier d. l. , nakamura m. , proceedings 3-d signatures in stellar explosions , ed .
p. hoeflich , p. kumar , c. wheeler , 2004 , cambridge university press , 219 merloni a. , fabian a. c. and ross r. r. , 2000 , mnras , 313 , 193 meyer f. , liu b. f. , meyer - hofmeister e. , 2007 , a&a , 463 , 1 meyer - hofmeister e. , liu b .
f , meyer f. , 2009 , a&a , 508 , 329 miller - jones j. c. a. , sivakoff g. r. , altamirano d. , et al . , 2012 ,
mnras , 421 , 468 miyamoto s. , kimura k. , kitamoto s. , et al . , 1991 , apj , 383 , 784 molteni d. , sponholz h. , chakrabarti s. k. , 1996 , apj , 457 , 805 motta s. , belloni t. , homan j. , 2009 , mnras , 400 , 1603 motta s. , munoz - darias t. , casella p. , et al . , 2011 , mnras , 418 , 2292 munoz - darias t. , motta s. and belloni t. , 2010 , mnras , 410 , 679 nandi a. , chakrabarti s. k. , vadawale s. v. , rao a. r. , 2001 , a&a , 380 , 245 nandi a. , debnath d. , mandal s. , chakrabarti s. k. , 2012 , a&a , 542 , 56 nandi , a. & radhika .
d , 2012 , cospar scientific assembly , mysore , india nandi , a. , radhika .
d , seetha . s. , 2013 , asi conference series , 8 , 71 ( arxiv : 1308.4567 ) nandi a. et al .
( in prep . ) okuda t. , teresi v. , molteni d. , 2007 , mnras , 377 , 1431 pooley g. g. , hjellming r. m. , 1999 , iauc , 7278 radhika .
d , nandi a. , s. seetha , 2013 , mmsai , 84 , 624 ( arxiv : 1301.7234 ) radhika .
( in prep . )
remilard r. a. et al . , 1999 , apj , 552 , 397 rodriguez j. , prat l. , 2008 , proceedings of science , 7th integral workshop ( arxiv : 0811:3519 ) rodriguez j. , varniere p. , 2011 ,
apj , 735 , 79 ryu d. , chakrabarti s. k. , molteni d. , 1997 , apj , 474 , 378 shakura n. i. , sunyaev r. a. , 1973 , a&a , 24 , 337 shaposhnikov n , titarchuk l. , 2007 , apj , 663 , 445 shaposhnikov n , titarchuk l.,2009 , apj , 699 , 453 sobczak g. j. , mcclintock j. e. , remilard r. a. , et al . , 2000 , apj , 531 , 537 soleri p. , belloni t. , casella p. , 2008 ,
mnras , 383 , 10989 stiele h. , belloni t. m. , kalemci e. , motta s. , 2013 , mnras , 429 , 2655 tagger m. , pellat r. , 1999 , a&a , 349 , 1003 titarchuk l. g. , 1994 , apj , 434 , 570 titarchuk , l. g. , fiorito , r. 2004 , apj , 612 , 988 titarchuk l. , shaposhnikov n. , & arefiev v. , 2007 , apj , 660 , 556 titarchuk l. , shaposhnikov n. , 2010 , apj , 724 , 1147 titarchuk l. , osherovich v. , 2000 , apj , 542 , l111 vadawale s. v. , rao a. r. , nandi a. , chakrabarti s. k. , 2001 , a&a , 370 , l17 wagner r. m. , schmidt g. d. , shrader c. r. , 1999 , iauc , 7279 wood a. , smith d. a. , marshall f. e. , swank j. h. , 1999 , iauc , 7274 yang j. , brocksopp c. , corbel s. , et al .
, 2010 , mnras , 409 , l64 zhang w. , jahoda k. , swank j. h. , et al .
, 1995 , apj , 449 , 930 | we re - investigated the ` spectro - temporal ' behaviour of the source xte j1859@xmath0226 in x - rays during its outburst phase in 1999 , by analysing the rxte pca / hexte data in 2 - 150 kev spectral band .
detailed analysis shows that the source evolves through different spectral states during its entire outburst as indicated by the variation in the spectral and temporal characteristics .
although the evolution pattern of the outburst followed the typical q - shaped profile , we observed an absence of ` canonical ' soft state and a weak presence of ` secondary ' emission during the decay phase of the outburst .
the broad - band spectra , modeled with high energy cutoff , shows that the fold - energy increases monotonically in the hard and hard - intermediate states followed by a random variation in the soft - intermediate state .
we attempted to estimate the mass of the source based on the evolution of quasi - periodic oscillation ( qpo ) frequencies during rising phase modeled with the propagating oscillatory shock solution , and from the correlation of photon index and qpo frequency .
it is also observed that during multiple ejections ( observed as radio flares ) the qpo frequencies are not present in the power spectra and there is an absence of lag in the soft to hard photons .
the disk flux increases along with a decrease in the high energy flux , implying the soft nature of the spectrum .
these results are the ` possible ' indication that the inner part of the disk ( _ i.e. , comptonized corona _ ) , which could be responsible for the generation of qpo and for the non - thermal comptonized component of the spectrum , is disrupted and the matter gets evacuated in the form of jet .
we attempted to explain the complex behaviour of ` spectro - temporal ' properties of the source during the entire outburst and the nature of the disk - jet connection before , during and after the ejection events in the context of two different types of accreting flow material , in presence of magnetic field .
black holes , accretion disks , radiation hydrodynamics , x - ray sources , stars : individual ( xte j1859@xmath0226 ) , disk - jet connection = 0.5 cm |
since the early days of high - energy gamma - ray astronomy it has been clear that the gamma - ray flux received at earth is dominated by emission from the galactic disk @xcite .
this emission is believed to arise from cosmic - ray interactions with the interstellar medium , which at gamma - ray energies @xmath0 mev are dominated by the decay of @xmath1 produced in collisions between cosmic - ray nuclei and the interstellar medium @xcite .
further contributions are from cosmic - ray electrons undergoing inverse compton scattering off interstellar soft photons and bremsstrahlung losses within the interstellar medium .
gamma - ray observations thus have the potential to map cosmic - ray acceleration sites in our galaxy which may ultimately help to identify the sources of cosmic - ray acceleration .
nearby galaxies have the advantage of being viewed from outside and so line of sight confusion , which complicates studies of emission from the galactic disk , is diminished .
this advantage is however somewhat offset by the limitations by the angular resolution and sensitivity of the instrument .
the large magellanic cloud ( lmc ) is thus an excellent target for studying the link between cosmic - ray acceleration and gamma - ray emission since the galaxy is nearby ( @xmath2 kpc ) @xcite , has a large angular extent of @xmath3 , and is seen at a small inclination angle of @xmath4 @xcite that avoids source confusion .
in addition , the lmc is relatively active , housing many supernova remnants , bubbles and superbubbles , and massive star forming regions that are all potential sites of cosmic - ray acceleration @xcite .
the egret telescope onboard the compton gamma - ray observatory ( _ cgro _ , 19912000 ) was the first to detect the lmc @xcite .
due to egret s limited angular resolution and limited sensitivity , details of the spatial structure of the gamma - ray emission could not be resolved , yet the observations showed some evidence for the spatial distribution being consistent with the morphology of radio emission . the large area telescope ( lat ) aboard _ fermi _ is providing now for the first time the capabilities to go well beyond the study of the integrated gamma - ray flux from the lmc @xcite . in this contribution
we present our first in - depth analysis of the lmc galaxy based on 11 months of continuous sky survey observations performed with _ fermi_/lat .
we put a particular emphasis on the determination of the spatial distribution of the gamma - ray emission , that , as we will show , reveals the distribution of cosmic rays in the galaxy . a more detailed discussion of the observations and their implications is given in @xcite .
the characteristics and performance of the lat aboard _ fermi _ are described in detail by @xcite .
the data used in this work amount to 274.3 days of continuous sky survey observations over the period august 8th 2008
july 9th 2009 during which a total exposure of @xmath5 @xmath6 s ( at 1 gev ) is obtained for the lmc .
events satisfying the standard low - background event selection ( `` diffuse '' events ) @xcite and coming from zenith angles @xmath7 ( to greatly reduce the contribution by earth albedo gamma rays ) are used . to further reduce the effect of earth albedo backgrounds , the time intervals when the earth was appreciably within the field of view ( specifically , when the center of the field of view was more than @xmath8 from the zenith ) are excluded from this analysis . furthermore , time intervals when the spacecraft was within the south atlantic anomaly are also excluded .
we further restrict the analysis to photon energies above 200 mev ; below this energy the effective area in the diffuse class " is relatively small and strongly dependent on energy .
all analysis is performed using the lat science tools package , which is available from the fermi science support center , using p6_v3 post - launch instrument response functions ( irfs ) .
these take into account pile - up and accidental coincidence effects in the detector subsystems that are not considered in the definition of the pre - launch irfs . at the galactic latitude of the lmc ( @xmath9 ) ,
the gamma - ray background is a combination of extragalactic and galactic diffuse emissions , some residual instrumental background and a number of point sources that primarily are associated to blazars .
we model the background using components for the diffuse galactic and the extragalactic and residual instrumental backgrounds and 6 point sources that all are associated with known blazars .
the galactic component is based on the lat standard diffuse background model gll_iem_v02 for which we keep the overall normalization as a free parameter .
the extragalactic and residual instrumental backgrounds are combined into a single component which has been taken as being isotropic .
the spectrum of this component is determined by fitting an isotropic component together with a model of the galactic diffuse emission and point sources to the data . also here
we leave the overall normalization of the component as a free parameter .
the 6 background blazars are modelled as point sources with power - law spectral shapes .
the flux and spectral power - law index of each source are left as free parameters of our background model and their values were determined from likelihood analysis . to investigate the spatial distribution of gamma - ray emission toward the lmc we show in fig .
[ fig : map ] an adaptively smoothed background subtracted counts map of the lat data .
the map clearly shows extended emission that is spatially confined to within the lmc boundaries which we trace by the iso column density contour @xmath10 h @xmath11 of neutral hydrogen in the lmc @xcite .
the total number of excess 200 mev 20 gev photons above the background in the lmc area amounts to @xmath12 counts whereas the background in the same area amounts to @xmath13 counts . with these statistics
, the extended gamma - ray emission from the lmc can be resolved into several components .
the brightest emission feature is located near @xmath14 , which is close to the massive star - forming region 30 doradus ( 30 dor ) that houses the two crab - like pulsars and @xcite .
excess gamma - ray emission is also seen toward the north and the west of 30 dor .
these bright regions are embedded into a more extended and diffuse glow that covers an area of approximately @xmath15 .
figure [ fig:30dor ] present a zoom of fig .
[ fig : map ] of the 30 dor region over which we overlay potential sources of gamma - ray emission . large region of the background subtracted counts map around the central star cluster r 136 in 30 dor .
stars show the locations of r 136 and of the pulsars and .
the circles show the 95% containment radius of sources s1 and s3 of the point source model , and of source g2 of the 2d gaussian shaped model ( see text ) . during month 4 of our dataset ,
a flaring point source occured near 30 dor , and we indicate the 95% containment radius of this source ; the diamond shows the location of the possible counterpart rx j0546.8@xmath166851 of the flaring source.,width=340 ] as next step , we assess the spatial distribution of the lmc emission using ( 1 ) simple parametrized geometrical models of the gamma - ray intensity distribution , and ( 2 ) spatial templates that trace the interstellar matter distribution in the lmc . we assume power - law spectral distributions for all models and keep the total flux and power law index as free parameters . we adjust
the spatial and spectral parameters of the models using a binned maximum likelihood analysis with spatial bins of @xmath17 and 60 logarithmically spaced energy bins covering the energy range 200 mev 20 gev .
we quantify the goodness - of - fit using the so - called _ test statistic _
( ts ) which is defined as twice the difference between the log - likelihood @xmath18 that is obtained by fitting the model on top of the background model to the data , and the log - likelihood @xmath19 that is obtained by fitting the background model only , i.e. @xmath20 .
first , we examine if the gamma - ray emission from the lmc can be explained with a combination of individual point sources . for this purpose
we add successive point sources to our model and optimize their locations , fluxes and spectral indices by maximizing the likelihood of the model .
we stop this procedure once the ts improvement after adding a further point source drops below 25 .
this happens after we added 5 point source to our model , resulting in ts @xmath21 .
the left panel of fig .
[ fig : geo ] shows the corresponding model counts map .
second , instead of using point sources we repeat the procedure with 2d gaussian shaped intensity profiles to build a geometrical model that is more appropriate for extended and diffuse emission structures .
we again stop the successive addition of 2d gaussian shaped sources once the ts improvement after adding a further source drops below 25 .
this occurs after two 2d gaussian shaped sources have been added to the model , resulting in ts @xmath22 .
we show the corresponding model counts map in the right panel of fig .
[ fig : geo ] .
obviously , the 2d gaussian model provides a larger ts than the point source model , suggesting that it better fits the data despite the smaller number of free parameter ( 10 for the 2d gaussian model compared to 20 for the point source model ) .
it is thus more likely that the lmc emission is indeed diffuse in nature , or alternatively , composed of a large number of unresolved and faint sources that can not be detected individually by _ fermi_/lat .
third , we fit the lat data to gas maps of neutral atomic hydrogen ( ) @xcite , molecular hydrogen ( ) @xcite , and ionized hydrogen ( ) @xcite . fitting the and
maps result in ts values of @xmath23 and @xmath24 , respectively , that are considerably worse than those obtained for the geometrical models .
apparently , the and maps provide rather poor fits to the data , indicating that the distribution of gamma rays does not follow the distribution of neutral hydrogen in the lmc .
this can already be seen from the left and mid panels of fig .
[ fig : tracer ] which show the gas maps after convolution with the lat point spread function . while the gamma - ray emissivity is highest in 30 dor and the northern part of the galaxy , the gas maps show a bright ridge that runs over @xmath25 along @xmath26 which coincides with the most prominent region of @xmath27co emission tracing giant molecular clouds in the lmc @xcite .
roughly @xmath28 of the total gas mass in the lmc is confined into this ridge @xcite , yet comparison with fig .
[ fig : map ] shows that most of the ridge is not luminuous in high - energy gamma rays .
the map , on the other hand , gives ts @xmath29 , which is very close to the ts values of the geometrical models .
the map thus provides the best fit among all of the gas maps to the lat data , and the right panel of fig .
[ fig : tracer ] indeed shows that the model is very similar to the 2d gaussian shaped model and also follows closely the observed distribution of gamma rays ( fig .
[ fig : map ] ) . to determine the spectrum of the gamma - ray emission from the lmc independently from any assumption on the spectral shape
, we fit our data in 6 logarithmically spaced energy bins covering the energy range 200 mev - 20 gev .
we obtain the total spectrum of the lmc by fitting the template to the data .
we also obtain separate spectra for the lmc disk ( g1 ) and for 30 dor ( g2 ) by fitting the 2d gaussian shaped model to the data . to determine the integrated gamma - ray flux we fit exponentially cut off power law spectral models of the form @xmath30 to the data .
we make these fits by means of a binned maximum likelihood analysis over the energy range 200 mev - 20 gev .
this results in an extrapolated @xmath31 mev photons flux of @xmath32 for the template which corresponds to an energy flux of @xmath33 ( systematic uncertainties in these estimates amount to less than 16% ) . dividing our spectra by the spatially integrated hydrogen column density @xmath34 h - atom @xmath11 sr of the lmc provides us with the differential gamma - ray emissivity per hydrogen atom .
we compare the resulting emissivity for the total lmc emission and the lmc disk only in fig .
[ fig : spectrum ] to the one obtained for the galactic local interstellar medium by @xcite .
apparently , the local galactic emissivity is between @xmath35 to @xmath36 times larger than the average emissivity of the lmc .
-decay ( long dashed ) , bremsstrahlung ( short dashed ) , and inverse compton emission ( dotted).,width=309 ] we compare the differential gamma - ray emissivities to a one - zone model of cosmic - ray interactions with the interstellar medium that takes into account @xmath1 decay following proton - proton interactions , bremsstrahlung from cosmic - ray electrons and inverse compton scattering of cosmic - ray electrons on lmc optical and infrared photons and cosmic microwave background photons @xcite . fitting this spectral model to our data using a binned maximum likelihood analysis gives an average cosmic - ray enhancement factor of @xmath37 for the entire lmc , and of @xmath38 for the lmc disk only .
systematic errors due to uncertainties in the effective area of the instrument amount to @xmath39 .
an additional systematic error of @xmath40 to @xmath41 comes from the uncertainty in the total gas mass of the lmc , which largely dominates the statistical and systematic measurements errors .
to reveal the sites of cosmic - ray acceleration in the lmc we map the cosmic - ray density variations in the galaxy by computing the gamma - ray emissivity as function of position .
we do this by dividing our background subtracted counts map by the @xmath42 map after convolution of the latter with the lat instrumental response function .
we normalise @xmath42 to a total lmc hydrogen mass of @xmath43 that takes into account the possible presence of dark gas that is not seen in radio surveys of @xcite .
we adaptively smooth @xcite the counts maps and used the resulting smoothing kernel distribution to smooth also the convolved @xmath42 map before the division to reveal significant structures at all possible scales , while supressing the noise that arises from the limited photon counting statistics .
the resulting emissivity maps are shown in fig .
[ fig : emissivitymap ] .
we superimpose on the images the interstellar gas distribution , as traced by @xmath42 , convolved with the lat instrumental response function , and also show the locations of potential particle acceleration sites , such as pulsars , supernova remnants , wolf - rayet stars and supergiant shells .
figure [ fig : emissivitymap ] reveals that the cosmic - ray density varies considerably over the disk of the lmc .
the gamma - ray emissivity is highest in 30 dor and the northern part of the galaxy , while the southern part and in particular the dense ridge of gas south of 30 dor seems basically devoid of cosmic rays .
these large variations confirm our earlier findings that the gamma - ray emission correlates little with the gas density in the lmc .
figure [ fig : emissivitymap ] suggests further that the cosmic - ray density correlates with massive star forming tracers , and in particular wolf - rayet stars and supergiant shells .
this finding is corroborated by the good fit of the gas map , which is probably the most direct tracer of massive star forming regions within a galaxy .
thus , the gamma - ray emissivity maps of the lmc support the idea that cosmic rays are accelerated in massive star forming regions as a result of the large amounts of kinetic energy that are input by the stellar winds and supernova explosions of massive stars into the interstellar medium .
our data reveal a relatively tight confinement of the gamma - ray emission to star forming regions , which suggests a relatively short diffusion length for gev protons .
observations of the lmc by _ fermi_/lat have for the first time provided a detailed map of high - energy gamma - ray emission from that galaxy .
our analysis reveals the massive star forming region 30 doradus as bright source of gamma - ray emission in the lmc in addition to fainter emission regions found in the northern part of the galaxy .
the gamma - ray emission from the lmc shows very little correlation with gas density .
a much better correlation is seen between gamma - ray emission and massive star forming regions , as traced by the ionizing gas , wolf - rayet stars and supergiant shells , and we take this as evidence for cosmic - ray acceleration in these regions .
this correlation supports the idea that cosmic rays are accelerated in massive star forming regions as a result of the large amounts of kinetic energy that are input by the stellar winds and supernova explosions of massive stars into the interstellar medium .
continuing observations of the lmc with _
fermi_/lat in the upcoming years will provide the photon statistics to learn more about the origin of the gamma - ray emission from that galaxy .
better statistics will help in identifying more individual emission components and may help to separate true point sources from the more diffuse emission that we expect from cosmic - ray interactions .
the _ fermi _ lat collaboration acknowledges generous ongoing support from a number of agencies and institutes that have supported both the development and the operation of the lat as well as scientific data analysis .
these include the national aeronautics and space administration and the department of energy in the united states , the commissariat lenergie atomique and the centre national de la recherche scientifique / institut national de physique nuclaire et de physique des particules in france , the agenzia spaziale italiana and the istituto nazionale di fisica nucleare in italy , the ministry of education , culture , sports , science and technology ( mext ) , high energy accelerator research organization ( kek ) and japan aerospace exploration agency ( jaxa ) in japan , and the k. a. wallenberg foundation , the swedish research council and the swedish national space board in sweden . additional support for science analysis during the operations phase is gratefully acknowledged from the istituto nazionale di astrofisica in italy and the and the centre national dtudes spatiales in france . | we report on observations of the large magellanic cloud with the fermi gamma - ray space telescope .
the lmc is clearly detected with the large area telescope ( lat ) and for the first time the emission is spatially well resolved in gamma - rays .
our observations reveal the massive star forming region 30 doradus as a bright source of gamma - ray emission in the lmc .
the observations furthermore show that the gamma - ray emission correlates little with the gas density of the lmc .
implications of this finding will be discussed . |
one of the key goals in modern cosmology is to study the assembly process of the first galaxies , and understand how the first stars and stellar systems formed at the end of the cosmic dark ages , a few hundred million years after the big bang . with the formation of the first stars , the so - called population iii ( pop iii )
, the universe was rapidly transformed into an increasingly complex , hierarchical system , due to the energy and heavy elements they released into the intergalactic medium ( igm ; for recent reviews , see barkana & loeb 2001 ; miralda - escud ; bromm & larson 2004 ; ciardi & ferrara 2005 ; glover 2005 ) .
currently , we can directly probe the state of the universe roughly a million years after the big bang by detecting the anisotropies in the cosmic microwave background ( cmb ) , thus providing us with the initial conditions for subsequent structure formation .
complementary to the cmb observations , we can probe cosmic history all the way from the present - day universe to roughly a billion years after the big bang , using the best available ground- and space - based telescopes . in between lies the remaining frontier , and the first galaxies are the sign - posts of this early , formative epoch .
there are a number of reasons why addressing the formation of the first galaxies and understanding second - generation star formation is important .
first , a rigorous connection between well - established structure formation models at high redshift and the properties of present - day galaxies is still missing .
an understanding of how the first galaxies formed could be a crucial step towards undertanding the formation of more massive systems .
second , the initial burst of pop iii star formation may have been rather brief due to the strong negative feedback effects that likely acted to self - limit this formation mode ( madau et al .
2001 ; ricotti & ostriker 2004 ; yoshida et al . 2004 ; greif & bromm 2006 ) .
second - generation star formation , therefore , might well have been cosmologically dominant compared to pop iii stars . despite their importance for cosmic evolution , e.g. , by possibly constituting the majority of sources for the initial stages of reionization at @xmath1 , we currently do not know the properties , and most importantly the typical mass scale , of the second - generation stars that formed in the wake of the very first stars . finally , a subset of second - generation stars , those with masses below @xmath2 , would have survived to the present day .
surveys of extremely metal - poor galactic halo stars therefore provide an indirect window into the pop iii era by scrutinizing their chemical abundance patterns , which reflect the enrichment from a single , or at most a small multiple of , pop iii supernovae ( sne ; christlieb et al .
2002 ; beers & christlieb 2005 ; frebel et al .
. stellar archaeology thus provides unique empirical constraints for numerical simulations , from which one can derive theoretical abundance patterns to be compared with the data . focusing on numerical simulations as the key driver of structure formation theory ,
the best strategy is to start with cosmological initial conditions , follow the evolution up to the formation of a small number ( @xmath3 ) of pop iii stars , and trace the ensuing expansion of sn blast waves together with the dispersal and mixing of the first heavy elements , towards the formation of second - generation stars out of enriched material ( greif et al .
2007 ; wise & abel 2007a ) . in this sense
some of the most pressing questions are : how does radiative and mechanical feedback by the very first stars in minihalos affect the formation of the first galaxies ? how and when does metal enrichment govern the transition to low - mass star formation ?
is there a critical metallicity at which this transition occurs ?
how does turbulence affect the chemical mixing and fragmentation of the gas ?
these questions have been addressed with detailed numerical simulations as well as analytic arguments over the last few years , and we here review some of the most recent work . for consistency , all quoted distances are physical , unless noted otherwise .
feedback by the very first stars in minihalos plays an important role for the subsequent build - up of the first galaxies . among the most prominent mechanisms
are ionizing and molecule - dissociating radiation emitted by massive pop iii.1 stars , as well as the mechanical and chemical feedback exerted by the first sne . in the next few sections ,
we briefly discuss these mechanisms in turn .
star formation in primordial gas is believed to produce very massive stars . during their brief lifetimes of @xmath4
, they produce @xmath5 ionizing photons per stellar baryon and thus have a significant impact on their environment ( bromm et al . 2001b ; schaerer 2002 ) . based on the large optical depth measured by the _ wilkinson microwave anisotropy probe _ ( _ wmap _ ) after one year of operation ,
wyithe & loeb ( 2003 ) suggested that the universe was reionized by massive metal - free stars .
even though the reionization depth according to _ wmap _ 5 decreased significantly ( komatsu et al . 2008 ; nolta et al .
2008 ) , recent reionization studies indicate that massive stars are still required ( schleicher et al .
considering different reionization scenarios with and without additional physics like primordial magnetic fields , they showed that stellar populations according to a scalo - type initial mass function ( imf ; scalo 1998 ) are ruled out within @xmath6 , unless very high star formation efficiencies of order @xmath7 are adopted . on the contrary , populations of very massive stars or mixed populations
can easily provide the required optical depth .
apart from their ionizing flux , pop iii.1 stars also emit a strong flux of h@xmath8-dissociating lyman - werner ( lw ) radiation ( bromm et al . 2001b ; schaerer 2002 ) .
thus , the radiation from the first stars dramatically influences their surroundings , heating and ionizing the gas within a few kiloparsec around the progenitor , and destroying the h@xmath8 and hd molecules locally within somewhat larger regions ( ferrara 1998 ; kitayama et al .
2004 ; whalen et al . 2004 ; alvarez et al . 2006 ; abel et al .
2007 ; johnson et al .
additionally , the lw radiation emitted by the first stars could propagate across cosmological distances , allowing the build - up of a pervasive lw background radiation field ( haiman et al .
the impact of radiation from the first stars on their local surroundings has important implications for the numbers and types of pop iii stars that form .
the photoheating of gas in the minihalos hosting pop iii.1 stars drives strong outflows , lowering the density of the primordial gas and delaying subsequent star formation by up to @xmath9 ( whalen et al .
2004 ; johnson et al . 2007 ; yoshida et al . 2007 ) .
furthermore , neighboring minihalos may be photoevaporated , delaying star formation in such systems as well ( shapiro et al .
2004 ; susa & umemura 2006 ; ahn & shapiro 2007 ; greif et al . 2007
; whalen et al .
the photodissociation of molecules by lw photons emitted from local star - forming regions will , in general , act to delay star formation by destroying the main coolants that allow the gas to collapse and form stars .
the photoionization of primordial gas , however , can ultimately lead to the production of copious amounts of molecules within the relic h ii regions surrounding the remnants of pop iii.1 stars ( figure 1 ; see also ricotti et al .
2001 ; oh & haiman 2002 ; nagakura & omukai 2005 ; johnson & bromm 2007 ) .
recent simulations tracking the formation of , and radiative feedback from , individual pop iii.1 stars in the early stages of the assembly of the first galaxies have demonstrated that the accumulation of relic h ii regions has two important effects .
first , the hd abundance that develops in relic h ii regions allows the primordial gas to re - collapse and cool to the temperature of the cmb , possibly leading to the formation of pop iii.2 stars in these regions ( johnson et al .
2007 ; yoshida et al . 2007 ; greif et al .
second , the molecule abundance in relic h ii regions , along with their increasing volume - filling fraction , leads to a large optical depth to lw photons over physical distances of the order of several kiloparsecs .
the development of a high optical depth to lw photons over such short length - scales suggests that the optical depth to lw photons over cosmological scales may be very high , acting to suppress the build - up of a background lw radiation field , and mitigating negative feedback on star formation .
even absent a large optical depth to lw photons , pop iii.1 stars in minihalos may readily form at @xmath10 . while star formation in more massive systems may proceed relatively unimpeded , through atomic line cooling , during the earliest epochs of star formation these atomic - cooling halos are rare compared to the minihalos which host individual pop iii stars .
although the process of star formation in atomic cooling halos is not well understood , for a broad range of models the dominant contribution to the lw background is from pop iii.1 stars formed in minihalos at @xmath11 .
therefore , at these redshifts the lw background radiation may be largely self - regulated , with pop iii.1 stars producing the very radiation which , in turn , suppresses their formation .
johnson et al .
2008 argue that there is a critical value for the lw flux , @xmath12 , at which pop iii.1 star formation occurs self - consistently , with the implication that the pop iii.1 star formation rate in minihalos at @xmath10 is decreased by only a factor of a few .
simulations of the formation of the first galaxies at @xmath13 which take into account the effect of a lw background at @xmath14 show that pop iii.1 star formation takes place before the galaxy is fully assembled , suggesting that the formation of the first galaxies does indeed take place after chemical enrichment by the first sn explosions ( greif et al .
2007 ; wise & abel 2007a ; johnson et al . 2008
; whalen et al .
2008b ) . and
hd molecules .
the light and dark shades of blue denote regions with a free electron fraction of @xmath15 and @xmath16 , respectively , while the shades of green denote regions with an h@xmath8 fraction of @xmath17 , @xmath18 , and @xmath19 , in order of decreasing brightness .
the regions with the highest molecule abundances lie within relic h ii regions , which thus play an important role for subsequent star formation , allowing molecules to become shielded from photodissociating radiation and altering the cooling properties of the primordial gas ( see johnson et al .
2007).,width=453 ] numerical simulations have indicated that pop iii.1 stars might become as massive as @xmath20 ( omukai & palla 2003 ; bromm & loeb 2004 ; yoshida et al .
2006 ; oshea & norman 2007 ) . after their main - sequence lifetimes of typically @xmath21
, stars with masses below @xmath22 are thought to collapse directly to black holes without significant metal ejection , while in the range @xmath23 a pair - instability supernova ( pisn ) disrupts the entire progenitor , with explosion energies ranging from @xmath24 , and yields of order @xmath25 ( heger & woosley 2002 ; heger et al .
less massive primordial stars with a high degree of angular momentum might explode with similar energies , but as jet - like hypernovae ( umeda & nomoto 2002 ; tominaga et al .
the significant mechanical and chemical feedback effects exerted by such explosions have been investigated with a number of detailed calculations , but these were either performed in one dimension ( salvaterra 2004 ; kitayama & yoshida 2005 ; machida et al . 2005 ; whalen et al 2008b ) , or did not start from realistic initial conditions ( bromm et al .
2003 ; norman et al .
recent work treated the full three - dimensional problem in a cosmological context at the cost of limited resolution , finding that the sn remnant propagated for a hubble time at @xmath26 to a final mass - weighted mean shock radius of @xmath27 , roughly half the size of the h ii region ( greif et al . 2007 ) . due to the high explosion energy ,
the host halo was entirely evacuated .
additional simulations in the absence of a sn explosion were performed to investigate the effect of photoheating and the impact of the sn shock on neighboring minihalos . for the case discussed in greif et al .
( 2007 ) , the sn remnant exerted positive mechanical feedback on neighboring minihalos by shock - compressing their cores , while photoheating marginally delayed star formation .
although a viable theoretical possibility , secondary star formation in the dense shell via gravitational fragmentation ( e.g. machida et al . 2005 ; mackey et al . 2003 ; salvaterra et al .
2004 ) was not observed , primarily due to the previous photoheating by the progenitor and the rapid adiabatic expansion of the shocked gas .
the dispersal of metals by the first sn explosions transformed the igm from a simple , pure h / he gas to one with ubiquituous metal enrichment .
the resulting cooling ultimately enabled the formation of the first low - mass stars
the key question is then when and where this transition occurred .
as indicated in the previous section , such a transition could only occur well after the explosion , as cooling by metal lines or dust requires the gas to re - collapse to high densities .
furthermore , the distribution of metals becomes highly anisotropic , since the shocked gas expands preferentially into the voids around the host halo .
due to the high temperature and low density of the shocked gas , dark matter ( dm ) halos with @xmath28 must be assembled to efficiently mix the gas .
for this reason , the first galaxies likely mark the formation environments of the first low - mass stars and stellar clusters ( see section 5 ) .
how massive were the first galaxies , and when did they emerge ?
theory predicts that dm halos containing a mass of @xmath29 and collapsing at @xmath30 were the hosts for the first bona fide galaxies .
these systems are special in that their associated virial temperature exceeds the threshold , @xmath31 , for cooling due to atomic hydrogen ( oh & haiman 2002 ) .
these so - called ` atomic - cooling halos ' did not rely on the presence of molecular hydrogen to enable cooling of the primordial gas .
in addition , their potential wells were sufficiently deep to retain photoheated gas , in contrast to the shallow potential wells of minihalos ( madau et al .
2001 ; mori et al . 2002 ; dijkstra et al .
these are arguably minimum requirements to set up a self - regulated process of star formation that comprises more than one generation of stars , and is embedded in a multi - phase interstellar medium . in this sense , we will term all objects with a viral temperature exceeding @xmath32 as a first galaxy
( see figure 2 ) .
an important consequences of atomic cooling is the softening of the equation of state below the virial radius , allowing a fraction of the potential energy to be converted into kinetic energy ( wise & abel 2007b ) .
perturbations in the gravitational potential can then generate turbulent motions on galactic scales , which are transported to the center of the galaxy . in this context
the distinction between two fundamentally different modes of accretion becomes important .
gas accreted directly from the igm is heated to the virial temperature and comprises the sole channel of inflow until cooling in filaments becomes important .
this mode is termed hot accretion , and dominates in low - mass halos at high redshift .
the formation of the virial shock and the concomitant heating are visible in figure 3 , where we show the hydrogen number density and temperature of the central @xmath33 ( comoving ) around the center of a first galaxy ( greif et al .
this case also reveals a second mode , termed cold accretion .
it becomes important as soon as filaments are massive enough to enable molecule reformation , which allows the gas to cool and flow into the nascent galaxy with high velocities .
these streams create a multitude of unorganized shocks near the center of the galaxy and could trigger the gravitational collapse of individual clumps ( figure 4 ) . in concert with metal enrichment by previous star formation in minihalos , chemical mixing might be highly efficient and could lead to the formation of the first low - mass star clusters ( clark et al .
2008 ) , in extreme cases possibly even to metal - poor globular clusters ( bromm & clarke 2002 ) .
some of the extremely iron - deficient , but carbon and oxygen - enhanced stars observed in the halo of the milky way may thus have formed as early as redshift @xmath34 .
( comoving ) of a simulation depicting the assembly of a first galaxy , shown at three different output times .
white crosses denote pop iii.1 star formation sites in minihalos , and the insets approximately delineate the boundary of the galaxy , further enlarged in figures 3 and 4 .
_ top row : _ the hierarchical merging of dm halos leads to the collapse of increasingly massive structures , with the least massive progenitors forming at the resolution limit of @xmath35 and ultimately merging into the first galaxy with @xmath36 .
the brightest regions mark halos in virial equilibrium according to the commonly used criterion @xmath37 .
although the resulting galaxy is not yet fully virialized and is still broken up into a number of sub - components , it shares a common potential well and the infalling gas is attracted towards its center of mass .
_ middle row : _ the gas generally follows the potential set by the dm , but pressure forces prevent collapse in halos below @xmath38 ( cosmological jeans criterion ) .
moreover , star formation only occurs in halos with virial masses above @xmath39 , as densities must become high enough for molecule formation and cooling . _
bottom row : _ the virial temperature of the first star - forming minihalo gradually increases from @xmath40 to @xmath41 , at which point atomic cooling sets in ( see greif et al .
2008b).,width=453 ] ( comoving ) of a simulation depicting the assembly of a first galaxy .
shown is the hydrogen number density ( _ left - hand side _ ) and temperature ( _ right - hand side _ ) in a slice centered on the galaxy .
the dashed lines denote the virial radius at a distance of @xmath42 .
hot accretion dominates where gas is accreted directly from the igm and shock - heated to @xmath41 .
in contrast , cold accretion becomes important as soon as gas cools in filaments and flows towards the center of the galaxy , such as the streams coming from the left- and right - hand side .
they drive a prodigious amount of turbulence and create transitory density perturbations that could in principle become jeans - unstable .
in contrast to minihalos , the initial conditions for second - generation star formation are highly complex ( see greif et al .
2008b).,width=453 ] ( comoving ) of a simulation depicting the assembly of a first galaxy .
shown is the divergence ( _ left - hand side _ ) and z - component of the vorticity ( _ right - hand side _ ) in a slice centered on the galaxy .
the dashed lines denote the virial radius at a distance of @xmath42 .
the most pronounced feature in the left - hand panel is the virial shock , where the ratio of infall speed to local sound speed approaches unity and the gas decelerates over a comparatively small distance .
in contrast , the vorticity at the virial shock is almost negligible .
the high velocity gradients at the center of the galaxy indicate the formation of a multitude of shocks where the bulk radial flows of filaments are converted into turbulent motions on small scales ( see greif et al .
related to the issue of metal enrichment , an important question is what controls the transition from a population of very massive stars to a distribution biased towards low - mass stars . in a seminal paper ,
bromm et al . ( 2001a ) performed simulations of the collapse of cold gas in a top - hat potential that included the metallicity - dependent effects of atomic fine - structure cooling . in the absence of molecular cooling
, they found that fragmentation suggestive of a present - day imf only set in at metallicities above a threshold value of @xmath43 .
however , they noted that the neglect of molecular cooling could be significant .
omukai et al .
( 2005 ) argued , based on the results of their detailed one - zone models , that molecular cooling would indeed dominate the cooling over many orders of magnitude in density .
the effects of molecular cooling at densities up to @xmath44 have been discussed by jappsen et al .
( 2007a ) in three - dimensional collapse simulations of warm ionized gas in minihalos for a wide range of environmental conditions .
this study used a time - dependent chemical network running alongside the hydrodynamic evolution as described in glover & jappsen ( 2007 ) .
the physical motivation was to investigate whether minihalos that formed within the relic h ii regions left by neighboring pop iii stars could form subsequent generations of stars themselves , or whether the elevated temperatures and fractional ionizations found in these regions suppressed star formation until larger halos formed . in this study , it was found that molecular hydrogen dominated the cooling of the gas for abundances up to at least @xmath45 .
in addition , there was no evidence for fragmentation at densities below @xmath46 .
jappsen et al .
( 2007b ) showed that gas in simulations with low initial temperature , moderate initial rotation , and a top - hat dm overdensity , will readily fragment into multiple objects , regardless of metallicity , provided that enough h@xmath47 is present to cool the gas .
rotation leads to the build - up of massive disk - like structures in these simulations , which allow smaller - scale fluctuations to grow and become gravitationally unstable
. the resulting mass spectrum of fragments peaks at a few hundred solar masses , roughly corresponding to the thermal jeans mass in the disk - like structure ( see figure 5 ) .
these results suggest that the initial conditions adopted by bromm et al .
( 2001a ) may have determined the result much more than might have been appreciated at the time . to make further progress in understanding the role that metal - line cooling plays in promoting fragmentation , it is paramount to develop a better understanding of how metals mix with pristine gas in the wake of the first galaxies . and
@xmath48 , from top to bottom ( see also jappsen 2007b ) .
the temperature evolution of primordial gas is very similar to the @xmath49 case , showing that metal - line cooling becomes important only for very high metallicities .
the fraction of low - mass fragments increases with higher metallicity , since more gas can cool to the temperature of the cmb before becoming jeans - unstable .
however , the fragments are still very massive , suggesting that metal - line cooling might not be responsible for the transition to low - mass star formation . instead , this transition might be governed by dust cooling occurring at higher densities . for more definitive conslusions
, one must perform detailed numerical simulations that follow the collapse to higher densities in a realistic cosmological environment.,width=453 ]
the discovery of extremely metal - poor subgiant stars in the galactic halo with masses below one solar mass ( christlieb et al . 2002 ; beers & christlieb 2005 ; frebel et al .
2005 ) indicates that the transition from primordial , high - mass star formation to the ` normal ' mode of star formation that dominates today occurs at abundances considerably smaller than the solar value . at the extreme end ,
these stars have iron abundances less than @xmath50 , and carbon or oxygen abundances that are still @xmath51 the solar value .
these stars are thus strongly iron deficient , which could be due to unusual abundance patterns produced by enrichment from pop iii stars ( umeda & nomoto 2002 ) , or due to mass transfer from a close binary companion ( ryan et al .
2005 ; komiya et al . 2007 ) .
recent work has shown that there are hints for an increasing binary fraction with decreasing metallicity ( lucatello et al .
. however , if metal enrichment is the key to the formation of low - mass stars , then logically there must be some critical metallicity @xmath52 at which the formation of low - mass stars first becomes possible .
however , the value of @xmath52 is a matter of ongoing debate .
as discussed in the previous sections , some models suggest that low - mass star formation becomes possible only once atomic fine - structure line cooling from carbon and oxygen becomes effective ( bromm et al .
2001a ; bromm & loeb 2003 ; santoro et al . 2006
; frebel et al . 2007 ) , setting a value for @xmath52 at around @xmath53 .
another possibility is that low - mass star formation is a result of dust - induced fragmentation occurring at high densities , and thus at a very late stage in the protostellar collapse ( schneider et al . 2002 ; omukai et al .
2005 ; schneider et al . 2006 ; tsuribe & omukai 2006 ) . in this model , @xmath54 , where much of the uncertainty in the predicted value results from uncertainties in the dust composition and the degree of gas - phase depletion ( schneider et al .
2002 , 2006 ) .
( 11.6,17.4 ) ( 0.0,11.6 ) of the gas cloud shortly before and shortly after the formation of the first protostar at @xmath55
. only gas at densities above @xmath56 is plotted .
the dynamical timescale at a density of @xmath57 is of the order of @xmath58 years .
dark dots indicate the location of protostars as identified by sink particles forming at @xmath59 .
note that without usage of sink particles to identify collapsed protostellar cores one would not have been able to follow the build - up of the protostellar cluster beyond the formation of the first object .
there are @xmath60 protostars when we stop the calculation at @xmath61 .
they occupy a region roughly a hundredth of the size of the initial cloud ( see clark et al .
2008).,title="fig:",width=219,height=219 ] ( 5.8,11.6 ) of the gas cloud shortly before and shortly after the formation of the first protostar at @xmath55 . only gas at densities above @xmath56 is plotted .
the dynamical timescale at a density of @xmath57 is of the order of @xmath58 years .
dark dots indicate the location of protostars as identified by sink particles forming at @xmath59 .
note that without usage of sink particles to identify collapsed protostellar cores one would not have been able to follow the build - up of the protostellar cluster beyond the formation of the first object .
there are @xmath60 protostars when we stop the calculation at @xmath61 .
they occupy a region roughly a hundredth of the size of the initial cloud ( see clark et al .
2008).,title="fig:",width=219,height=219 ] ( 0.0 , 5.8 ) of the gas cloud shortly before and shortly after the formation of the first protostar at @xmath55
. only gas at densities above @xmath56 is plotted .
the dynamical timescale at a density of @xmath57 is of the order of @xmath58 years .
dark dots indicate the location of protostars as identified by sink particles forming at @xmath59 .
note that without usage of sink particles to identify collapsed protostellar cores one would not have been able to follow the build - up of the protostellar cluster beyond the formation of the first object .
there are @xmath60 protostars when we stop the calculation at @xmath61 .
they occupy a region roughly a hundredth of the size of the initial cloud ( see clark et al .
2008).,title="fig:",width=219,height=219 ] ( 5.8 , 5.8 ) of the gas cloud shortly before and shortly after the formation of the first protostar at @xmath55
. only gas at densities above @xmath56 is plotted .
the dynamical timescale at a density of @xmath57 is of the order of @xmath58 years .
dark dots indicate the location of protostars as identified by sink particles forming at @xmath59 .
note that without usage of sink particles to identify collapsed protostellar cores one would not have been able to follow the build - up of the protostellar cluster beyond the formation of the first object .
there are @xmath60 protostars when we stop the calculation at @xmath61 .
they occupy a region roughly a hundredth of the size of the initial cloud ( see clark et al .
2008).,title="fig:",width=219,height=219 ] ( 0.0,0.0 ) of the gas cloud shortly before and shortly after the formation of the first protostar at @xmath55
. only gas at densities above @xmath56 is plotted .
the dynamical timescale at a density of @xmath57 is of the order of @xmath58 years .
dark dots indicate the location of protostars as identified by sink particles forming at @xmath59 .
note that without usage of sink particles to identify collapsed protostellar cores one would not have been able to follow the build - up of the protostellar cluster beyond the formation of the first object .
there are @xmath60 protostars when we stop the calculation at @xmath61 .
they occupy a region roughly a hundredth of the size of the initial cloud ( see clark et al .
2008).,title="fig:",width=219,height=219 ] ( 5.8,0.0 ) of the gas cloud shortly before and shortly after the formation of the first protostar at @xmath55
. only gas at densities above @xmath56 is plotted .
the dynamical timescale at a density of @xmath57 is of the order of @xmath58 years .
dark dots indicate the location of protostars as identified by sink particles forming at @xmath59 .
note that without usage of sink particles to identify collapsed protostellar cores one would not have been able to follow the build - up of the protostellar cluster beyond the formation of the first object .
there are @xmath60 protostars when we stop the calculation at @xmath61 .
they occupy a region roughly a hundredth of the size of the initial cloud ( see clark et al .
2008).,title="fig:",width=219,height=219 ] ( 1.9,11.9 ) ( 7.7,11.9 ) ( 1.9,6.1 ) ( 7.7,6.1 ) ( 1.9,0.3 ) ( 7.7,0.3 ) in recent work , clark et al .
( 2008 ) focused on dust - induced fragmentation in the high - density regime , with @xmath62 .
they modeled star formation in the central regions of low - mass halos at high redshift adopting an equation of state ( eos ) similar to omukai et al .
( 2005 ) , finding that enrichment of the gas to a metallicity of only @xmath63 dramatically enhances fragmentation .
a typical time evolution is illustrated in figure 6 .
it shows several stages in the collapse process , spanning a time interval from shortly before the formation of the first protostar ( as identified by the formation of a sink particle in the simulation ) to @xmath64 years afterwards . during the initial contraction
, the cloud builds up a central core with a density of about @xmath65 .
this core is supported by a combination of thermal pressure and rotation .
eventually , the core reaches high enough densities to go into free - fall collapse , and forms a single protostar .
as more high angular momentum material falls to the center , the core evolves into a disk - like structure with density inhomogeneities caused by low levels of turbulence .
as it grows in mass , its density increases . when dust - induced cooling sets in , it fragments heavily into a tightly packed protostellar cluster within only a few hundred years .
one can see this behavior in particle density - position plots in figure 7 .
the simulation is stopped @xmath64 years after the formation of the first stellar object ( sink particle ) . at this point
, the core has formed @xmath60 stars .
the time between the formation of the first and second protostar is roughly @xmath66 years , which is two orders of magnitude higher than the free - fall time at the density where the sinks are formed .
note that without the inclusion of sink particles , one would only have been able to capture the formation of the first collapsing object which forms the first protostar : the formation of the accompanying cluster would have been missed entirely .
( 18,12.5 ) ( 1.0,6.5 ) simulation , the graphs show the densities of the particles , plotted as a function of their position . note that for each plot , the particle data has been centered on the region of interest .
results are plotted for three different output times , ranging from the time that the first star forms ( @xmath55 ) to @xmath67 years afterwards .
the densities lying between the two horizontal dashed lines denote the range over which dust cooling lowers the gas temperature .
we also plot the mass function for a metallicity of @xmath63 and mass resolution @xmath68 and @xmath69 , respectively .
note the similarity between the results of the low - resolution and high - resolution simulations .
the onset of dust cooling in the @xmath63 cloud results in a stellar cluster which has a mass function similar to that for present - day stars , in that the majority of the mass resides in the low - mass objects .
this contrasts with the @xmath70 and primordial clouds , in which the bulk of the cluster mass is in high - mass stars ( see clark et al .
2008).,title="fig:",width=192,height=192 ] ( 7.0,6.5 ) simulation , the graphs show the densities of the particles , plotted as a function of their position .
note that for each plot , the particle data has been centered on the region of interest .
results are plotted for three different output times , ranging from the time that the first star forms ( @xmath55 ) to @xmath67 years afterwards .
the densities lying between the two horizontal dashed lines denote the range over which dust cooling lowers the gas temperature .
we also plot the mass function for a metallicity of @xmath63 and mass resolution @xmath68 and @xmath69 , respectively .
note the similarity between the results of the low - resolution and high - resolution simulations .
the onset of dust cooling in the @xmath63 cloud results in a stellar cluster which has a mass function similar to that for present - day stars , in that the majority of the mass resides in the low - mass objects .
this contrasts with the @xmath70 and primordial clouds , in which the bulk of the cluster mass is in high - mass stars ( see clark et al .
2008).,title="fig:",width=192,height=192 ] ( 1.0,0.5 ) simulation , the graphs show the densities of the particles , plotted as a function of their position .
note that for each plot , the particle data has been centered on the region of interest .
results are plotted for three different output times , ranging from the time that the first star forms ( @xmath55 ) to @xmath67 years afterwards .
the densities lying between the two horizontal dashed lines denote the range over which dust cooling lowers the gas temperature .
we also plot the mass function for a metallicity of @xmath63 and mass resolution @xmath68 and @xmath69 , respectively .
note the similarity between the results of the low - resolution and high - resolution simulations .
the onset of dust cooling in the @xmath63 cloud results in a stellar cluster which has a mass function similar to that for present - day stars , in that the majority of the mass resides in the low - mass objects .
this contrasts with the @xmath70 and primordial clouds , in which the bulk of the cluster mass is in high - mass stars ( see clark et al .
2008).,title="fig:",width=192,height=192 ] ( 7.0,0.5 ) simulation , the graphs show the densities of the particles , plotted as a function of their position .
note that for each plot , the particle data has been centered on the region of interest .
results are plotted for three different output times , ranging from the time that the first star forms ( @xmath55 ) to @xmath67 years afterwards .
the densities lying between the two horizontal dashed lines denote the range over which dust cooling lowers the gas temperature .
we also plot the mass function for a metallicity of @xmath63 and mass resolution @xmath68 and @xmath69 , respectively .
note the similarity between the results of the low - resolution and high - resolution simulations .
the onset of dust cooling in the @xmath63 cloud results in a stellar cluster which has a mass function similar to that for present - day stars , in that the majority of the mass resides in the low - mass objects .
this contrasts with the @xmath70 and primordial clouds , in which the bulk of the cluster mass is in high - mass stars ( see clark et al .
2008).,title="fig:",width=201,height=206 ] the fragmentation of low - metallicity gas in this model is the result of two key features in its thermal evolution .
first , the rise in the eos curve between densities @xmath71 and @xmath72 causes material to loiter at this point in the gravitational contraction .
a similar behavior at densities around @xmath73 is discussed by bromm et al .
( 2001a ) .
the rotationally stabilized disk - like structure , as seen in the plateau at @xmath74 in figure 7 , is able to accumulate a significant amount of mass in this phase and only slowly increases in density .
second , once the density exceeds @xmath75 , the sudden drop in the eos curve lowers the critical mass for gravitational collapse by two orders of magnitude .
the jeans mass in the gas at this stage is only @xmath76 .
the disk - like structure suddenly becomes highly unstable against gravitational collapse and fragments vigorously on timescales of several hundred years .
a very dense cluster of embedded low - mass protostars builds up , and the protostars grow in mass by accretion from the available gas reservoir .
the number of protostars formed by the end of the simulation is nearly two orders of magnitude larger than the initial number of jeans masses in the cloud setup . because the evolutionary timescale of the system is extremely short the free - fall time at a density of @xmath57 is of the order of 10 years
none of the protostars that have formed by the time that the simulation is stopped have yet commenced hydrogen burning .
this justifies neglecting the effects of protostellar feedback in this study .
heating of the dust due to the significant accretion luminosities of the newly - formed protostars will occur ( krumholz 2006 ) , but is unlikely to be important , as the temperature of the dust at the onset of dust - induced cooling is much higher than in a typical galactic protostellar core ( @xmath77 or more , compared to @xmath78 in the galactic case ) .
the rapid collapse and fragmentation of the gas also leaves no time for dynamo amplification of magnetic fields ( tan & blackman 2004 ) , which in any case are expected to be weak and dynamically unimportant in primordial and very low metallicity gas ( widrow 2002 ) .
however , other authors suggest that the biermann battery effect may amplify weak initial fields such that the magneto - rotational instability can influence the further collapse of the star ( silk & langer 2006 ) .
simulations by xu et al .
( 2008 ) show that this effect yields peak magnetic fields of @xmath79 in the center of star - forming minihalos .
jets and outflows may reduce the final stellar mass by @xmath80 ( machida et al .
2006 ) . in the presence of primordial fields ,
the magnetic pressure may even prevent star formation in minihalos and thus increase the mass scale of star - forming objects ( schleicher et al .
2008a , b ) .
the mass functions of the protostars at the end of the @xmath63 simulations ( both high and low resolution cases ) are shown in figure 7 .
when the simulation is terminated , collapsed cores hold @xmath81 of gas in total .
the mass function peaks somewhere below @xmath82 and ranges from below @xmath83 to about @xmath84 .
this is not the final protostellar mass function .
the continuing accretion of gas by the cluster will alter the mass function , as will mergers between the newly - formed protostars ( which can not be followed using our current sink particle implementation ) .
protostellar feedback in the form of winds , jets and h ii regions may also play a role in determining the shape of the final stellar mass function .
however , a key point to note is that the chaotic evolution of a bound system such as this cluster ensures that a wide spread of stellar masses will persist
. some stars will enjoy favorable accretion at the expense of others that will be thrown out of the system ( as can be seen in figure 6 ) , thus having their accretion effectively terminated ( see the discussions in bonnell & bate 2006 ; bonnell et al .
the survival of some of the low - mass stars formed in the cluster is therefore inevitable .
the forming cluster represents a very extreme analogue of the clustered star formation that we know dominates in the present - day universe ( lada & lada 2003 ) .
a mere @xmath64 years after the formation of the first object , the cluster has formed @xmath60 stars ( see figure 6 ) . these occupy a region of only around @xmath85 , or @xmath86 , in size , roughly a hundredth of the size of the initial cloud . with @xmath81
accreted at this stage , the stellar density is @xmath87 .
this is about five orders of magnitude greater than the stellar density in the trapezium cluster in orion ( hillenbrand & hartmann 1998 ) and about a thousand times greater than that in the core of 30 doradus in the large magellanic cloud ( massey & hunter 1998 ) .
this means that dynamical encounters will be extremely important during the formation of the first star cluster .
the violent environment causes stars to be thrown out of the denser regions of the cluster , slowing down their accretion .
the stellar mass spectrum thus depends on both the details of the initial fragmentation process ( e.g. as discussed by clark & bonnell 2005 ; jappsen et al .
2005 ) as well as dynamical effects in the growing cluster ( bonnell et al .
2001 , 2004 ) .
this is different to present - day star formation , where the situation is less clear - cut and the relative importance of these two processes may vary strongly from region to region ( krumholz et al . 2005 ; bonnell & bate 2006 ; bonnell et al .
2007 ) . in future work
, it will be important to assess the validity of the initial conditions adopted for the present study , ideally by performing cosmological simulations that simultaneously follow the formation of the first galaxies and the metal enrichment by primordial sne in minihalos .
understanding the formation of the first galaxies marks the frontier of high - redshift structure formation .
it is crucial to predict their properties in order to develop the optimal search and survey strategies for the _ jwst_. whereas _ ab - initio _ simulations of the very first stars can be carried out from first principles , and with virtually no free parameters , one faces a much more daunting challenge with the first galaxies .
now , the previous history of star formation has to be considered , leading to enhanced complexity in the assembly of the first galaxies .
one by one , all the complex astrophysical processes that play a role in more recent galaxy formation appear back on the scene . among them
are external radiation fields , comprising uv and x - ray photons , as well as local radiative feedback that may alter the star formation process on small scales .
perhaps the most important issue , though , is metal enrichment in the wake of the first sn explosions , which fundamentally alters the cooling and fragmentation properties of the gas .
together with the onset of turbulence ( wise & abel 2007b ; greif et al .
2008b ) , chemical mixing might be highly efficient and could lead to the formation of the first low - mass stars and stellar clusters ( clark et al .
2008 ) . in this sense
a crucial question is whether the transition from pop iii to pop ii stars is governed by atomic fine - structure or dust cooling .
theoretical work has indicated that molecular hydrogen dominates over metal - line cooling at low densities ( jappsen et al .
2007a , b ) , and that fragment masses below @xmath88 can only be attained via dust cooling at high densities ( omukai et al .
2005 ; clark et al .
on the other hand , observational studies seem to be in favor of the fine - structure based model ( frebel et al .
2007 ) , even though existing samples of extremely metal - poor stars in the milky way are statistically questionable ( christlieb et al . 2002 ; beers & christlieb 2005 ; frebel et al . 2005 ) .
moreover , these studies assume that their abundances are related to primordial star formation a connection that is still debated ( lucatello et al . 2005 ; ryan et al .
2005 ; komiya et al . 2007 ) . in light of these uncertainties , it is essential to push numerical simulation to ever lower redshifts and include additional physics in the form of radiative feedback , metal dispersal ( greif et al .
2008a ) , chemisty and cooling ( e.g. glover & jappsen 2007 ) and the effects of magnetic fields ( e.g. xu et al . 2008 ) .
we are confident that a great deal of interesting discoveries , both theoretical and observational , await us in the rapidly growing field of early galaxy formation .
the authors would like to thank the organisers of the iau symposium 255 for a very enjoyable and stimulating conference .
drgs thanks the lgfg for financial support .
pcc acknowledges support by the deutsche forschungsgemeinschaft ( dfg ) under grant kl 1358/5 .
rsk thanks for partial support from the emmy noether grant kl 1358/1 .
vb acknowledges support from nsf grant ast-0708795 .
drgs , pcc , rsk and tg acknowledge subsidies from the dfg sfb 439 , galaxien i m frhen universum .
drgs and tg would like to thank the heidelberg graduate school of fundamental physics ( hgsfp ) for financial support .
the hgsfp is funded by the excellence initiative of the german government ( grant number gsc 129/1 ) . | the formation of the first galaxies at redshifts @xmath0 signaled the transition from the simple initial state of the universe to one of ever increasing complexity .
we here review recent progress in understanding their assembly process with numerical simulations , starting with cosmological initial conditions and modelling the detailed physics of star formation . in this context
we emphasize the importance and influence of selecting appropriate initial conditions for the star formation process .
we revisit the notion of a critical metallicity resulting in the transition from primordial to present - day initial mass functions and highlight its dependence on additional cooling mechanisms and the exact initial conditions .
we also review recent work on the ability of dust cooling to provide the transition to present - day low - mass star formation .
in particular , we highlight the extreme conditions under which this transition mechanism occurs , with violent fragmentation in dense gas resulting in tightly packed clusters . |
the study of the density matrices of identical particles ( bosons or fermions ) moving freely in a box @xcite is generalized in this paper to the case of identical particles in a parabolic confinement potential with either harmonic interactions between the particles , or with an anisotropy induced by a homogeneous magnetic field on top of the parabolic confinement .
this model , giving rise to repetitive gaussian integrals , allows to derive an analytical expression for the generating function of the partition function . for an ideal gas of non - interacting particles in a parabolic well , this generating function coincides with the grand - canonical partition function . with interactions ,
the calculation of this generating function circumvents the constraints on the summation over the cycles of the permutation group .
moreover , it allows one to calculate the canonical partition function recursively for the system with harmonic two - body interactions .
the theory is developed both for fermions and for bosons . in view of the recent interest in bose - einstein condensation in a trap @xcite ,
more attention has been payed to the boson case in the discussion of the results .
the model system discussed here has already been studied in the context of quantum dots with operator techniques , and the eigenvalues and eigenstates were calculated including the effect of harmonic two - body interactions and in the presence of a magnetic field @xcite . however , to the best of our knowledge neither the boson case nor the thermodynamics seems to have been analyzed previously .
it should also be mentioned that the idea of first expanding the hilbert space to the configuration space and then projecting onto the appropriate subspace by group theoretical means has been used recently @xcite in the context of quantum dots to study the ground state correlations for fermions and bosons .
another motivation to perform the present analytical calculations lies in our path integral formulation of the density matrices for @xmath0 particles@xcite .
indeed , particles in a parabolic potential are a favorite testing ground for the path integral method@xcite .
it should be noted that for our formulation the existence of a positive measure over a well defined domain in the @xmath1 configuration space is essential in view of any algorithmic approach to the problem . in the present paper , which is in essence analytical , integrations over the configuration space
are performed .
the reason is that the extension of the state space to the configuration space makes the gaussian integrals tractable .
the permutation symmetry leads to summations over the cycles which are performed using the generating function technique , which is one of the main results of the present paper . the model of @xmath0 identical particles in a parabolic
well , in the presence of a magnetic field and with harmonic repulsive or attractive two - body interactions has its intrinsic value , since it constitutes an exactly soluble idealization of atoms in a magnetic trap .
it should be stressed that the association of identical particles with each three oscillator degrees of freedom makes the model @xmath2 . without bose - einstein or fermi - dirac statistics ,
i.e. for `` distinguishable '' particles , the model is equivalent with @xmath3 one dimensional oscillators , because each degree of freedom decouples in such a way that there is no difference in statistical behavior between @xmath3 @xmath4 oscillators and @xmath0 @xmath2 oscillators @xcite .
this paper is organized as follows : the calculation technique is explained in the next section . in the subsequent section we repeat the same calculation for the model with a homogeneous magnetic field . in section
iv the results for @xmath4 bosons and fermions are given .
in section v bosons in @xmath5 are analyzed in some detail , and in the last section the conclusions are given .
in this section we calculate the partition function of @xmath0 identical particles with the following lagrangian including one - body and two - body potentials @xmath6 ( atomic units are used . )
the potentials can be rewritten in terms of the center - of - mass coordinate @xmath7 and coordinates @xmath8 describing the coordinates of the particles measured from the center of mass @xmath9 from which @xmath10 with @xmath11 the requirement that @xmath12 has to be positive expresses the stability condition that the confining potential has to be strong enough to overcome the repulsion between the particles .
if an harmonic interparticle * attraction * is considered , the eigenfrequency @xmath12 would become @xmath13 and no stability condition has to be imposed on the confining potential .
notice that these transformations do neither diagonalize the lagrangian nor the hamiltonian , because the coordinates @xmath14 are _ not _ independent of the center - of - mass coordinate .
since the system consists in each direction of 1 degree of freedom with frequency @xmath15 and @xmath16 degrees of freedom with frequency @xmath17 the propagator @xmath18 for * distinguishable * particles ( indicated by the subscript @xmath19 for 3 dimensions and @xmath20 in 1 dimension ) can be calculated from the action expressed in the imaginary time variable , and it is of course a product of the propagators @xmath21 per component : @xmath22 where the column vector @xmath23 contains the @xmath24-components of the particles , i.e. @xmath25 and similarly for @xmath26 and @xmath27 knowing the propagator @xmath28 of a single harmonic oscillator with frequency @xmath29@xmath30 one finds for the 1-dimensional propagator @xmath21 of the @xmath0 distinguishable oscillators in the interacting system : @xmath31 where the factor @xmath32 in @xmath33 accounts for the mass @xmath0 ( in atomic units ) of the center .
the denominator in ( [ eqkdexplicit ] ) compensates for the fact that @xmath16 instead of @xmath0 degrees of freedom of frequency @xmath12 are available .
the 3-dimensional propagator @xmath34 ( [ eqkddef ] ) for @xmath0 distinguishable oscillators of the interacting system is , according to ( [ eqkdseparate ] ) and ( [ eqkdexplicit ] ) , given by @xmath35 @xmath36 where @xmath37 denotes a point in the configuration space @xmath38 i.e. @xmath39 . the symmetrized density matrix @xmath40 for 3d identical particles ( indicated by the subscript @xmath41 can be obtained by using the following projection , with @xmath42 denoting the permutation matrix : @xmath43 where @xmath44 for bosons and @xmath45 for fermions . it should be emphasized that @xmath42 acts on the particle indices , not on the components of @xmath46 separately .
the partition function is then readily obtained by integrating over the configuration space @xmath47 the remaining part of this section will be devoted to the explicit evaluation of this integral for the partition function .
the integration proceeds in 3 stages : the first stage deals with the center - of - mass treatment , the second one concerns the cyclic decomposition , and in the third step the summation over the cycles will be performed .
the center - of - mass coordinate @xmath7 does not only depend on the coordinates of all the particles , but it also has its own propagator . therefore substituting @xmath7 by its expression in terms of the particle positions and
then performing the integration seems not to be the most adequate way to deal with the integration over the configuration space .
instead , the following identity is used for the formal treatment of @xmath7 as an independent coordinate , at the expense of additional integrations : @xmath48 fourier transformation of the @xmath49-function then leads to @xmath50 where @xmath51 is a @xmath3 dimensional row vector . applying this transformation to the partition function @xmath52 and rearranging the factors one obtains @xmath53 this transformation makes @xmath7 independent of the particle positions relative to the center of mass .
the real dependence on the relative positions is reintroduced by the fourier transform .
it should be noted that the explicit dependence of the propagator ( [ eqkdproduct ] ) on @xmath54 and the presence of the factor @xmath55 are consequences of the two - body interactions .
the next step is to rewrite the sum over the permutations as a sum over all possible cycles .
this will be done in the next subsection .
an excellent example of such a decomposition into cycles has been given by feynman @xcite for a system of non - interacting particles in a box .
a permutation can be broken up into cycles .
suppose that a particular permutation contains @xmath56 cycles of length@xmath57 the positive integers @xmath56 and @xmath58 then have to satisfy the constraint @xmath59 furthermore , the number @xmath60 of cyclic decompositions with @xmath61 cycles of length @xmath62 @xmath63 @xmath56 cycles of length @xmath64 @xmath65 is known to be @xmath66 a cycle of length @xmath58 will be obtained from @xmath67 permutations .
therefore the sign factor @xmath68 can be decomposed as @xmath69 combining these result originating from the permutation symmetry one obtains @xmath70 @xmath71 the @xmath49-function expresses that the decomposition is cyclic .
it is obvious that @xmath72 which allows to analyze @xmath73 from its 1-dimensional constituents : @xmath74 using the semigroup property @xcite of the harmonic oscillator propagator @xmath75 all integrations but one can be performed @xmath76 where @xmath77 the integral ( [ eqkl1d ] ) is the propagator @xmath78 of a driven harmonic oscillator with the lagrangian @xmath79 studied in @xcite .
it should be noted again that without two - body interactions the driving force ( [ eqdriving ] ) is lacking . taking over the result from @xcite and integrating over the configuration space one obtains @xmath80 after straightforward algebra one obtains for the 1d function @xmath81 : @xmath82 and for its 3d extension : @xmath83 using ( [ eqmlsum ] ) one then is left with a sixfold integral for the partition function @xmath84{l } \int d{\bf r}\int \frac{d{\bf k}}{\left ( 2\pi \right ) ^3}e^{i{\bf k}\cdot { \bf r}}\frac{k\left ( \sqrt{n}{\bf r},\beta |\sqrt{n}{\bf r},0\right ) _ \omega } { k\left ( \sqrt{n}{\bf r},\beta |\sqrt{n}{\bf r},0\right ) _
w}\exp \left ( -\frac 1{4n}\frac{k^2}w\frac{1+e^{-\beta w}}{1-e^{-\beta w}}\right ) \\
\times \sum_{m_1\cdots m_n}\prod_\ell \frac{\xi ^{\left ( \ell -1\right ) m_\ell } } { m_\ell ! \ell ^{m_\ell } } \left ( \frac 1{2\sinh \frac 12\ell \beta w}% \right ) ^{3m_\ell } \end{array } .\ ] ] both the integrations over @xmath85 and @xmath7 are gaussian , leading to the following series for @xmath52 : @xmath86 without two - body interactions ( @xmath87 ) , @xmath88 is the partition function of a set of identical oscillators .
the partition function @xmath52 only differs from it by a center - of - mass correction and the actual values of @xmath89 the remaining summation over the cycles involves the constraint ( [ eqmlsum ] ) , which however can be removed by the use of the generating function technique , which will be considered in the next subsection . concentrating on the explicit dependence of @xmath88 on @xmath90 ( with @xmath12 considered as a parameter ) , one can construct the following generating function @xmath91 with @xmath92 by definition .
the partition function @xmath93 can then be obtained by taking the appropriate derivatives of @xmath94 with respect to @xmath95 assuming that the series for @xmath94 is convergent near @xmath96 : @xmath97 the summation over the number of cycles with length @xmath58 is now unrestricted , and can easily be performed : @xmath98 this series can be rewritten into the more familiar form @xmath99 it should be noted in view of the remarks at the end of the preceding section concerning @xmath88 , that @xmath100 is also the generating function of a model without two - body interactions . in that case @xmath12
equals @xmath101 and @xmath102 coincides with the well known @xcite grand canonical partition function of a set of identical particles in a parabolic well .
starting from the expression for @xmath102 derived in the previous subsection for the interacting model , a recursion relation can be obtained for @xmath88 . introducing : @xmath103 for brevity in the notations ,
we observe that @xmath104 considering next @xmath105 the product rule and an elementary binomial expansion can be used to find @xmath106 the corresponding 1-dimensional version of this recurrence relation ( indicated with the subscript @xmath107 to distinguish it from the 3d case with capital subscript ) becomes @xmath108 leading to the following partition functions in closed form for one - dimensional bosons and one - dimensional fermions @xmath109 it is easy to check that these partition functions are the solution of the recurrence relation for @xmath110 with @xmath111 for bosons and @xmath45 for fermions .
however we did not find a systematic method to obtain analytical solutions of this type of recurrence relations .
e.g. for the 3d case we had to rely on numerical schemes , as will be discussed below .
but at this stage , it is worthwhile first to consider the presence of an homogeneous magnetic field , as the origin of anisotropy in our model of @xmath0 identical oscillators .
the lagrangian of @xmath0 particles in a confining parabolic potential in the presence of a magnetic field is @xmath112 where @xmath113 is the cyclotron frequency . for this model ,
the calculations of the preceding section can in essence be repeated .
first the propagator for distinguishable particles is calculated
. the next step will be the projection on the irreducible representation of the permutation group , and performing the cyclic decomposition .
then the generating function is introduced to circumvent the constraints on the partition in cycles , and finally the summation over the cycles is performed .
the fact that the energy spectrum and the wavefunction can be calculated when harmonic interparticle interactions are included @xcite indicates that the propagator and the partition function for the model in a magnetic field with two - body interaction can be obtained using our methods .
the calculation technique is very demanding , therefore it seemed appropriate to illustrate the method at the hand of the simple model ( [ eqlmagn ] ) .
for distinguishable particles , the many - particle propagator is a product of one - particle propagators .
the one - particle propagator @xmath114 of this model can be calculated by stochastic techniques@xcite or by path integral techniques@xcite .
the evaluation is somewhat lengthy but straightforward , resulting eventually in @xmath115{l } \sqrt{\frac \omega { 2\pi \sinh \beta \omega } } \frac s{2\pi \sinh \beta s}% \exp \left\ { -\frac{\omega \left ( \left ( z^2+\left ( z^{\prime } \right ) ^2\right ) \cosh \beta \omega -2zz^{\prime } \right ) } { 2\sinh \beta \omega } % \right\ } \\ \times \exp \left\ { -\frac s2\frac{\left ( x^2+y^2+(x^{\prime } ) ^2+(y^{\prime } ) ^2\right ) \cosh
\beta s-2\left ( xx^{\prime } + yy^{\prime } \right ) \cosh \frac 12\beta \omega } { \sinh \beta s}\right\ } \\ \times \exp \left\ { -i\left ( \frac 12\omega _ c\left ( xy - x^{\prime } y^{\prime } \right ) -s\frac{\sinh \frac 12\beta \omega } { \sinh \beta s}\left ( y^{\prime } x - yx^{\prime } \right ) \right ) \right\ } , \end{array}\ ] ] with the eigenfrequency @xmath116 given by @xmath117 the partition function corresponding to this single - particle propagator is obtained by integrating over the configuration space : @xmath118 ^{-1 } \label{eqzmagn}\ ] ] which coincides with the partition function of a 3d harmonic oscillator if @xmath119 for @xmath0 identical particles , the propagator becomes @xmath120 with the corresponding partition function given by @xmath121 using the cyclic decomposition and the semigroup property as before , followed by the relaxation of the constraint ( [ eqmlsum ] ) on the number of cycles , one finds for the generating function @xmath122 because in this model with a magnetic field no two - body interactions were taken into account , this expression equals the grand - canonical partition function of the model if @xmath123 is interpreted as the fugacity , with @xmath124 the chemical potential .
we have thus obtained an expression for the grand - canonical partition function in the absence of two - body interactions for @xmath0 identical harmonic oscillators , both with and without a magnetic field . for a system with harmonic two - body interactions , repulsive or attractive ,
expressions for the canonical partition function are obtained ( in the absence of a magnetic field ) . the corresponding thermodynamic properties ( grand - canonical ground state occupancy , canonical and grand - canonical specific heat and average square radius ) of the confined system
will be calculated in the next section .
in the preceding sections we have obtained the grand canonical partition function of identical harmonic oscillators in a confining parabolic potential well without a magnetic field [ equation ( [ eqksinomagn ] ) ] and in the presence of a magnetic field [ equation ( [ eqksimagn ] ) ] .
for the same system the canonical partition function is obtained for an attractive as well as for a repulsive harmonic two - body interaction [ equation ( [ eqzgeneral ] ) ] .
the expressions are given for bosons and for fermions . in this section
we will analyze the thermodynamical properties of the confined system in the case of motion in 1d .
it should be noted that for the non - interacting case in 1d , we recover the results of ref . .
to the best of our knowledge , the interacting case has not been analyzed up to now for identical particles .
for distinguishable particles it has been studied before @xcite . in 1d , the explicit expressions for @xmath125 largely facilitate the analysis : the free energy @xmath126 the internal energy @xmath127 and the specific heat @xmath128 can be readily obtained .
the results are summarized in table i. the subscript @xmath129refers to the center - of - mass contribution .
the center - of - mass correction @xmath130 to the specific heat which is an excess specific heat and therefore may be negative is shown in fig .
1 . in fig .
2 we plot the contribution @xmath131 to the specific heat , due to the internal degrees of freedom .
note that @xmath131 is the same for 1d fermions and bosons ; bosonization implies this result @xcite .
the spatial extension of the cloud of interacting identical particles can be obtained by applying the hellman - feynman theorem to the free energy : @xmath132 the difference @xmath133 between the mean square length of the 1d fermion and boson system turns out to be : @xmath134 it is proportional to the number of particles , and inversely proportional to the frequency @xmath12 of the internal degrees of freedom .
the calculation of the thermodynamic properties or the mean square radius of the 3d cloud of identical particles is substantially complicated by the lack of an explicit expression in closed form for the partition function .
numerical methods have therefore to be used .
we concentrate this analysis on the boson case . for fermions
a different scheme will have to be developed and will be published later . the recurrence relation ( [ zirecur ] ) is not directly accessible for numerical computation , as can easily be seen by evaluating the expected dominant factor @xmath135 for bosons . for a relatively low temperature and a moderate number of particles ,
say @xmath136 and @xmath137 this factor is as small as @xmath138 we therefore isolate this factor using the following scaling for the partition function @xmath139 and rewrite the recurrence relation ( [ zirecur ] ) in terms of the activity @xmath140 ( see e.g. @xcite ) defined as @xmath141 where @xmath142 have been introduced for convenience .
the recurrence relation ( [ zirecur ] ) then becomes after some manipulations @xmath143 the corresponding recurrence relation for the internal energy of the internal degrees of freedom becomes @xmath144 the temperature scale used to express @xmath145 is @xmath146 the routine used to calculate the values @xmath147 and @xmath148 from these recurrence relations ( [ eqrhorecur ] ) and ( [ equrecur ] ) is displayed in table ii .
the values of @xmath147 for @xmath149 are shown in fig . 3 for the three temperatures @xmath150 1 and 2 . for @xmath151 it turns out that @xmath152 is about 8 times larger than @xmath153 for @xmath154 approaching @xmath155 which makes it extremely difficult to deal numerically with the values of the partition function rather than with the proportionality factors .
once @xmath140 is known , the internal energy @xmath156 is readily obtained from ( [ equrecur ] ) . the results for @xmath157 100 and 1000
are shown in fig .
the specific heat in the canonical ensemble can also be calculated from these parameters .
this is shown in fig .
5 , clearly illustrating the effect of condensation for a finite number of particles . anticipating the next subsection , it should be noted that the specific heat in the canonical ensemble for @xmath0 particles and without repulsive two - body interactions is identical to the specific heat in the grand canonical ensemble with the average number of particles given by the same value of @xmath158 the reason why we compare both ensembles without repulsive interactions is the requirement that in the grand canonical ensemble the system should be stable for any large number of particles , which is not the case in the gaussian model with repulsion , because the confining potential can only accommodate a finite number of particles as a consequence of ( [ eqeigenw ] )
. another interesting consequence of the repulsive interactions follows from the dependence of the condensation temperature @xmath159 on the number of particles , which obeys the following scaling law taking ( [ eqeigenw ] ) into account : @xmath160 for the case of attractive interactions and no confinement potential , @xmath159 is proportional to @xmath161 the condensation temperature for both cases is plotted in fig .
6 . for the gaussian model , the case of harmonic attraction does not pose any problem because in this model there is no sign of an `` extra '' collapse due to the nature of the interaction .
the only consequence seems to be that the condensation occurs at a much higher temperature than would be the case without two - body interactions or with a repulsive harmonic interaction .
the thermodynamic properties in the grand canonical ensemble of an ideal bose gas in a parabolic well and in the presence of a magnetic field can be obtained directly from ( [ eqzmagn ] ) and ( [ eqksimagn ] ) .
substituting @xmath162 by the fugacity @xmath163 the gibbs free energy @xmath164 for the boson case ( @xmath44 ) becomes : @xmath165 after a power series expansion of the denominators , the summation over the cycle lengths @xmath58 can be performed : @xmath166 as expected from a bose - einstein distribution with energy levels @xmath167 the average number of particles @xmath168 is given by @xmath169 and the fugacity can be eliminated in favor of the ground state occupancy @xmath170 : @xmath171 restoring the cyclic summation for reasons of numerical convergence , one obtains for the average number of particles @xmath172 which can be cast in the form of the following numerically tractable series @xmath173 for given @xmath155 standard numerical techniques can be used to determine @xmath174 , $ ] and hence the ground state occupancy @xmath175 which is shown in fig . 7 for @xmath176 and in fig . 8 for @xmath177 as a function of @xmath178 for several values of @xmath158 it turns out that the parabolic well is more important than the anisotropy due to the presence of the magnetic field .
the anisotropy only moderately influences the temperature dependence of the ground state occupancy .
the internal energy @xmath179 becomes @xmath180 and its numerical evaluation once @xmath181 is determined presents no numerical difficulties .
the resulting specific heat is shown in fig . 9 for @xmath182
comparison with fig .
5 reveals that the anisotropy due to the magnetic field essentially broadens the peak in the specific heat near the condensation temperature , but does not substantially alter the structure .
in this paper we applied the method of symmetrical density matrices , developed by feynman for a system of non - interacting particles in a box , to a system of harmonically interacting particles in a confining parabolic potential .
the interaction could be taken into account , due to the gaussian nature of the propagators , allowing integration over the configuration space .
the symmetrization resulting from the projection of the propagators for distinguishable particles on the appropriate representation of the permutation group gives rise to a series which could be summed using the generating function technique . without these generating functions
the calculation has to be restricted to a limited number of particles @xcite . using them
not only the grand canonical partition function @xmath94 could be obtained in the parameter range of the model where @xmath94 is well defined , but also the canonical partition functions @xmath183 for a given number @xmath0 of identical particles could be obtained as a recursion of partition functions of a smaller number of particles for the interacting system . the recurrence relation for the activity
@xmath184 i.e. the proportionality factor between @xmath183 and @xmath185 allows for an accurate numerical treatment of the thermodynamical quantities of the model , such as the internal energy and the specific heat . also the thermodynamic properties of the same model in the presence of a homogeneous magnetic field
could be investigated along these lines .
a detailed analysis of the ideal gas in a confining parabolic potential with anisotropy due to a magnetic field was presented ; special attention was payed to the relation between the number of condensed atoms , the magnetic field , the strength of the confinement potential and the total number of particles .
the relationship between the parameters of our model and the characteristics of atomic traps @xcite lies beyond the scope of the present paper .
it should be mentioned that in the absence of two - body interactions , the generating function can be identified as the grand canonical partition function . in that case
the specific heat as obtained from the grand canonical ensemble for an average number of @xmath186 particles calculated using the chemical potential equals the specific heat obtained from the canonical ensemble with the number of particles @xmath0 given by @xmath187 for a more elaborate discussion we refer to @xcite .
it should also be mentioned that in this the model it is assumed that the spin degrees of freedom are fixed .
this simplifying assumption is imposed by the symmetrization method , which becomes more involved if the spin degrees of freedom depend on the configuration of the particles . in summary ,
the partition function of a general gaussian model for bosons and fermions , with or without a magnetic field , has been calculated analytically , and the thermodynamical properties of this model have been studied .
particular attention has been given to the bose - einstein condensation in the presence of two - body interactions , repulsive and attractive , and in the presence of a magnetic field .
part of this work is performed in the framework of the nfwo projects no .
2.0093.91 , 2.0110.91 , g. 0287.95 and wo.073.94n ( wetenschappelijke onderzoeksgemeenschap , scientific research community of the nfwo on `` low - dimensional systems '' ) , and in the framework of the european community program human capital and mobility through contracts no .
chrx - ct93 - 0337 and chrx - ct93 - 0124 .
one of the authors ( f.b . ) acknowledges the national fund for scientific research for financial support .
.partition function , free energy , internal energy and specific heat of 1d identical particles in a parabolic confining potential with harmonic two - body interactions.[table1 ] [ cols="<,<,<,<,<",options="header " , ] .... subroutine recur(nmax ,
b , rho , u ) dimension rho(0:nmax),u(0:nmax ) rho(1)=1 u(1)=1.5*(1+b)/(1-b ) if ( nmax.le.1 ) return do n=1,nmax rho(n)=0 do m=0,n-2 rho(n)=(1-b**(m+1))**3/rho(m+1 ) * * ( rho(n)+ ( ( 1-b)/(1-b**(n - m ) ) ) * * 3 ) enddo rho(n)=(1+rho(n))/n * ( ( 1-b**n)/(1-b))**3 u(n)=0 do m=0,n-2 du=1.5*(n - m)*(1+b**(n - m))/(1-b**(n - m ) ) + u(m ) u(n)=(1-b**(m+1))**3/rho(m+1 ) * * ( u(n)+du * ( ( 1-b)/(1-b**(n - m ) ) ) * * 3 ) enddo u(n)=1./n / rho(n)*((1-b**n)/(1-b))**3 * * ( u(n)+1.5*(1+b)/(1-b)+u(n-1 ) ) enddo end .... fig . 1 : : : center - of - mass correction @xmath130 to the specific heat of 1d fermions and bosons
( see table i ) in units of the boltzmann constant @xmath188 as a function of the reduced temperature @xmath189 for various ratios of @xmath190 fig . 2 : : : specific heat contribution @xmath191 per particle for 1d fermions and bosons
( see table i ) in units of the boltzmann constant @xmath188 as a function of @xmath192 with @xmath193 the reduced temperature @xmath194 for several values of @xmath0 .
fig . 3 : : : values of @xmath147 for @xmath149 from the recurrence relation ( [ eqrhorecur ] ) for the three temperatures @xmath195 @xmath196 and @xmath197 .
fig . 4 : : : internal energy @xmath198 for @xmath199 as a function of @xmath200 fig . 5 : : : specific heat @xmath201 per particle in units of the boltzmann constant @xmath188 for @xmath199 as a function of @xmath200 fig . 6
: : : critical temperature @xmath159 as a function of the number of particles @xmath0 for a repulsive ( full line ) and an attractive ( dashed line ) two - body interaction .
: : : relative ground state occupancy @xmath202 as a function of @xmath203 with @xmath193 the reduced temperature @xmath204 for @xmath205 and for several values of @xmath158 fig . 8 : : : relative ground state occupancy @xmath202 as a function of @xmath203 with @xmath193 the reduced temperature @xmath204 for @xmath206 and for several values of @xmath0 .
: : : specific heat @xmath201 per particle in units of the boltzmann constant @xmath188 as a function of @xmath207 with @xmath193 the reduced temperature @xmath204 for @xmath208 and for several values of @xmath158 | a generalization of symmetrized density matrices in combination with the technique of generating functions allows to calculate the partition function of identical particles in a parabolic confining well .
harmonic two - body interactions ( repulsive or attractive ) are taken into account .
also the influence of a homogeneous magnetic field , introducing anisotropy in the model , is examined .
although the theory is developed for fermions and bosons , special attention is payed to the thermodynamic properties of bosons and their condensation . |
a multibeam receiver operating at a wavelength of 20 cm has recently been installed on the 64-m parkes radio telescope in australia .
this receiver permits simultaneous observation of 13 regions of sky , increasing the speed of surveys by the same factor . while the main motivation for the development of this system was to survey the sky for hi in the local universe
@xcite , it also enables very efficient surveys for pulsars .
the central observing frequency of 1374mhz is higher than usual for all - sky pulsar surveys , but is ideal for surveys at low galactic latitudes . at lower frequencies ,
sensitive surveys of the galactic plane are limited by the high background temperature , as well as broadening of the pulses caused by dispersion and scattering of the pulsed emission by the interstellar electron plasma .
all of these effects are strongly frequency - dependent and are much reduced at 1374mhz . the jodrell bank and parkes surveys of clifton et al .
( 1992 ) and johnston et al .
( 1992 ) were centred close to this frequency and both were particularly successful at finding many distant and young pulsars , several of which are also x - ray and/or gamma - ray emitters and some possibly associated with supernova remnants .
we are undertaking a survey for pulsars in an @xmath4-wide strip along the galactic plane using the multibeam receiver on the parkes telescope .
this survey is proving remarkably successful and the discovery of the first @xmath5 pulsars is presented elsewhere ( manchester et al .
, in preparation ) .
we briefly describe the experimental details of the survey in section 2 .
detailed timing observations to determine accurate pulsar positions , pulse periods , period derivatives and other parameters are an essential follow - up to any pulsar survey . timing observations and analysis procedures for the multibeam survey pulsars are described in section 3 .
finally , section 4 describes a particularly interesting pulsar , psr j1811@xmath01736 .
this pulsar , one of the first discovered in the survey , is in a very eccentric orbit with a companion which is probably another neutron star .
pulsars in such binary systems in the galactic plane are rare , there being only three known which are definite ( psrs j1518 + 4904 , b1534 + 12 and b1913 + 16 ) and one which is possible ( psr b1820@xmath011 ) .
such systems are of great interest because both stars act essentially as point masses and they make excellent systems for studying their dynamics with high precision ( e.g. damour & taylor 1992 )
. in particular , the discovery of more systems like these will allow additional sensitive tests of general relativity and alternative theories of gravity , more precise measurements of neutron star masses , and better estimates of the rate of coalescence of neutron - star binaries , events which are likely to be detected by future gravitational - wave detectors .
the survey of the galactic plane for pulsars using the multibeam receiver on the parkes radio telescope commenced in 1997 august .
the survey observations are made in a 288-mhz band centred on 1374mhz .
receivers for each of the 13 beams are dual - channel cryogenic systems sensitive to orthogonal linear polarisations .
the full width at half - power of each beam is approximately 14 arcmin and , at high galactic latitudes , the system noise temperature of each channel is about 21k .
a large filterbank system was constructed to process the data from the 13 beams .
the filterbank has 96 3-mhz channels covering the 288-mhz bandwidth for each polarisation of each beam , making a total of 2496 channels .
detected signals from individual frequency channels are added in polarisation pairs , high - pass filtered with a cutoff at approximately 0.2hz , integrated and 1-bit digitised every 250@xmath6s , and recorded on magnetic tape ( dlt ) for subsequent analysis .
it is planned that the survey will cover the region @xmath7 about the galactic plane from longitude @xmath8 to @xmath9 .
the relatively long observation time of 35min per pointing , combined with the excellent receiver noise performance and the wide bandwidth , gives a high sensitivity , with a limiting flux density for long - period pulsars of about 0.15mjy .
this is about seven times better than the previously most sensitive surveys of the galactic plane ( clifton et al .
1992 ; johnston et al .
survey observations are made on a hexagonal grid of pointings to give complete sky coverage , with adjacent beams overlapping at the half - power points .
a total of 2136 pointings of the 13 beams are required to cover the survey region . at the time of writing , approximately 60 per cent of these pointings have been observed .
analysis procedures are similar to those used in previous surveys ( e.g. manchester et al .
1996 ) with de - dispersion followed by transformation into the modulation frequency domain using a fast fourier transform ( fft ) and harmonic summing to improve sensitivity to the typically narrow pulses .
some periodic interference is present in the data and this is removed primarily by ignoring signals found repeatedly in undedispersed power spectra , both in simultaneous data from different beams as well as from single beams at many different times .
processing is currently being carried out on networks of workstations .
initially , a single coherent fft has been performed over each 35-min data set , giving reduced sensitivity to pulsars in short - period binary systems in which the period is varying due to the doppler effects of the pulsar motion .
the data are now being reprocessed in order to search for such accelerated signals .
pulsar candidates from the initial processing are scheduled for reobservation in order to confirm their identity as pulsars .
we use the centre beam of the multibeam system to make five 6-minute observations , at the nominal position and four surrounding points . in most cases ,
this gives detections at two or three positions , confirming the pulsar and allowing an improved position to be determined with an accuracy of about 2 arcmin . where no detection results from these observations , confirmation is attempted by means of a 35-min observation made at the nominal position .
the improved positions usually permit much shorter observation times in the follow - up timing measurements , because of the higher signal - to - noise ratio obtained when the pulsar is in the centre of the telescope beam .
furthermore , they allow a smaller range of parameter space to be searched in obtaining coherent timing solutions , reducing the overall number of observations required . at the time of writing ,
about 90 per cent of the observed pointings have been processed using the `` non - acceleration '' search code , whilst analysis with the `` acceleration '' search code has just started .
confirmation observations have been made on most of the better candidates , resulting in the detection of about 400 previously unknown pulsars , most of which are relatively distant , the median distance being @xmath10 kpc .
more than 150 previously known pulsars have also been detected .
the current discovery rate is an unprecedented one pulsar per hour of survey observing time . this rate will decline as regions further from the galactic plane are surveyed , but we estimate the survey will detect at least 600 previously unknown pulsars , nearly doubling the number known before the survey commenced , and providing a significant database for many different follow - up studies .
after confirmation , each pulsar is subjected to a series of timing observations at either the parkes 64-m telescope or the lovell 76-m telescope at jodrell bank .
almost all of the detected pulsars north of declination @xmath11 are being timed at jodrell bank . at parkes ,
data are recorded using just the central beam of the multibeam system .
the jodrell bank observations are made in a 96-mhz band centred on 1376mhz .
dual - channel cryogenic systems receiving orthogonal circular polarisations are used , each channel having a system noise temperature of about 30k at high galactic latitude . each of the two polarisation signals are down - converted and fed into a multi - channel filterbank consisting of 32 3-mhz filters , before digitisation .
data from both parkes and jodrell bank are de - dispersed and synchronously folded at the topocentric pulsar period for @xmath12min to form sub - integrations ; the total integration time per observation ranges between 2 and 30min , dependent upon the pulsar flux density .
each pulse profile obtained by summing over an observation is convolved with a high signal - to - noise ratio `` standard profile '' for the corresponding pulsar , producing a topocentric time - of - arrival ( toa ) .
these are then processed using the tempo program ( see http://pulsar.princeton.edu/tempo ) which converts them to barycentric toas at infinite frequency and performs a multi - parameter fit for the pulsar parameters .
barycentric corrections are obtained using the jet propulsion laboratory de200 solar - system ephemeris @xcite . except for especially interesting cases ,
we make timing observations of each pulsar over about 12 months , resulting in an accurate position , period , period derivative and dispersion measure ( dm ) .
this provides the basic parameters necessary for follow - up studies such as investigations of the galactic distribution of pulsars , studies of the interstellar medium and high - energy observations
. these observations also reveal pulsars which are members of binary systems . after detection , such pulsars
are observed more intensively to determine their binary parameters . of the new pulsars detected so far ,
seven have already been identified as members of binary systems and one of these is described in the next section .
( 12,10 ) ( -2,11 )
psr j1811@xmath01736 was initially observed on 1997 august 3 and revealed as a candidate with a 104-millisecond period and dm of 477@xmath13pc . it was confirmed as a pulsar on 1997 december 19 .
this observation gave a period which was substantially different from the initial discovery period , indicating a possible binary nature .
this was confirmed by a series of observations at jodrell bank during the beginning of 1998 which demonstrated that the pulsar was a member of a binary system in a highly eccentric orbit .
[ fig : perphase ] shows the measured period variation of psr j1811@xmath01736 through the orbital period of 18.8 days .
subsequently , we performed a phase - coherent analysis of the arrival times over a 16-month period .
the results are summarised in table 1 .
errors given in parentheses refer to the last quoted digit and are twice the formal standard error . using the taylor & cordes ( 1993 ) model for the electron density distribution in the galaxy and the measured dm , we obtain an estimated distance for psr j1811@xmath01736 of @xmath14kpc . with the measured flux density at 1400mhz of @xmath15mjy ( table 1 ) , the luminosity of this pulsar at 1400mhz is @xmath16mjykpc@xmath17 .
the pulsar has not yet been detected at other frequencies , mainly because of the large amount of scattering at lower frequencies .
however , we can get a rough estimate of its luminosity at @xmath18mhz by assuming a typical pulsar spectral index of @xmath19 .
we obtain @xmath20mjykpc@xmath17 , indicating that this is a relatively luminous pulsar , significantly above any low - luminosity cut - off in the luminosity distribution of galactic disk pulsars at 400mhz of 1mjykpc@xmath17 or lower ( lorimer et al .
1993 ; lyne et al .
1998 ) .
ll + + right ascension ( j2000 ) & @xmath21 + declination ( j2000 ) & @xmath22 + dispersion measure ( @xmath13 pc ) & 477(10 ) + period ( s ) & 0.104181954734(3 ) + epoch of period ( mjd ) & 51050.0 + period derivative & @xmath23 + orbital period ( days ) & 18.779168(4 ) + projected semi - major axis ( lt - s ) & 34.7830(8 ) + eccentricity & 0.82802(2 ) + epoch of periastron ( mjd ) & 51044.03702(3 ) + angle of periastron ( degrees ) & 127.661(2 ) + rate of advance of periastron ( @xmath24yr@xmath25 ) & 0.009(2 ) + rate of change of projected semi - major axis & @xmath26 + r.m.s . timing residual ( ms ) & 1.0 + flux density at 1400 mhz ( mjy ) & 0.7(2 ) + interstellar scattering at 1376mhz ( ms ) & 21(3 ) + + galactic longitude ( degrees ) & 13.4 + galactic latitude ( degrees ) & 0.7 + distance ( kpc ) & @xmath27 + characteristic age ( yr ) & @xmath28 + surface magnetic field ( g ) & @xmath29 + total system mass ( m@xmath2 ) & 2.6(9 ) + mass function ( m@xmath2 ) & 0.128 + ( 12,10 ) ( -3,11 ) the average pulse profile of psr j1811@xmath01736 is shown in fig .
[ fig : profile ] .
it is very asymmetric , with a rapidly rising leading edge with an approximately exponential tail following the peak .
the tail is detectable throughout the period and is characteristic of the effects of multi - path propagation in the interstellar medium , more usually observed at lower frequencies . in such scattering ,
the timescale of the exponential , @xmath30 , varies with frequency , @xmath31 , approximately as @xmath32 . over the 288-mhz band of the parkes receiver , @xmath30
is found to vary by about a factor of two , consistent with such a variation .
the amount of scattering is unusually large for a pulsar with this dm observed at around 1400mhz : the taylor & cordes ( 1993 ) model predicts @xmath33ms , while the observed @xmath30 is about 20ms .
this accounts for the rather large rms timing residual reported in table 1 .
as listed in table 1 , the spin parameters for psr j1811@xmath01736 imply a large `` characteristic age '' of @xmath34yr , and a relatively low implied surface dipole magnetic field strength of @xmath35 g .
such spin parameters are typical of the known double - neutron star systems ( taylor et al . 1995 ; nice , sayer , & taylor 1996 ) .
this pulsar is located below the `` spin - up line '' on a period magnetic field diagram ( see , e.g. lyne & graham - smith , 1998 , p115 or phinney and kulkarni 1994 , fig .
1 ) , suggesting that it was spun - up at an earlier stage in its history .
( 12,10 ) ( 0.5,0 ) in addition to the standard astrometric , spin , and keplerian orbital parameters , a satisfactory fit to the timing data requires inclusion of the rate of orbital precession of the pulsar in the fitted parameters , giving @xmath36yr@xmath25 . assuming that the companion star is a compact object , that tidal effects are negligible and that the measured rate of advance of periastron is due to general relativistic effects alone , we have ( e.g. will 1993 ) : @xmath37 where @xmath38s , @xmath39 is the binary period , @xmath40 the binary eccentricity , and @xmath41 and @xmath42 are the pulsar and companion masses , in solar masses
. the measurement of @xmath43 yields a total system mass of @xmath44 m@xmath3 , or an average component mass of @xmath45m@xmath3 .
this compares with the average mass of neutron stars measured in other binary pulsar systems of @xmath46m@xmath3 @xcite .
the constraints on the masses of the two stars can be most readily seen in fig .
[ fig : m1m2 ] .
the region under the convex curve is excluded because of the requirement that @xmath47 in the `` mass function '' , @xmath48 where @xmath49 is the projected semi - major axis of the pulsar orbit . from fig .
[ fig : m1m2 ] we see that @xmath50m@xmath51 and that @xmath52m@xmath51 .
these constraints on the individual and average masses of the system , the lack of eclipses , the low magnetic field of the pulsar , and the high binary eccentricity , all suggest that the pulsar companion is another neutron star .
if so , this is only the fourth or fifth double - neutron star system discovered in the galactic disk ( e.g. nice , sayer & taylor 1996 ) .
if the companion were a main sequence or giant star , in addition to relativistic effects , classical effects could also contribute to the measured value of @xmath53 ( e.g. lai , bildsten & kaspi 1995 ; kaspi et al .
1996 ; wex 1998 ) . for a pulsar of mass @xmath54m@xmath51 , the relativistic contribution always dominates .
however , for a small range of companion masses around 1m@xmath51 , some remaining @xmath53 could in principle be attributed to a classical quadrupole moment @xmath55 in the companion star .
the neutron star should raise a tide in a companion having radius @xmath56 such that @xmath57 where @xmath58 is the apsidal constant and @xmath59 is the instantaneous centre - of - mass separation @xcite .
however , given the orbital parameters and estimates for @xmath58 from stellar modeling @xcite , the values of @xmath55 permitted by the observations are much smaller than would be expected .
thus , a main - sequence or giant companion should result in a much larger @xmath53 than is observed .
this conclusion is not very sensitive to the choice of @xmath41 .
we note that a rapidly rotating main sequence star companion , with spin axis sufficiently misaligned with the orbital angular momentum , could result in a spin - induced quadrupole that precesses the orbit in the retrograde direction , cancelling the tidal effect .
however , this is most unlikely as it requires fine - tuning the spin and tidal effects to be surprisingly close in magnitude .
recently , the companion in another double neutron - star candidate system ( psr b2303 + 46 ) has been shown to be a white dwarf @xcite .
it is therefore important to assess what other constraints exist on the nature of the companion .
searches of a number of optical telescope archives , revealed that the only available images were from the digital sky survey v band , obtained at the uk schmidt telescope .
no sources are visible in the vicinity of the pulsar down to the plate limit of m@xmath60 . unfortunately
, this can only conclusively rule out a red giant companion and a deeper image would be required to test if the companion is a main sequence star ( m@xmath61 ) or a white dwarf ( m@xmath62 ) .
we note that there is no evidence of occultation of the pulsar at any phase of the orbit ( fig .
the closest approach of the two stars depends upon the unknown inclination angle @xmath63 , but is several solar radii , making the occultation of the pulsar by a solar - mass main - sequence companion star possible , but rather unlikely .
these observations do not therefore distinguish between a main - sequence and a compact companion .
however , the whole orbit is smaller than a giant , ruling out the possibility of such a companion .
the absence of significant tidal or quadrupole effects on @xmath64 suggests that the companion star is compact and therefore either another neutron star or a white dwarf .
such binary systems are generally thought to evolve from two initially massive stars .
the formation of a neutron star in the supernova collapse of the primary is expected in due course to be followed by a common - envelope and spiral - in stage . during this time , it is detectable as an accreting high - mass x - ray binary system ( e.g. verbunt 1993 ) , prior to the formation of a second neutron star or a white dwarf . if the pulsar is the second - formed compact object , it is not clear how the magnetic field became so small , an attribute which is generally thought to arise during mass - transfer .
if the pulsar is the original neutron star , the ensuing mass - transfer would have reduced the magnetic field . however , this would also have circularised the orbit which would have remained that way if the companion evolved to a white dwarf .
since the orbit is eccentric , we must conclude that it was probably the formation of a second neutron star which gave rise to the eccentricity . in the future ,
this binary system is not likely to evolve significantly . because of its relatively large orbital size
, it does not emit significant amounts of gravitational radiation : using the formulae of peters ( 1964 ) , we calculate that the coalescence time of the psr j1811@xmath01736 system is about @xmath65yr .
therefore , systems such as this make a negligible contribution to the neutron - star merger - rate calculations which are important for gravity - wave detectors such as ligo .
we gratefully acknowledge the technical assistance provided by george loone , tim ikin , mark leach and all the staff at the parkes observatory in developing the pulsar multibeam system .
the parkes radio telescope is part of the australia telescope which is funded by the commonwealth of australia for operation as a national facility managed by csiro .
f. camilo gratefully acknowledges support from the european commission through a marie curie fellowship under contract no .
erb fmbi ct961700 .
v. m. k. is an alfred p. sloan research fellow . | we are undertaking a high - frequency survey of the galactic plane for radio pulsars , using the 13-element multibeam receiver on the 64-m parkes radio telescope .
we describe briefly the survey system and some of the initial results .
psr j1811@xmath01736 , one of the first pulsars discovered with this system , has a rotation period of 104ms .
subsequent timing observations using the 76-m radio telescope at jodrell bank show that it is in an 18.8-day , highly - eccentric binary orbit .
we have measured the rate of advance of periastron which indicates a total system mass of @xmath1m@xmath2 , and the minimum companion mass is about 0.7m@xmath3 .
this , the high orbital eccentricity and the recycled nature of the pulsar suggests that this system is composed of two neutron stars , only the fourth or fifth such system known in the disk of the galaxy .
0= = 0 @=@ 0= = 0 ! = ! = methods : observational pulsars : general pulsars : individual ( j1811@xmath01736 ) pulsars : searches pulsars : timing |
the existence of a heavy neutral gauge boson @xmath1 is theoretically motivated by most extended electroweak scenarios attempting to overcome the conceptual difficulties of the standard model @xcite .
the knowledge of the @xmath1 parameters , such as the mass and couplings to ordinary fermions , would be essential in order to test these extended theories .
numerous strategies to evidence manifestations of the @xmath1 in experiments at high energy @xmath4 and hadronic colliders have been developed . due to the present ` direct ' lower mass limit from the tevatron , @xmath5 gev @xcite ,
only ` indirect ' manifestations of the @xmath1 can be searched for at the energy of lep2 .
the preferred channel in this case should be the electron - positron annihilation into fermion pairs , where the @xmath1 contributes already at the born level @xcite . in this regard
, particularly convenient should be the annihilation into lepton pairs @xmath6 where , assuming lepton universality , the number of @xmath1 parameters is significantly reduced , leading to characteristic , model - independent potential signals .
these effects manifest themselves as deviations of observables from the standard model ( sm ) predictions , to be compared to the experimental data , not only as regards the size of the signals , but also their characteristic signs and energy dependences . in the case of
no observed signal within the experimental accuracy , the result of the analysis can be expressed as bounds on the @xmath1 parameters to a conventionally specified confidence level .
similar effects in process ( [ proc ] ) can arise from the independent physical mechanism represented by anomalous gauge boson couplings , that can contribute even at the tree level @xcite .
clearly , in the case of no observed effect , upper limits on the anomalous gauge couplings can be obtained from the data . on the other hand , it would be desirable to have some criterion to distinguish the two sources of such nonstandard effects , in the case these were observed .
the aim of this paper is to present a discussion of the general properties of the deviations from the sm predictions due to the @xmath1 and to the anomalous gauge couplings , at the energies of lep2 .
in particular , some criterion to distinguish the two kinds of signal is worked out . moreover ,
the constraints on the relevant parameters are derived either from the present lep2 data and in the case of the foreseeable ones .
we start from the general neutral - current interaction including the @xmath1 : @xmath7 where @xmath8 ; @xmath9 ( with @xmath10 ) and @xmath11 are the @xmath12 and @xmath1 gauge couplings , respectively .
the currents can be written as @xmath13 here , @xmath14 and @xmath15 are projectors onto the left- and right - handed electron helicity states , so that @xmath16 and @xmath17 . the standard model ( sm ) couplings are @xmath18 in general
, the unpolarized differential cross section for electron - positron annihilation into a fermion - antifermion pair , @xmath19 , can be written in born approximation including @xmath20 , @xmath12 and @xmath1 exchanges , as : @xmath21 , \label{cross}\ ] ] where @xmath22 is the angle between the initial electron and the outgoing fermion . for our purposes , it is convenient to split @xmath23 and @xmath24 as : @xmath25 in eq .
( [ f ] ) : @xmath26 and the deviation due to the @xmath1 is @xmath27 here , the gauge boson propagators are @xmath28 , @xmath29 , @xmath30 , and the fermionic coupling constants are normalized as : @xmath31 the total width @xmath32 can be assessed from a sum over the partial widths : @xmath33.\label{width}\ ] ] the leptonic channel ( [ proc ] ) has the nice phenomenological feature that , assuming leptonic @xmath34 universality , only three independent nonstandard parameters are needed : the couplings @xmath35 , @xmath36 and the mass @xmath37 .
more parameters would be needed in the description of @xmath38 .
concentrating on process ( [ proc ] ) at lep2 energies with @xmath39 , the two conventionally used observables are the total cross section @xmath40 with @xmath41 , and the forward - backward asymmetry @xmath42 where the forward and backward cross sections are defined by @xmath43 and @xmath44 , respectively . as it will appear in the sequel , in order to identify @xmath1 effects in process ( [ proc ] ) it may be convenient to consider , alternatively to @xmath45 , just the difference @xmath46 between the forward and backward cross sections . to discuss the distinctive features of nonstandard physical effects induced either by the exchange of a @xmath30 or by anomalous gauge couplings , we introduce also the new observables @xmath2 and @xmath3 defined as the differences of cross sections integrated in suitable ranges of @xmath47 : @xmath48 where @xmath49 is determined from the condition that the kinematical coefficients multiplying @xmath23 and @xmath24 in eq .
( [ cross ] ) be the same after integration over @xmath47 . the condition is @xmath50 and gives the equation for @xmath51 @xmath52 whose solution is @xmath53 , corresponding to @xmath54 . in the case of a reduced kinematical range of the scattering angle such as , e.g. , @xmath55 , one derives @xmath56 .
one can write @xmath2 and @xmath3 in terms of @xmath23 and @xmath24 as @xmath57 where , to a very good precision : @xmath58 the new , independent , observables @xmath59 are simple combinations of the conventional ones @xmath60 and @xmath45 . indeed , from eqs .
( [ crosstot ] ) , ( [ afb ] ) , ( [ sigmapm ] ) and ( [ pt * ] ) : @xmath61 so that @xmath2 and @xmath3 can be measured either directly according to eqs . ( [ sigma+ ] ) and ( [ sigma- ] ) or indirectly by means of @xmath60 and @xmath45 .
the previous formulae for the differential cross section ( [ cross ] ) , as well as those for all of the other observables , are still valid to a very good ( improved born ) approximation which includes one - loop electro - weak radiative corrections , with the following replacements @xcite , @xcite : @xmath62 where only the main contribution to @xmath63 , coming from the _ top _ mass , has been taken into account .
this parameterization uses the best known sm parameters @xmath64 , @xmath65 , and @xmath66 .
moreover , the energy dependence of the @xmath12-width should be accounted for by the replacement of the @xmath12 propagator : @xmath67 in our analysis , the improved born approximation with @xmath68 gev and @xmath69 gev is used . while the numerical results are obtained from the exact formulae , for the sake of a simplified presentation of the analysis it is convenient to make a few , harmless , assumptions : _ i
_ ) since the typical lower bound for the @xmath1 boson mass , is much larger than the lep2 energy , it suffices to take into account @xmath1 interference effects only , the pure @xmath1 exchange contributions being negligible ; _ ii _ ) since in the sm @xmath70 , in the expressions of the deviations from the sm due to the @xmath1 we neglect @xmath71 with respect to @xmath72 . in addition , one can neglect the imaginary part of the @xmath1 boson propagator . in the approximations _
i _ ) and _ ii _ ) , the deviation of the cross section from the sm prediction , @xmath73 , reads @xmath74 where the two terms represent the @xmath75 and @xmath76 interference , respectively , and @xmath77 , @xmath78 are the leptonic vector and axial - vector couplings to the @xmath1 defined in eq .
( [ coupli ] ) . in eq .
( [ delsigma ] ) , only squares of leptonic couplings appear , so that one has the general property that the sign of the deviations of the cross section from the sm prediction due to the @xmath1 is definitely negative at the lep2 energy , because both terms within the round brackets are positive . from eq .
( [ df ] ) , it is clear that this is independent of the simplifying assumption _ ii _ ) above .
also , such sign definiteness holds only for process ( [ proc ] ) , assuming universality between initial and final leptons @xmath79 or @xmath80 , and not for the the process @xmath81 , where additional assumptions on the fermion coupling constants would be needed .
the deviation of @xmath46 in ( [ afb ] ) from the sm prediction , @xmath82 , has a similar expression : @xmath83 and at lep2 it generally has the same negative sign property for arbitrary leptonic @xmath1 couplings as the cross section @xmath60 .
such sign property of @xmath1 effects for these observables can be used to attempt a model - independent distinction of the @xmath1 from analogous anomalous gauge coupling effects , if some deviations were observed . concerning the forward - backward asymmetry ( [ afb ] ) , the deviation from the sm prediction is expressed as @xmath84\chi_{z'}\ ] ] and , in contrast to the case of @xmath60 and @xmath46 , at lep2 energy it can be either positive or negative , depending on the actual values of the @xmath1 couplings .
the deviation of @xmath2 from the sm prediction has the expression @xmath85 eq . ( [ deltasig+ ] ) shows that the dependence of @xmath86 on the @xmath1 parameters is characterized by the expression @xmath87 , which is a negative quantity at @xmath88 for arbitrary @xmath1 couplings .
in addition to the definite negative sign , @xmath86 has the property of depending on the total center of mass energy @xmath89 _ via _ the factor @xmath90 , which is completely determined by well - known sm parameters .
1a shows the energy dependence of the relative deviation @xmath91 . in this numerical example , we have considered a ` sequential ' @xmath1 @xcite with couplings @xmath92 , @xmath93 and with masses @xmath94 gev and @xmath95 gev . in this figure , one should notice the characteristic sign correlations of @xmath96 at the different energies expected from a @xmath1 ( @xmath97 is sequentially negative , positive and then negative again , with increasing @xmath89 ) .
as regards @xmath98 , it has in common with @xmath2 the property that the deviation from the sm prediction @xmath99 has an energy dependence completely determined in terms of sm parameters .
however , in contrast to @xmath96 , the sign and magnitude of @xmath100 is determined by @xmath101 , for which both the positive and the negative signs are generally possible .
1b shows the energy dependence of the relative deviation @xmath102 , which , similar to @xmath86 , has the characteristic sign correlations at the different energies . in the case of fig . 1b ,
we take a @xmath1 with same mass values as in fig . 1a , but with couplings such that @xmath103 ( to give an example ) .
.model - independent properties of deviations induced by a @xmath1 at lep2 energy [ cols="^,^,^",options="header " , ] in addition , to assess future possibilities of improving the sensitivity , we work out the same example by using the expected final luminosity at lep2 , assuming for the systematic uncertainty , in this case , @xmath104 @xcite ) . the numerical analysis has been performed by means of the program zefit , which has to be used along with zfitter @xcite . in fig .
2 we show the results from @xmath105 , and compare the bounds on the couplings ( [ coupl ] ) obtainable from current and future experiments at lep2 , with inputs from table 2 .
the contours are derived from the combination of both observables , @xmath106 and @xmath3 , and correspond to two standard deviations ( @xmath107 ) . from the numerical point of view , a comparison of allowed bounds on ( @xmath108 ) depicted in fig .
2 with those obtained from a similar analysis of @xmath60 and @xmath45 indicates that the new observables @xmath2 and @xmath3 have a slightly better sensitivity .
this reflects the fact that , according to ( [ deltasig+ ] ) , ( [ dsigm ] ) and ( [ coupl ] ) , the deviations @xmath96 and @xmath109 are proportional to combinations of @xmath1 coupling constants ( @xmath110 , respectively ) rather than the individual ones , so that a one - parameter fit applies in this case , rather than a two - parameter one .
moreover , we may remark that the above analysis is a model - independent one , as it starts from rather general considerations .
therefore , it can usefully complement the conventional analysis of @xmath1 couplings based on the observables @xmath60 and @xmath45 .
in this regard , we now discuss a specific case of models with anomalous gauge couplings .
this case was considered in ref .
@xcite , specifically concentrating on the @xmath111 invariant lagrangian with only dimension six operators . of these operators , four ones ( denoted as @xmath112 , @xmath113 , @xmath114 , and @xmath115 ) affect the neutral current and charged current amplitudes at tree level , and five more induce effects at the one - loop level . of these five operators , three ones
lead to anomalous trilinear vector boson couplings . as it was shown in ref .
@xcite , the quadratically and logarithmically divergent contributions of the latter five operators to neutral and charged current amplitudes are equivalent to a renormalization of the four operators @xmath112 , @xmath113 , @xmath114 , and @xmath115 which contribute at tree level . although the number of parameters of this model is apparently rather large , it was shown in ref .
@xcite that in the @xmath12-peak subtracted representation only two parameters remain in the process @xmath116 , since @xmath117 and @xmath118 are fully reabsorbed in the input parameters @xmath119 and @xmath120 . in the @xmath12-peak subtracted representation ,
the observables @xmath60 and @xmath46 are given by @xcite : @xmath121 , \label{sigma_gamma}\\ \sigma_{\mu \mu}^{(z ) } ( s)&=&\tilde\sigma_{pt } \hskip 2pt \left |
\chi_z \right |^2 a^4\hskip 2pt \left [ 1 -2 r(s ) - 16 s_w c_w \frac{\tilde{v}_l}{1+\tilde{v}^2_l } v(s ) \right ] , \label{sigma_z}\\ \sigma_{\mu \mu}^{(\gamma z ) } ( s)&=&\tilde\sigma_{pt } \hskip 2pt 2 { \rm re } \chi_z \hskip 2pt a^2\frac{\tilde{v}^2_l}{1+\tilde{v}^2_l } \hskip 3pt \times \nonumber \\ & \times & \hskip 3pt \left [ 1 + \tilde{\delta } \alpha ( s ) - r(s ) - 8 \hskip 2pt \frac{s_w c_w}{\tilde{v_l } } v(s ) \right ] , \label{sigma_gammaz}\end{aligned}\ ] ] and @xmath122 , \label{fb_z}\\ \sigma_{\rm fb}^{(\gamma z)}(s ) & = & \frac{3}{4}\tilde\sigma_{pt } \hskip 2pt 2{\rm re } \chi_z \hskip 2pt a^2\frac{1}{1+\tilde{v}^2_l } \hskip 2pt \left [ 1 + \tilde{\delta } \alpha ( s ) - r(s ) \right ] . \label{fb_gammaz}\end{aligned}\ ] ] here : @xmath123 , @xmath124 , @xmath125 , @xmath126 as in ( [ impborn ] ) , and @xmath127 .
contributions of anomalous gauge couplings ( agc ) to the above equations can be written as : @xmath128 , \label{alpha } \\ & & r^{(agc ) } ( s ) = \frac{8\pi \alpha } { s^2_w c^2_w } \hskip 2pt \frac{(s - m^2_z)}{\lambda^2 } \left [ c^4_w f^r_{dw } + s^4_w f^r_{db } \right ] , \label{r}\\ & & v^{(agc ) } ( s ) = \frac{8\pi \alpha } { s_w c_w } \hskip 2pt \frac{(s - m^2_z)}{\lambda^2 } \left [ c^2_w f^r_{dw } - s^2_w f^r_{db } \right ] .
\label{v}\end{aligned}\ ] ] in these equations , @xmath129 is the mass scale of the new interaction , and @xmath130 and @xmath131 are the renormalized couplings that are associated to the six - dimensional effective operators @xmath132\left [ d^\mu , \hat{w}^{\nu \rho } \right ] \right ) , \\ & & { \cal o}_{db } = -\frac{g'^2}{2 } \left ( \partial_\mu b_{\nu \rho } \right ) \left(\partial^\nu b^{\nu \rho}\right),\end{aligned}\ ] ] that are involved in the effective lagrangian @xcite and survive in the @xmath12-peak subtracted representation .
it is convenient to express the deviations of the cross section and of the forward - backward cross section from the sm prediction , at energies away from the @xmath12 peak , in the approximation based on the simplified assumption _ ii _ ) , and neglecting numerically irrelevant small contributions .
the result is : @xmath133 \label{crossagc}\end{aligned}\ ] ] and @xmath134 .
\label{fbagc}\end{aligned}\ ] ] the corresponding expressions for the new observables @xmath2 and @xmath3 can be easily derived from the relation ( [ relation ] ) and the formulae for @xmath60 and @xmath46 .
the result is : @xmath135 and @xmath136 fig .
3a shows the relative deviation @xmath137 induced by the anomalous gauge couplings with ( @xmath138)= @xmath139 and @xmath140 tev , as numerical examples .
in contrast to the @xmath1 case ( fig .
1a ) , the sign of the deviation due to anomalous gauge couplings is not defined , i.e. , it may be either positive or negative at energies up to lep2 .
however , as eq .
( [ sigma+agc ] ) shows , characteristic energy correlations of the deviation occur also in this case .
indeed , for certain values of the parameters ( @xmath141 ) the energy behaviour of the deviation due to anomalous gauge couplings is the same as that in fig .
1a for the @xmath1 , while for other choices of the parameters ( @xmath142 ) the behaviour is opposite to the @xmath1 case .
3b shows the energy behaviour of @xmath143 for the same choice of the parameters .
such behaviour is qualitatively similar to the @xmath1 case , in particular regarding the energy correlations of the sign of the deviation .
moreover , from eq .
( [ sigma - agc ] ) , @xmath109 depends on the couplings @xmath131 only , and therefore just a one - parameter fit to current ( and expected ) experimental data is required for this parameter in our analysis . concerning the distinction of the two potential sources of nonstandard effects considered here , @xmath1 vs anomalous gauge couplings
, one can use the negative sign property at lep2 of @xmath96 as well as @xmath144 and @xmath145 , which holds for the former case ( table 1 ) but not for the latter one .
the hatched area in the ( @xmath130,@xmath131 ) plane of fig .
4 corresponds to the values of anomalous gauge couplings such that the deviations of eqs .
( [ crossagc])-([sigma+agc ] ) are negative at lep2 .
this is a ` confusion ' region , where both nonstandard mechanisms can induce the same deviations from the sm prediction , and therefore can not be distinguished through the property of negative sign deviation of the observables introduced here . on the contrary , the white area on the left of the hatched one
is determined by values of the anomalous couplings such that the corresponding ( positive ) deviations are not allowed to the @xmath1 and , in this sense , the source of the nonstandard effect can be distinguished .
another condition can be derived by considering that eqs .
( [ deltasig+ ] ) and ( [ dsigm ] ) imply , for the @xmath1 , that @xmath146 with , of course , @xmath147 . for , e.g. , @xmath148 gev , the above ratio
is limited by @xmath149 taking into account eqs .
( [ sigma+agc ] ) and ( [ sigma - agc ] ) for the case of the anomalous gauge couplings : @xmath150 and , comparing with the numerical bound ( [ ratio ] ) , one finds that the actual ` confusion ' region between the two kinds of mechanism is limited by the lines 4 and 5 of fig . 4 , and therefore is somewhat reduced . finally , the contours in fig .
5 represent the upper bounds on @xmath130 and @xmath131 allowed by the combination of @xmath2 and @xmath3 at the 2-@xmath151 level , in the case of no deviation observed from the sm , using the current lep2 data of table 2 and the expected experimental uncertainty with the future full luminosity .
numerical values of the anomalous gauge couplings included in the area between the two ellipses give rise to deviations from the sm that are , potentially , still in the reach of future measurements at lep2 .
the values of the anomalous couplings within the shaded area produce deviations which can also originate from the @xmath1 , and in this sense identify the so - called ` confusion ' area .
aab and aap gratefully acknowledge the support of the university of trieste .
99 for a review see , e.g. , l. hewett and t. g. rizzo , phys . rep .
* c183 * ( 1989 ) 193 .
valls , report fermilab - conf-97/135-e .
physics at lep2 , proceedings of the workshop , geneva , switzerland ( 1996 ) , cern 96 - 01 , g. altarelli , t. sjstrand and f. zwirner , eds .
see , e.g. , k. hagiwara , s. ishihara , r. szalapski and d. zeppenfeld , phys .
b283 * ( 1992 ) 353 ; phys .
d * 48 * ( 1993 ) 2182 .
m. consoli , w. hollik and f. jegerlehner , _ in _ z physics at lep1 , ed .
g. altarelli , r. kleiss and c. verzegnassi , vol.1 , p.7 , 1989 .
g. altarelli , r. casalbuoni , d. dominici , f. feruglio and r. gatto , nucl .
* b342 * ( 1990 ) 15 .
g. altarelli , b. mele and m. ruiz altaba , z. phys .
c * 45 * ( 1989 ) 109 ; ( e ) c * 47 * ( 1990 ) 676 .
a. leike , z. phys .
c * 62 * ( 1994 ) 265 .
p. osland and a.a .
pankov , phys . lett .
* b403 * ( 1997 ) 93 ; * b406 * ( 1997 ) 328 .
l3 collaboration , m. acciarri et al .
, preprint cern - ppe/97 - 52 , 1997 .
opal collaboration , k. ackerstaff et al .
, preprint cern - ppe/97 - 101 , 1997 .
aleph collaboration , contribution to the 1997 eps - hep conference , jerusalem , august 19 - 26,1997 , ref .
delphi collaboration , a. behrmann et al .
, preprint delphi 97 - 132 conf 110 , 1997 . s. riemann , fortran program zefit version 4.2 ; + d. bardin et al . , preprint cern - th .
6443/92 , cern , 1992 . f. m. renard and c. verzegnassi , phys .
d * 52 * ( 1995 ) 1369 ; d * 53 * ( 1996 ) 1290 .
* fig . 1a * relative deviation @xmath137 induced by a ` sequential ' @xmath1 with @xmath152 and @xmath153 .
@xmath154 ( line 1 ) and @xmath155 ( line 2 ) .
notice @xmath156 at @xmath157 and @xmath158 .
1b * relative deviation @xmath159 .
lines 1 and 2 correspond to @xmath160 , @xmath154 and @xmath155 , respectively .
lines 3 and 4 correspond to @xmath161 and @xmath154 , @xmath155 , respectively .
notice @xmath162 at @xmath158 and @xmath163 . *
fig . 2 * upper bounds on the @xmath1 couplings ( @xmath164 , @xmath165 ) determined by @xmath105 at the @xmath166 level .
the solid contour is obtained from @xmath0 with inputs summarized in table 2 .
the dashed line is obtained with @xmath167 and @xmath168 . *
4 * straight lines in the two - dimensional plane ( @xmath138 ) corresponding to @xmath174 ( line 1 ) , @xmath175 ( line 2 ) and @xmath176 ( line 3 ) .
the hatched area is the ` confusion ' area between @xmath1 and anomalous gauge couplings effects when the deviations @xmath144 , @xmath145 and @xmath96 are negative at lep2 .
the shaded area is the reduced ` confusion ' area using also the solutions of ( [ ratio ] ) and ( [ ragc ] ) ( lines 4 and 5 ) . * fig . 5 * upper bounds on the anomalous gauge couplings ( @xmath138 ) determined by @xmath105 at the @xmath166 level . input parameters for solid and dotted lines are as in fig . 2 .
the shaded area is the ` confusion ' one , where @xmath1 and anomalous gauge couplings can produce the same deviation . | in the case of the process @xmath0 , we study the sensitivity to either @xmath1 or anomalous gauge coupling effects of two new observables , @xmath2 and @xmath3 , conveniently defined for such kind of analysis .
we discuss general properties of the deviations of these observables from the standard model predictions , and derive the model - independent bounds on the relevant parameters allowed by present and foreseeable lep2 data .
we also discuss the possibility of separating the two kinds of effect .
* @xmath1 and anomalous gauge coupling effects at lep2 + and their separation * + international centre for theoretical physics , trieste , italy a. a. pankov + international centre for theoretical physics , trieste , italy + istituto nazionale di fisica nucleare , sezione di trieste , trieste , italy n. paver + dipartimento di fisica teorica , universit di trieste , trieste , italy + istituto nazionale di fisica nucleare , sezione di trieste , trieste , italy |
the magnetars ( comprising the anomalous x - ray pulsars and soft gamma repeaters ; axps and sgrs ) are a small group of x - ray pulsars ( about twenty objects ) , with spin periods between 0.3 and 12 s. their bright x - ray emission ( @xmath1 ) is marginally explained by the commonly accepted emission models for isolated pulsars or by accretion from a companion star . their inferred magnetic fields assuming the spin - down dipolar loss formula ( b = 3.2 @xmath210@xmath3 g
, @xmath4 is the spin period and @xmath5 is the first derivative of spin period ) appear to be in general as high as @xmath6 g . due to these high magnetic fields , the emission of magnetars is thought to be powered by the decay and the instability of their strong fields ( duncan & thompson 1992 ; thompson & duncan 1993 , thompson , lyutikov , & kulkarni 2002 ) .
the powerful x - ray output is usually well modelled by thermal emission from the neutron star hot surface ( about 0.20.6kev ) , reprocessed in a twisted magnetosphere through resonant cyclotron scattering , a process favored only under these extreme magnetic conditions . on top of their persistent x - ray emission , magnetars emit very peculiar flares and outbursts on several timescales ( from fractions of seconds to years ) releasing a large amount of energy ( @xmath7 erg ) .
these flares are probably caused by large - scale rearrangements of the surface / magnetospheric field , either accompanied or triggered by displacements of the neutron - star crust ( somehow analogous to quakes on the stellar surface ) .
see mereghetti ( 2008 ) , rea & esposito ( 2011 ) , and olausen & kaspi ( 2014 ) for recent reviews .
thanks to _ integral _ , _ suzaku _ and _ nustar _ we now have good spectra for magnetars in the hard x - ray energy range ( kuiper et al . 2004 ; enoto et al .
2010 ; an et al .
2015 ) .
the gamma - ray emission ( 0.1300gev band ) from 13 magnetars has been searched using the first 17 months of _ fermi_-lat data ( abdo et al .
this search resulted in upper limits and no pulsations were found either .
taking advantage of about 5 more years of data , in this paper we re - analyze the _
fermi_lat observations for the previous objects , as well as for the 7 magnetars that were discovered in the meantime .
furthermore , the better response of the instrument at lower energies brought about by pass 8 ( atwood et al .
2013 ) allows us to produce a more stringent search . in sec.[obs
] we report on the _ fermi_-lat analysis procedure , we summarize our results in sec.[results ] and [ timing ] , and provide a discussion of our findings in sec.[discussion ] .
[ table1 ] lcccccc + + source & ts&ts & 0.1@xmath81 gev & 1@xmath810 gev & 0.1@xmath810 gev & 6-years internal list + & & ( abdo et al .
2010a ) & @xmath9 ( 1.5 ) & @xmath9 ( 3.5 ) & @xmath9 ( 2.5 ) & srcs within 3@xmath10 + + + + & 0.0 & 2.3 & @xmath110.18 & @xmath110.05 & @xmath110.15 & 3 + & 0.0 & 3.6 & @xmath110.43 & @xmath110.07 & @xmath110.29 & 3 + swiftj1822.3@xmath81606 & 0.0 & & @xmath110.88 & @xmath110.13 & @xmath110.43 & 19 + swiftj1834.9@xmath80846@xmath12@xmath13 & 0.0 & & @xmath111.06 & @xmath11 0.99 & @xmath11 0.45 & 19 + & 0.0 & 0.0 & @xmath110.51 & @xmath110.53 & @xmath111.10 & 35 + xtej1810@xmath8197 & 0.0 & 13.1 & @xmath111.00 & @xmath110.89 & @xmath113.65 & 17 + & 0.8 & & @xmath111.98 & @xmath110.63 & @xmath111.00 & 26 + 1e1841@xmath8045 @xmath13 & 1.0 & 7.5 & @xmath111.05 & @xmath110.78 & @xmath11 2.02 & 23 + 3xmmj185246.6@xmath14003317@xmath13 & 2.5 & & @xmath111.32 & @xmath110.49 & @xmath111.88 & 23 + @xmath13 & 2.8 & 15.6 & @xmath110.21 & @xmath110.12 & @xmath110.13 & 13 + sgr1806@xmath820 & 2.9 & 2.8 & @xmath112.38 & @xmath110.50 & @xmath110.67 & 15 + sgr1935@xmath142154 & 3.5 & & @xmath110.11 & @xmath110.22 & @xmath110.17 & 6 + @xmath13 & 3.8 & 16.3 & @xmath110.51 & @xmath110.08 & @xmath110.40 & 4 + & 4.2 & 36.2 & @xmath111.40 & @xmath110.90 & @xmath112.84 & 14 + sgr1900@xmath1414@xmath13 & 4.5 & 0.0 & @xmath11 0.63 & @xmath110.49 & @xmath11 2.07 & 13 + & 6.0 & & @xmath110.96 & @xmath110.45 & @xmath110.89 & 27 + & 6.6 & 32.1 & @xmath112.14 & @xmath110.96 & @xmath112.51 & 15 + sgr1833@xmath80832 @xmath12@xmath13 & 18.9 & & @xmath112.75 & @xmath111.29 & @xmath110.87 & 19 + @xmath13 & 23.8 & 36.0 & @xmath112.87 & @xmath111.75 & @xmath113.34 & 29 + cxouj171405.7@xmath8381031@xmath15 & 51.0@xmath16 & & @xmath110.27 & @xmath110.53 & @xmath110.80 & 11 +
the _ fermi_-lat data included in this paper covered 74 months from 2008 august 4 to 2014 october 1 , which greatly extended the 17 months data coverage of abdo et al .
( 2010a ) .
the analysis of _
fermi_-lat data was performed using the _ fermi _ science tools , 09 - 34 - 01 release .
events from the pass 8 source class were selected .
the p8 source v5 instruments response functions ( irfs ) were used for the analysis .
all gamma - ray photons within an energy range of 0.1300 gev and within a circular region of interest ( roi ) of 10@xmath17 radius centered on each source analyzed were used for analysis . to reject contaminating gamma rays from the earth s limb , we selected events with zenith angle @xmath11 100@xmath17 , or @xmath11 90@xmath17 for the sources with high possibility of contamination ( high - latitude sources with delc.@xmath18 45@xmath17 , underlined in table 1 ) .
the gamma - ray flux and spectral results presented in this work were calculated by performing a binned maximum likelihood fit using the science tool _ gtlike_. the binned maximum likelihood analysis was carried out on a 14@xmath191@xmath214@xmath191 square region with a pixel size of 0@xmath191 centered on each magnetar .
30 bins in 100 mev300 gev were used in the analysis . the spectral - spatial model constructed to perform the likelihood analysis includes galactic and isotropic diffuse emission components ( gll_iem_v06.fits " , acero et al .
2016b , and iso_p8r2_source_v6_v06.txt " , respectively ) as well as known gamma - ray sources within 15@xmath17 from the magnetars , following the internal 6-years source list which was based on `` pass 8 '' _
fermi_-lat data covering from 2008 august 4 to 2014 august 4 .
the spectral parameters were fixed to the catalog values , except for the sources within 3@xmath17 from the magnetars . for these latter sources ,
all the spectral parameters were left free .
magnetars in table 1 were modelled with a simple power law with all spectral parameters allowed to vary . the maximum likelihood test statistic ( ts )
was employed to evaluate the significance of the gamma - ray fluxes coming from the magnetars .
the test statistic is defined as ts=@xmath20 , where @xmath21 is the maximum likelihood value for a model without an additional source ( the null hypothesis " ) and @xmath22 is the maximum likelihood value for a model with the additional source at a specified location . a larger ts indicates that the null hypothesis is incorrect ( i.e. , a source is detected ) .
the square root of the ts is approximately equal to the detection significance of a given source .
we set the detection threshold to be ts@xmath1825 .
the 95@xmath23 flux upper limits for magnetars are calculated following helene s method ( helene 1983 ) assuming a power - law spectral shape . in the 0.110 gev energy range
, we fixed the photon index at 2.5 , as in abdo et al .
( 2010a ) .
in 0.11 gev and 110 gev ranges , the upper limits are evaluated using a photon index fixed at 1.5 and 3.5 respectively , mimicking a spectral cutoff at @xmath24 1gev , as is common in pulsar spectra .
apart from the photon index , other spectral parameters in the upper limits calculation are the same as in the _ gtlike _ analysis . for extended source analysis
, we followed the method as described in lande et al .
we assumed the source to be spatially extended with a symmetric disk model and fitted its position and extension with the _ pointlike _ analysis package ( kerr , 2011 ) between 0.1 and 300 gev . following the method in lande et al .
( 2012 ) , the ts value of the putative extension is defined as ts@xmath25=2(@xmath26-@xmath27 ) , in which @xmath28 and @xmath29 were the _ gtlike _ global log likelihood of the point source and extended source hypothesis , respectively .
we set the threshold for claiming the source to be spatially extended as ts@xmath3016 corresponding to a significance of @xmath24 4 @xmath31 .
ts maps shown in this paper are produced with the pointlike analysis tool . in most cases ,
fermi_-lat data could be well modelled by the sources from the internal 6-years source list . in case of excesses in the residual ts map
, additional point sources were added following a similar method to that described in caliandro et al .
( 2013 ) , section 3.3.3 .
[ 1e2259 ] [ swift1834 ]
we studied the _ fermi_lat observations of all confirmed magnetars from the mcgill online magnetar catalog ( olausen & kaspi , 2014 ) , except for the extra - galactic ones and galactic center magnetars due to the difficulty of resolving them from their respective environments ( see abdo et al .
2010b , 2010c ) .
those are : sgr 0526@xmath866 in the large magellanic cloud , cxou j010043.1@xmath8721134 in the small magellanic cloud and sgr 1745@xmath82900 at the galactic center .
we list in table 1 the results for the 20 selected magnetars .
most of the magnetars are clearly undetected , having ts values @xmath32 in the _ gtlike _ analysis .
this is the case of : sgr0418@xmath145729 , 4u0142@xmath14614 , swiftj1822.3@xmath81606 , psrj1622@xmath84950 , 1e1048.1@xmath85937 , xtej1810@xmath8197 , sgr 1806@xmath820 , sgr1935@xmath142154 , cxou j164710.2@xmath8455216 , and 1e1547.0@xmath85408 .
no gamma - ray source is associated with these 10 magnetars in the ts maps .
upper limits are evaluated for the 10 magnetars .
the uncertainties of the _ fermi_lat effective area and of the galactic diffuse emission are the two main sources of systematics that can affect the evaluation of the upper limits to a different extent on individual sources .
these systematic effects are estimated by repeating the upper limit analysis using modified irfs that bracket the effective areas and changing the normalization of the galactic diffuse model artificially by @xmath336% , similar to abdo et al .
( 2010a ) , for all magnetars considered in this paper .
energy dispersion is not considered in the data analysis , which may be important below 100 mev , but not expected to produce significant changes in the 100 mev300 gev energy range considered in this work .
the results of this analysis are reported in table 1 .
below we comment in a case by case basis on all magnetar locations that resulted in a high ts value since they are either associated with or adjacent to a _
fermi_-lat detected snr , or located in a crowded region .
[ cxouj1714 ] snr kes 73 is associated with 1e1841@xmath8045 , which was reported as an extended gamma - ray source with a disk radius of 0@xmath1970@xmath340@xmath1903 and ts@xmath3519 by acero et al .
( 2016a ) .
a gamma - ray point source from the internal 6-years source list is located only 0.16@xmath17 away from 1e1841@xmath8045 ( figure [ 1e1841 ] ) , which was excluded from the spatial model .
the _ pointlike _ ts map of 1e1841@xmath8045 showed an extended source peaked near the magnetar ( figure [ 1e1841 ] , top left panel ) .
the source is not present in the third _ fermi _ source catalog ( acero et al .
2015a , 3fgl hereafter ) but it is spatially coincident the source from the internal 6-years source list mentioned above .
we estimated the extension of the gamma - ray source .
an extended disk with power law spectral shape has been fitted to the data . a disk of radius 0@xmath1932@xmath330@xmath1903 located at r.a.@xmath36280@xmath1921@xmath330@xmath1903
, decl.@xmath374@xmath1989@xmath330@xmath1903 is favored over a point - like source with ts@xmath25=54.4 .
the gamma - ray source is significantly detected as an extended source with an overall ts value of 323.2 .
the spectral parameters are reported in table [ table2 ] .
thus , the extended gamma - ray source detected is most likely the gev counterpart of snr kes 73 .
the ts map as well as the radio map from the vla galactic plane survey ( vgps , stil et al .
2006 ) are shown in figure [ 1e1841 ] , left and middle panel of the corresponding row .
snr kes 73 is clearly seen as a shell - like source in the radio map around 1e1841@xmath8045 ( figure [ 1e1841 ] , middle panel ) .
the gamma - ray extension detected is larger than the radio size of snr kes 73 and there is a radio complex which plausibly also contributes to the extended emission , leading to a larger radius than the radio shell .
however , the extended gamma - ray emission peaks at snrs kes 73 and we propose that it is the main contributor .
in order to search for gamma - ray emission from 1e1841@xmath8045 , we need to exclude snr kes 73 .
we modelled snr kes 73 as an extended source with spatial parameters fixed at the above values and performed the likelihood analysis .
1e1841@xmath8045 is not detected and the ts map is shown in figure [ 1e1841 ] . by assuming a power law spectral model , we derived 95% flux upper limits in the 0.11 gev , 110 gev and 0.110 gev energy bands .
the results are reported in table 1 . in 3fgl ,
a bright gamma - ray source 3fgl j2301.2 + 5853 is located only 0@xmath1902 away from 1e2259@xmath14586 . in the internal 6-years source list , two bright gamma - ray point sources are located near 1e2259@xmath14586 , which are both positionally coincident with snr ctb 109 ( figure [ 1e2259 ] , middle panel ) . for a detailed analysis of the 1e2259@xmath14586 region , we excluded the two sources from the spatial model .
the _ pointlike _ ts map of 1e2259@xmath14586 showed a source spatially coincident with the magnetar ( figure [ 1e2259 ] , left panel ) . to check for the possible extension of the gamma - ray source
, we performed the likelihood analysis in a similar way as was done for 1e1841@xmath8045 .
a disk of radius 0@xmath1920@xmath330@xmath1902 located at r.a.@xmath36345@xmath1946@xmath330@xmath1902 , decl.@xmath3658@xmath1988@xmath330@xmath1902 is favored over a point - like source with ts@xmath25=23.0 ( figure [ 1e2259 ] , left panel ) . the gamma - ray source is significantly detected as an extended source with an overall ts value of 74.5 .
the spectral parameters are reported in table [ table2 ] .
snr ctb 109 is associated with 1e2259@xmath14586 , which is reported as a point - like source by acero et al .
( 2016a ) and an extended source by castro et al .
( 2012 ) with a consistent extension .
the ts map as well as the radio map from the canadian galactic plane survey ( cgps , taylor et al .
2003 ) of the 1e2259@xmath14586 region are shown in figure [ 1e2259 ] , middle panel .
snr ctb 109 is clearly seen as a shell - like structure in the radio map around 1e2259@xmath14586 .
the disk radius derived from the gamma - ray data is consistent with the radio size of snr ctb 109 .
thus , the extended gamma - ray source detected around 1e2259@xmath14586 is most likely the gev counterpart of snr ctb 109 .
after we modelled snr ctb 109 as an extended source with spatial parameters fixed at the above values , 1e2259@xmath14586 is not detected ( figure [ 1e2259 ] , right panel ) .
the 95% flux upper limits in the 0.11 gev , 110 gev and 0.110 gev energy bands are reported in table 1 .
[ sgr1900 ] [ 3xmm ] swiftj1834.9@xmath80846 is associated with snr w41 and is in the same sky region as sgr1833@xmath80832 . as a result ,
the likelihood analysis for the two sources are carried out together .
snr w41 was detected by magic and h.e.s.s . at tev energies ( aharonian et al .
2005 ; albert et al .
2006 ) and gev emission is expected . the point - like gamma - ray source 3fgl j1834.5@xmath80841 from 3fgl and two point - like gamma - ray sources from the internal 6-years source list
are located in the vicinity of the two magnetars , and are positionally coincident with snr w41 ( figure [ swift1834 ] , middle panel ) .
snr w41 was once reported as an extended gamma - ray source with a disk radius of 0@xmath1933 by acero et al .
( 2016a ) , and by castro et al .
( 2013 ) . using a two - dimensional gaussian profile , the extension of snr w41 was reported as @xmath31=0@xmath1915 by abramowski et al .
( 2015 ) . for a detailed analysis of swiftj1834.9@xmath80846 and sgr1833@xmath80832 as well as exploring the possible gamma - ray extension of snr w41
, we excluded the two sources in the internal 6-years source list from the spatial model .
the likelihood analysis of this region showed an extended gamma - ray source spatially associated with swiftj1834.9@xmath80846 and sgr1833@xmath80832 ( figure [ swift1834 ] ) . to check for its possible extension
, we performed the likelihood analysis in a similar way as done for 1e1841@xmath8045 .
a disk of radius 0@xmath1931@xmath330@xmath1902 located at r.a.@xmath36 278@xmath1964@xmath330@xmath1902 , decl.@xmath36@xmath88@xmath1972@xmath330@xmath1902 is favored over a point - like source with ts@xmath25=180.4 .
the gamma - ray source is significantly detected as an extended source with an overall ts value of 508.6 .
the spectral parameters are reported in table [ table2 ] .
the spectral parameters are consistent with abramowski et al .
( 2015 ) and the extension is about twice the @xmath31 in abramowski et al .
( 2015 ) , which are consistent considering the different extension model used uniform disk model and two - dimensional gaussian profile ( lande et al .
the ts map as well as the radio map from vgps around swiftj1834.9@xmath80846 and sgr1833@xmath80832 are shown in figure [ swift1834 ] , left and middle panels .
the shell - like structure of snr w41 , which is associated with swiftj1834.9@xmath80846 , is apparent in figure [ swift1834 ] , middle panel .
no radio source is associated with sgr1833@xmath80832 .
the gamma - ray source detected is most likely the gev counterpart of snr w41 .
we modelled snr w41 as an extended source with spatial parameters fixed at the above values .
swiftj1834.9@xmath80846 and sgr1833@xmath80832 are not detected ( figure [ swift1834 ] , right panel ) .
the 95% flux upper limits in the 0.11 gev , 110 gev and 0.110 gev energy bands are reported in table 1 .
sgr1627@xmath841 is associated with snr [email protected] .
the snr was detected as an extended gamma - ray source with a disk radius of 0@xmath1929@xmath330@xmath1903 and ts@xmath3543 by acero et al .
( 2016a ) , and by castro et al .
the point - like gamma - ray source 3fgl j1636.2@xmath84734 from 3fgl , and two point - like gamma - ray sources from the internal 6-years source list are all in the vicinity of sgr1627@xmath841 and are positionally coincident with snr [email protected] ( figure [ sgr1627 ] .
middle panel ) . for a detailed analysis of sgr1627@xmath841
as well as exploring the possible gamma - ray extension of snr [email protected] , we excluded the two sources in the internal 6-years source list from the spatial model .
the likelihood analysis of this region showed an extended gamma - ray source positionally coincident with sgr1627@xmath841 .
to check for the possible extension of the gamma - ray source , we performed the likelihood analysis . since 3fgl j1636.2@xmath84734 and one of the above two sources
are described by a logparabola model in 3fgl and the internal 6-years source list , we adopted an extended disk with logparabola spectral shape to fit the data .
a disk of radius 0@[email protected] located at r.a.@xmath36249@xmath1900@xmath330@xmath1901 , decl.@xmath36@xmath847@xmath1954@xmath330@xmath1901 is favored over a point - like source with ts@xmath25=133.2 . the gamma - ray source is significantly detected as an extended source with an overall ts value of 1942.2 .
the spectral parameters are reported in table [ table2 ] .
the ts map as well as the radio map from the molonglo galactic plane survey 2nd epoch ( mgps-2 , murphy et al . , 2007 ) are shown in figure [ sgr1627 ] left and middle panels .
the radio complex of snr [email protected] is clearly seen in the radio map around sgr1627@xmath841 .
the gamma - ray source detected is most likely the gev counterpart of snr [email protected] .
we modelled snr [email protected] as an extended source with spatial parameters fixed at the above values .
sgr1627@xmath841 is not detected ( figure [ sgr1627 ] , right panel ) .
the 95% flux upper limits in the 0.11 gev , 110 gev and 0.110 gev energy bands are reported in table 1 .
cxouj171405.7@xmath8381031 is associated with snr ctb 37b while one gamma - ray point source from the internal 6-years source list is located within snr ctb 37b ( s3 in figure [ cxouj1714 ] , middle panel ) .
the gamma - ray point source 3fgl j1714.5@xmath83832 from 3fgl and one gamma - ray source from the internal 6-years source list ( s1 in figure [ cxouj1714 ] , middle panel ) , which are @xmath24 0@xmath194 away from cxouj171405.7@xmath8381031 , are positionally coincident with snr ctb 37a .
another two point sources ( s2 and s3 in figure [ cxouj1714 ] ) from the internal 6-years source list are in the vicinity .
the radio map from mgps-2 of this region is shown in figure [ cxouj1714 ] , middle panel and the two snrs are apparent in the radio map . although snr ctb 37a was once reported as a point source by acero et al .
( 2016a ) , to exclude its influence on the analysis of cxouj171405.7@xmath8381031 , we estimated its possible extension .
s1 , s2 and s3 were removed from the spatial mode . to check for the possible extension of snr ctb 37a , we performed the likelihood analysis as for the case of sgr1627@xmath841 .
since 3fgl j1714.5@xmath83832 and s1 were described by a logparabola model in 3fgl and the internal 6-years source list respectively , we adopted an extended disk with logparabola spectral shape to fit the data .
a disk of radius 0@xmath1918@xmath330@xmath1901 located at r.a.@xmath36258@xmath1965@xmath330@xmath1901 , decl.@xmath36@xmath838@xmath1955@xmath330@xmath1901 is favored over a point - like source with ts@xmath25=116.3 .
the gamma - ray source is significantly detected as an extended source with an overall ts value of 2597.2 .
the spectral parameters are reported in table [ table2 ] .
the ts map of snr ctb 37a is shown in figure [ cxouj1714 ] , left panel .
snr ctb 37a is bright and cxouj171405.7@xmath8381031 is not seen in the ts map ( figure [ cxouj1714 ] , left panel ) . in order to search for gamma - ray emission from cxouj171405.7@xmath8381031 ,
we need to exclude the gamma - ray emission from snr ctb 37a .
we modelled snr ctb 37a as an extended source with spatial parameters fixed at the above values , included s2 and cxouj171405.7@xmath8381031 in the spatial model and performed the likelihood analysis .
in the residual map ( figure [ cxouj1714 ] , right panel ) , cxouj171405.7@xmath8381031 is positionally coincident with the peak of a gamma - ray source .
we checked for its possible extension but it is not significant ( ts@xmath25=5.42 ) . by assuming a power law spectral shape ,
the _ gtlike _ analysis of this gamma - ray source resulted in a ts value of 51.0 and a photon index of [email protected] .
the flux level is estimated as @xmath38 erg @xmath39s@xmath40 in the 0.1300 gev range .
we explored the presence of a spectral cut - off but it is not significant ( below 3 @xmath31 ) .
we could not further distinguish if the source is associated with snr ctb 37b or cxouj171405.7@xmath8381031 . by assuming a power law spectral model , we derived 95% flux upper limits in the 0.11 gev , 110 gev and 0.110 gev energy bands .
the results are reported in table 1 .
sgr0501@xmath144516 is adjacent to snr hb9 .
the radio map from vgps of this region is shown in figure [ sgr0501 ] , middle panel .
the shell - like structure of snr hb9 is clearly seen .
the snr was detected as an extended gamma - ray source with a disk radius of 1@xmath192@xmath330@xmath193 and a logparabola spectral shape by araya ( 2014 ) .
two point - like gamma - ray sources from the internal 6-years source list are positionally coincident with snr hb9 ( s1 and s2 in figure [ sgr0501 ] , middle panel ) . for a detailed analysis of sgr0501@xmath144516
as well as exploring the possible gamma - ray extension of snr hb9 , we excluded s1 and s2 from the spatial model .
the likelihood analysis of this region showed an extended gamma - ray source positionally coincident with snr hb9 . to check for the possible extension of the gamma - ray source
, we performed the likelihood analysis as in the case of sgr1627@xmath841 .
we adopted an extended disk with logparabola spectral shape to fit the data , as in araya ( 2014 ) .
a disk of radius 1@xmath1932@xmath330@xmath1905 located at r.a.@xmath3675@xmath1939@xmath330@xmath1905 , decl.@xmath41@xmath846@xmath1950@xmath33 0@xmath1906 is favored over a point - like source with ts@xmath25=60.4 .
the gamma - ray source is significantly detected as an extended source with an overall ts value of 294.3 .
the spectral parameters are reported in table [ table2 ] .
we modelled snr hb9 as an extended source with spatial parameters fixed at the above values .
sgr0501@xmath144516 is not detected ( figure [ sgr0501 ] , right panel ) .
the 95% flux upper limits in the 0.11 gev , 110 gev and 0.110 gev energy bands are reported in table 1 .
400910 in 0.1 - 300 gev with a new source added in the analysis ( cyan cross ) .
the x - axis and y - axis are r.a . and decl .
referenced to j2000 .
the magnetars are shown with a green cross while sources from the internal 6-years source list are shown with a white cross .
the spatial models used to produce ts maps are described in the text . ]
sgr1900@xmath1414 is adjacent to snr [email protected] and snr [email protected] .
the radio map from vgps of this region is shown in figure [ sgr1900 ] , middle panel .
one point - like gamma - ray source from the internal 6-years source list is located only 0@xmath1907 away from sgr1900@xmath1414 ( s1 in figure [ sgr1900 ] , middle panel ) . for a detailed analysis of sgr1900@xmath1414
, we excluded the source from the spatial model .
the likelihood analysis of this region showed an extended gamma - ray source positionally coincident with sgr1900@xmath1414 , snr [email protected] and snr [email protected] . to check for the possible extension of the gamma - ray source
, we performed the likelihood analysis in a similar way as done for 1e1841@xmath8045 .
a disk of radius 0@xmath1960@xmath330@xmath1906 located at r.a.@xmath36287@xmath1916@xmath330@xmath1905 , decl.@xmath419@xmath1926@xmath330@xmath1905 is favored over a point - like source with ts@xmath25=54.7 . the gamma - ray source is significantly detected as an extended source with an overall ts value of 259.6 .
the spectral parameters are reported in table [ table2 ] . besides snr [email protected] and snr [email protected] , there are other radio sources positionally coincident with this extended gamma - ray source , which may also have contributed to the gamma - ray emission .
we modelled it as an extended source with spatial parameters fixed at the above values .
sgr1900@xmath1414 is not detected ( figure [ sgr1900 ] , right panel ) .
the 95% flux upper limits in the 0.11 gev , 110 gev and 0.110 gev energy bands are reported in table 1 .
3xmmj185246.6@xmath14003317 is near snr kes 79 which is spatially coincident with one gamma - ray source from the internal 6-years source list .
the radio map from vgps of this region is shown in figure [ 3xmm ] , middle panel .
the shell - like structure of snr kes 79 is clearly seen .
the snr was detected as a gamma - ray source by auchettl et al .
the likelihood analysis of this region showed an extended gamma - ray source positionally coincident with snr kes 79 ( figure [ 3xmm ] , left panel ) .
for a detailed analysis of 3xmmj185246.6@xmath14003317 as well as exploring the possible gamma - ray extension of snr kes 79 , we excluded the gamma - ray source in the internal 6-years source list from the spatial model . to check for the possible extension of the gamma - ray source
, we performed the likelihood analysis in a similar way as done for 1e1841@xmath8045 .
a disk of radius 0@xmath1963@xmath330@xmath1913 located at r.a.@xmath36 283@xmath1910@xmath330@xmath1903 , decl.@xmath360@xmath1963@xmath330@xmath1914 is favored over a point - like source with ts@xmath25=108.5 . the gamma - ray source is significantly detected as an extended source with an overall ts value of 339.5 .
the extension of the gamma - ray source is larger than the size of the radio shell of snr kes 79 ( figure [ 3xmm ] , middle panel ) , which may originate from the interaction between snr kes 79 and the dense molecular clouds around ( auchettl et al . 2014 ) .
the spectral parameters are reported in table [ table2 ] .
we modelled it as an extended source with spatial parameters fixed at the above - quoted values .
3xmmj185246.6@xmath14003317 is not detected ( figure [ 3xmm ] , right panel ) .
the 95% flux upper limits in the 0.11 gev , 110 gev and 0.110 gev energy bands are reported in table 1 .
1rxsj170849.0@xmath8400910 is in a crowded region on the galactic plane .
an additional point source modelled by a simple power law was added to the spacial model .
the best position of the additional source was determined with _ pointlike _ as r.a.@xmath36 257@xmath1986@xmath330@xmath1904 , decl.@xmath36 - 39@xmath1958@xmath330@xmath1904 .
there is no significant ts excess in the ts map .
the _ gtlike _
ts value for 1rxsj170849.0@xmath8400910 is 6.6 and no clearly resolved source was seen coincident with it ( figure [ 1rxs ] , right panel ) .
the 95% flux upper limits in the 0.11 gev , 110 gev and 0.110 gev energy bands are reported in table 1 .
the new source yields a ts value of 197.7 , a spectral index of @xmath420.01 and a flux level of @xmath4310@xmath44 erg @xmath39s@xmath40 .
[ table1 ] llllccccc + + snr name & ts & ra ( j2000 ) & dec ( j2000 ) & ts@xmath25 & radius & spectral index & fluxes & associated / adjacent magnetars + & & degree & degree & & degree & & & + + + + ctb 109 & 74.5 & [email protected] & [email protected] & 23.0 & [email protected] & [email protected] & 1.44@xmath33 0.28 & 1e2259@xmath14586/ + [email protected] ; [email protected] & 259.6 & [email protected] & [email protected] & 54.7 & [email protected] & [email protected] & [email protected] & /sgr1900@xmath1414 + hb9@xmath45 & 294.3 & 75.39 @xmath33 0.05 & [email protected] & 60.4 & [email protected] & [email protected] & [email protected] & /sgr0501@xmath144516 + kes 73@xmath16 & 323.2 & [email protected] & @[email protected] & 54.4 & [email protected] & [email protected] & [email protected] & 1e1841@xmath8045/ + kes 79 & 339.5 & [email protected] & [email protected] & 108.5 & [email protected] & [email protected] & [email protected] & /3xmmj185246.6@xmath14003317 + w41 & 508.6 & [email protected] & @[email protected] & 180.4 & [email protected] & [email protected] & [email protected] & swiftj1834@xmath80846/sgr1833@xmath80832 + [email protected]@xmath45 & 1942.2 & [email protected] & @[email protected] & 133.2 & [email protected]&[email protected] & [email protected] & sgr1627@xmath841/ + ctb 37a@xmath45 & 2597.2 & 258.65 @xmath330.01 & @[email protected] & 116.3 & [email protected] & [email protected] & [email protected] & /cxou j171405.7@xmath8381031 +
pulsar rotational phases for each gamma - ray photon that passed the selection criteria could be calculated using tempo2 ( hobbs et al .
2006 ) with the _ fermi _ plug - in ( ray et al .
the significance of gamma - ray pulsations was evaluated with the weighted h - test ( de jager et al .
1989 ; de jager & b@xmath46sching 2010 ) .
however , the prerequisite is to have an available precise ephemeris for the magnetars , valid during the gamma - ray data coverage of _ fermi_-lat . because of the noisy timing nature of magnetars , valid ephemerides during the _
fermi_-lat data coverage were not available for all the magnetars studied . adopting the ephemeris measured by dib et al .
( 2014 ) with 16 years of rxte data , we searched for gamma - ray pulsations from 1e1841@xmath8045 , 1e2259@xmath14586 , 1rxsj170849.0@xmath8400910 and 4u0142@xmath14614 using h - test ( hou et al .
we searched for gamma - ray pulsations for all photons below 10 gev , exploring different values for the minimum energy ( from 100 mev to 2 gev in a step of 100 mev ) and radius ( from 0@xmath191 to @xmath47 in a step of 0@xmath191 ) of the _ fermi_-lat data .
we first considered _
fermi_-lat data which are contemporaneous with rxte which operated until the end of 2011 , and then extended the last available ephemeris to the rest of _
fermi_-lat data . no statistically significant detection of pulsations was found .
the most relevant ( yet insignificant ) result is obtained for the magnetar 4u0142@xmath14614 using _
fermi_-lat data with available ephemeris coverage . in the 0.410 gev energy range with a data radius of 1@xmath193 ,
we obtain 3.7 @xmath31 before and 1.7 @xmath31 after trials .
in this paper , we searched for gamma - ray emission from magnetars using 6 years of _ fermi_-lat data , which greatly extended the results of abdo et al .
( 2010a ) . with this larger data
set , and with a total of 20 magnetars searched , we can confirm that no gamma - ray emission is detected for any of the sources studied .
twelve out of twenty magnetars are singled out by a _
gtlike _ analysis as having ts values below 25 ( table 1 ) .
cxou j171405.7@xmath8381031 is positionally coincident with a point source .
however , we could not further distinguish if the emission originates from the magnetar or from snr ctb 37b .
for all the studied magnetars we derived the deepest upper limits to date in the 0.1300 gev energy range .
gamma - ray pulsations were searched for 4 magnetars having reliable ephemerides covering most of the _ fermi_-lat data span , but no periodicity was significantly observed . by applying an outer - gap model to magnetars , cheng & zhang ( 2001 ) and zhang & cheng ( 2002 ) predicted detectable gamma - ray emission from sgr 1900 + 14 and five others axps within one year of _
fermi_-lat observations .
these predicted flux levels are much beyond the currently imposed upper limits .
admittedly , these models are based on a number of assumptions and parameters ( see vigan et al .
( 2015a ) for a detailed discussion ) .
however , it would be unrealistic to suppose that these parameters ( e.g. , the inclination angle ) should be such to secure non - detections in all cases .
this point was analyzed by tong et al .
( 2010 ) when considering the reasons for the early non - detection of 4u 0142 + 61 .
for this particular case , even doubling the distance and using an outer gap location at 10 stellar radii or beyond the flux would have been larger than the upper limit found ( see tong et al .
2010 for details ) .
takata et al .
2013 proposed a possible scenario for gev gamma - ray emission from magnetars , in which the magnetic energy released from crust cracking could be carried into outer magnetosphere by alfv@xmath48n waves . in this case , the predicted gamma - ray flux for 1e2259@xmath14586 is higher than the upper limits we derived .
vigan et al .
( 2015b ) proposed an outer - gap model for gamma - ray pulsars which follows the particle dynamics to consistently compute the emission of synchro - curvature radiation .
they have applied this model to both the phase - averaged and phase - resolved spectra of the brightest pulsars . by fitting the model to all observed _ fermi_-lat pulsar spectral data
, they found a strong correlation between the accelerating electric field and the magnetic field at the light cylinder ( b@xmath49 , vigan et al .
if the correlations they found hold up to the magnetar regime ( where at the light cylinder ( lc ) , b@xmath49 @xmath24 10@xmath40 - 10@xmath50 g , p @xmath24 2 - 10 s ) , the accelerating electric field in the gap would be just too weak to provide particles energetic enough as to emit gamma rays .
their predictions that magnetars should not emit gamma rays via synchro - curvature radiation is consistent with the observational results of this paper .
analyzing the gamma - ray emission of the regions around magnetars , we have significantly detected 7 snrs , four of which are believed to be associated to the spatially coincident magnetar .
we have now studied them using pass 8 data , updating their morphology and spectral parameters with respect to the 3fgl catalog ( acero et al .
2015 ) and acero et al .
the gamma - ray morphologies and luminosities of these magnetars are in line with what is expected for remnants associated with normal radio pulsars at the same age ( see also martin et al .
2014 for an x - ray comparison ) .
the _ fermi _ lat collaboration acknowledges generous ongoing support from a number of agencies and institutes that have supported both the development and the operation of the lat as well as scientific data analysis .
these include the national aeronautics and space administration and the department of energy in the united states , the commissariat lenergie atomique and the centre national de la recherche scientifique / institut national de physique nuclaire et de physique des particules in france , the agenzia spaziale italiana and the istituto nazionale di fisica nucleare in italy , the ministry of education , culture , sports , science and technology ( mext ) , high energy accelerator research organization ( kek ) and japan aerospace exploration agency ( jaxa ) in japan , and the k. a. wallenberg foundation , the swedish research council and the swedish national space board in sweden .
additional support for science analysis during the operations phase is gratefully acknowledged from the istituto nazionale di astrofisica in italy and the centre national dtudes spatiales in france .
we acknowledge the support from the grants aya2015 - 71042-p , sgr 2014 - 1073 and the national natural science foundation of china via nsfc-11473027 , nsfc-11503078 , nsfc-11133002 , nsfc-11103020 , xtp project xda 04060604 and the strategic priority research program
the emergence of cosmological structures " of the chinese academy of sciences , grant no .
nr is further supported an nwo vidi award a.2320.0076 , and via the european cost action mp1304 ( newcompstar ) .
99 abdo , a. a. , ackermann , m. , ajello , m. , et al .
2010a , apj , 725 , 73 abdo , a. a. , ackermann , m. , ajello , m. , et al .
2010b , a&a , 512 , 7 abdo , a. a. , ackermann , m. , ajello , m. , et al .
2010c , a&a , 523 , 46 acero , f. , ackermann , m. , ajello , m. , et al .
2015 , apjs , 218 , 23 ( 3fgl ) acero , f. , ackermann , m. , ajello , m. , et al .
2016a , apjs , 224 , 8 acero , f. , ackermann , m. , ajello , m. , et al .
2016b , apjs , 223 , 26 abramowski , a. , aharonian , f. , ait benkhali , f. , et al .
2015 , a&a , 574 , 27 aharonian , f. , akhperjanian , a. g. , aye , k .-
, et al . 2005 , science , 307 , 1938 albert , j. , aliu , e. , anderhub , h. , et al .
2006 , apj , 643 , 53 an , h. ; archibald , r. , f. , hasco@xmath51 t , r. , et al .
2015 , apj , 807 , 93 araya , m , 2014 , mnras , 1 , 860 cheng , k. s. , & zhang , l. 2001 , apj , 562 , 918 caliandro , g. a. , hill , a. b. , torres , d. f. , et al . 2013 .
mnras , 436 , 740 castro , d. , slane , p. , elission , d. c. , patnaude , d. j. , 2012 , apj , 756 , 88 .
castro , d. , slane , p. , carlton , a. , e. figueroa - feliciano , 2013 , apj , 774 , 36 .
de jager , o. c. , & b@xmath46sching , i. 2010 , a&a , 517 , l9 de jager , o. c. , raubenheimer , b. c. , & swanepoel , j. w. h. 1989 , a&a , 221 , 180 dib , r. & kaspi , v. , m. , apj , 784 , 37 duncan , r. c. , & thompson , c. , 1992 , apj , 392 , 9 enoto , t. , nakazawa , k. , makishima , k. , et al .
2010 , , 722 , l162 helene , o. , 1983 , nimpr , 212 , 319 hou , x. , smith , d. a. , guillemot , l. , et al . , 2014 ,
a&a , 570 , 44 kerr , m. 2011 , phd thesis , univ .
washington kuiper , l. , hermsen , w. & mendez , m. , 2004 , apj , 613 , 1173 lande , j. , ackermann , m. , allafort , a. , et al . , 2012 ,
apj , 756 , 5 mereghetti , s , 2008 , a & arv , 15 , 225 stil , j. m. , taylor , a. r. , dickey , j. m. , et al .
2006 , aj 132 , 1158 takata , j. , wang , y. , wu , e. m. h. , cheng , k. s. , 2013 , mnras , 431 , 2645 taylor , a. r. , gibson , s. j. , peracaula , m. , et al . , 2003 ,
aj , 125 , 3145 | we report on the search for gamma - ray emission from 20 magnetars using 6 years of _ fermi _ large area telescope ( lat ) observations .
no significant evidence for gamma - ray emission from any of the currently - known magnetars is found .
we derived the most stringent upper limits to date on the 0.110 gev emission of galactic magnetars , which are estimated between @xmath02 .
gamma - ray pulsations were searched for the four magnetars having reliable ephemerides over the observing period , but none were detected .
on the other hand , we also studied the gamma - ray morphology and spectra of seven supernova remnants associated or adjacent to the magnetars . |
the age , chemical and kinematic distributions of galactic stellar populations provide powerful constraints on models of the formation and evolution of the milky way .
the galactic globular clusters ( ggc ) constitute an especially useful case because the stars within individual clusters are coeval and spatially distinct . but
uncertainties in the determination of distances to ggcs , along with uncertainties in the determination of their physical characteristics such as age or metallicity can result when the reddenig produced by interstellar dust is not properly taken into account @xcite . in this context ,
differential reddening across the field of the cluster has proven to be difficult to map , and the properties of ggcs that suffer high and patchy extinction , especially those in the direction of the inner galaxy , have not been accurately measured @xcite .
several authors have used different methods to try to map differential extinction in ggcs before : @xcite have used colors of variable rrlyrae stars to create the reddening maps ; @xcite and @xcite have used photometric studies of the stars in the horizontal branch ( hb ) ; @xcite and @xcite have used photometric studies of main sequence ( ms ) , subgiant branch ( sgb ) , and red giant branch ( rgb ) stars . in this paper
we describe a new dereddening technique , based on this last approach , but with the improvement of calculating the extinction not on a predetermined grid but on a star by star basis . in section 2
we extensively explain this dereddening method and in section 3 we apply it to an example cluster , m62 . in the next papers of this series we plan to apply this dereddening method to a sample of clusters in the direction of the inner galaxy that in principle can be affected by this effect , and analyze the obtained results . in paper
ii we will present a new photometric database consisting of a sample of 25 inner ggcs observed using the magellan 6.5 m telescope and the _ hubble space telescope _
( _ hst _ ) , describe the processes we follow to obtain an accurate astrometric and photometric calibration of the stars in these clusters , and , after applying the dereddening technique , give the extinction maps along the field of the clusters in the sample , along with new cleaner , differentially dereddened color - magnitude diagrams ( cmd ) of these ggcs . in paper iii we will provide an analysis of the stellar populations of these clusters based on their dereddened cmds . finally , in paper iv
, we will characterize the interstellar medium along the low - latitude lines of sight were the clusters lie , based on the properties of the extinction maps we derived .
our dereddening technique is composed of five iterative steps ( see figure [ figflowchart ] ) : * we assign a probability to the stars in our observed regions to belong to the cluster or to the galactic field depending on their positions in the sky and in the cmd ( see section [ sectprob ] ) .
* we build a ridgeline for the stellar population of the observed cluster .
the ridgeline represents the evolutionary path that a given star is going to follow in the cmd ( see section [ sectridge ] ) . *
we assign an individual extinction value to every star , based on its displacement from the ridgeline along the reddening vector ( see section [ sectext ] ) . *
we smooth the different individual color excesses over the observed field to generate an extinction map ( see section [ sectmap ] ) . *
we apply the extinction values from the map to the stars in our observations to construct a dereddened cmd ( see section [ sectcmd ] ) .
we repeat these steps iteratively .
every iteration the ridgeline is more accurately defined , and therefore the extinction map and the dereddened cmd are also more precise .
we end up the process when there is a convergence in the calculated ridgeline .
this technique is based on work by @xcite and @xcite , and pioneered by @xcite , but unlike them , we do not divide the field in a grid of well established subregions a priori , but use a non - parametric approximation to smooth the information about the reddening in the field provided by every star , without such hard edges .
the success of this analysis depends critically on the assumption that the stellar populations are uniform within individual gcs .
recent studies tend to suggest that this is not strictly true , especially in the most massive ggcs and ggcs with an extreme blue horizontal branch ( ebhb ) @xcite .
in the cases discovered , it is conjectured that the spread in helium content between stars in the cluster is important , but the spread in age and metallicity seems to be small @xcite , making the spread of the stars in the sgb and upper rgb regions of the cmd also small .
since most of the information in our method comes from stars in these regions of the cmds , our technique should not be significantly affected .
what is more , in our method variations in the reddening must be spatially related , which should not be the case if there are multiple helium enriched populations within the cluster .
because of these reasons and since our sample only includes a few massive and ebhb clusters , we expect the differential populations effects to be comparatively minor and not significantly affect our dereddening approach .
also it is important to realize that our dereddening method does not establish the absolute extinction toward a target cluster , so to estimate the absolute extinction in each case we need to use other methods , e.g. , comparison of our differentially dereddened cmds with isochrone models , or use of reddening estimates from rr lyr stars in our fields . from our photometric studies we can assign probabilities to the stars in our observations to belong to the clusters , based on @xmath0 , their position in the sky with respect to the center of the cluster , and also on ( @xmath1 ) , their color and magnitude position on the cmd .
these probabilities will be used in the next steps of our technique to give higher weight to the information provided by stars with high probabilities of being cluster members . before going any further ,
it is convenient to explain the notation that we are going to use , and to remember the basic rules of probability calculation .
the marginal or unconditional probability of an event a happening is expressed @xmath2 .
the joint probability of an event a and an event b happening at the same time is expressed @xmath3 .
finally , the conditional probability of an event a happening given the occurrence of some other event b is expressed @xmath4 , and can be written as @xmath5 or , using bayes theorem , as @xmath6 also , notice that in many parts of our analysis it is going to be more convenient to use probability density functions , @xmath7 , to express the probability .
these functions describe the relative likelihood for a random variable x to occur at a given point in the sampled space @xmath8 and satisfy @xmath9 @xmath10 @xmath11 whenever needed in this work , the probability density functions are calculated non - parametrically using _ locfit_. _
@xcite is a local likelihood estimation software implemented in the r statistical programming language ( see appendix ) .
_ locfit _ does not constrain the probability density functions globally , i.e. , it is non - parametric , but assumes that locally , in a local window around a certain point in the sample space , the functions can be well approximated by a polynomial . hence the main inputs that _
locfit _ requires are the positions of the observed elements in the sample space , and a parameter to define the local window where the function is approximated parametrically , i.e. , a smoothing factor given by the maximum of two elements : a bandwidth generated by a nearest neighbor fraction , and a constant bandwidth .
now , if we suppose the variable @xmath12 indicates membership to the cluster ( @xmath13 indicates observation of a member , and @xmath14 indicates observation of a non - member ) , in the next paragraphs we calculate @xmath15 , the conditional probability of the stars being members of the cluster , given their position in the sky and in the cmd . actually , it is more convenient for us to calculate first @xmath16 and @xmath17 , the conditional probabilities of the stars to belong to the cluster given _ just _ their position in the sky , and _ just _ their position in the cmd , and only after attempt to calculate @xmath15 . the conditional probability of the stars being members of the cluster , given _
just _ their position in the sky as a function of the distance to the center of the cluster , @xmath16 , can be written , using bayes theorem , as @xmath18 for any given cluster , both @xmath19 and @xmath20 are unknown .
but since @xmath20 is the ratio of the observed cluster stars to the total observed stars , we can rewrite @xmath21 as @xmath22 where @xmath23 is the number of observed member stars as a function of the distance to the cluster center , and @xmath24 is the total number of observed stars as a function of the distance to the cluster center . we still do not know @xmath23 , but if we divide both numerator and denominator in @xmath25 by @xmath26 , the area coverage at a given @xmath0 of the field of view ( fov ) in our observation of the cluster , we can rewrite @xmath25 as @xmath27 where @xmath28 is the surface density of observed member stars as a function of the distance to the cluster center , and @xmath29 is the total surface density of observed stars as a function of the distance to the cluster center .
now , if we assume an empirical king profile @xcite for the cluster , we can write the surface density of member stars as @xmath30^{1/2}}-\frac{1}{[1+(r_t / r_c)^2]^{1/2}}\right)^2 & \mbox{if } & r \leq r_t \\ 0 & \mbox{if } & r > r_t \end{array } \right.\ ] ] where @xmath31 and @xmath32 are the core radius and the tidal radius of the cluster .
if we make another assumption and consider a constant surface density of stars for the non - member population , which is reasonable due to the small size of our field , then @xmath33 and we have the following functional form for the total surface density distribution of stars observed in our fov @xmath34 and , according to @xmath35 , we can write the conditional probability of the stars being members of the cluster , given their position in the sky , as @xmath36 or , alternatively , @xmath37 therefore to calculate @xmath16 we need to find the coefficients @xmath38 and @xmath39 , since @xmath40 can be obtained taking the values for @xmath31 and @xmath32 for every cluster in our sample provided in the harris catalog @xcite . these two remaining coefficients , @xmath38 and @xmath39 , can be easily found by a least squares fit of @xmath40 to the observed surface density of stars @xmath29 ( see @xmath41 , figure [ figprobdis ] , and table [ tabking ] ) . to find the observed @xmath29 , we need to first calculate the number of stars @xmath24 , and then divide it by the area coverage @xmath26 .
as we mentioned when passing from @xmath21 to @xmath25 , the number of stars @xmath24 is just the probability density @xmath42 of stars in our observation , multiplied by the total number of observed stars .
we calculated the probability density @xmath42 of stars by feeding _ locfit _ with the radial positions of the stars and a smoothing factor with a constant bandwidth of 0.25 arcmin .
the surface coverage @xmath26 in our observation is not trivial to find , since our fov is a square not centered in the cluster center . in order to quickly calculate this area
, we create a grid of points equally spaced over our fov , calculate the probability density @xmath42 of points , and then multiply by the area of the fov .
we calculated the probability density @xmath42 of points by feeding _ locfit _ with the radial positions of the points in the grid and a smoothing factor with a constant bandwidth of 0.25 arcmin .
the conditional probability of the stars being members of the cluster , given _
just _ their position in the color magnitude diagram , @xmath17 , can be written , using bayes theorem , as @xmath43 for any given cluster , we do not know the true distribution in magnitude and color in our observations of genuine gc members , but we can model the distribution of the field , non - member stars .
we use the model of the galaxy described in @xcite , from now on referred as the besanon model , to obtain @xmath44 .
we can now rewrite @xmath45 as @xmath46 where @xmath44 is obtained feeding _ locfit _ with the position in the cmd of the galactic field stars provided by the besanon model and a smoothing factor with a nearest neighbor ratio of 0.01 , @xmath47 is obtained feeding _ locfit _ with the position in the cmd of the stars in our observations and a smoothing factor with a nearest neighbor ratio of 0.01 , and
@xmath48 is just the ratio of the total number of modeled non - member stars for an area equal to the fov of the observation , to the total number of observed stars , which include field and cluster stars . to take care of the different variation ranges in color and in magnitude for the stars in the cmd a scale ( see appendix ) of 1 in color to 5 in magnitude
is also provided to _
locfit _ as an input parameter to calculate @xmath44 and @xmath47 . in figure [ figprobdcm ]
we can see the different @xmath17 for the stars in m62 , one of the clusters in our sample .
similar methods to calculate @xmath17 have been employed by @xcite based on calculations by @xcite and @xcite , and also by @xcite based on calculations by @xcite and @xcite .
in contrast to these other studies we do not eliminate stars from our analysis based on these probabilities , but downweight them . also , as we show at the end of this section , we do not just use @xmath17 , but @xmath15 in our following analysis .
given the importance of the besanon model ( and the cmds it provides ) in analyzing the galactic field component of our observations and calculating @xmath49 , we describe here in more detail the input parameters that we use in the web interface of the besanon group to model the non - member stellar populations in the vicinity of the sampled clusters . * *
field of view .
* we use the small field option from the web interface , in which the modeled stars are all supposed to be at the same coordinates ( see table [ tabbesancon ] ) , implicitly assuming that the field star density gradient across the fov is negligible .
the same assumption was made in the calculation of @xmath50 ( see @xmath51 ) . to obtain higher statistical significance in the models
, we can increase the selection area to obtain a larger number of stars in the model .
@xmath44 is independent of the area and the number of stars used , since it is normalized ( see @xmath52 ) . in practice , we set this parameter to select @xmath53 stars ( see table [ tabbesancon ] ) .
we have to notice though , that when calculating @xmath48 in @xmath54 we need to normalize the number of stars in the model to the fov .
* * extinction law .
* we apply _ locfit _ to map @xmath47 as a function of @xmath39 and @xmath55 using the cmds from our observations . from the modeled stars
, we can also use _ locfit _ to map @xmath44 as a function of @xmath39 and @xmath55 , realizing that this map changes depending on what extinction @xmath56 we assumed for our modeled data .
so if we define @xmath57 as @xmath58\ ] ] we need to realize that this function depends also on the extinction assumed for the modeled cmds , i.e. , @xmath59 . for the fields studied , we choose the modeled stars to simulate no interstellar extinction in the beginning
. then we move the density maps of the models along the reddening vector in increments of extinction @xmath60 , calculating @xmath57 for every increment .
the value of @xmath56 where @xmath57 is a maximum is the value we choose for the absolute extinction of our observation .
this @xmath56 is generally equal or within a few hundredths of a magnitude to the one provided in @xcite .
the extinction is then simulated in the model as a cloud with an extinction equal to this @xmath56 at distance of 0 pc ( see table [ tabbesancon ] ) . * * characteristics of the stars used in the model . *
we allow all ages , galactic components , and spectral types provided by the besanon model .
we provide a limit only in the intervals of magnitudes , equal to the ones of our observations ( table [ tabbesancon ] ) .
we also model our observational errors @xmath61 with an exponential function of the apparent magnitude _
m _ in each band : @xmath62
so the models will take the photometric error into account providing a more realistic approximation ( see table [ tabbesancon ] ) .
certain problems arise in this procedure of calculating the conditional probabilities @xmath16 and @xmath17 , including : * we have to take into account the completeness factor of our photometric observations . in general
, completeness decreases with magnitude somewhat smoothly at the faint end , but suddenly at the saturation limit .
because of this , and since we are trying to calculate the probabilities comparing our observations with models ( king profile and besanon model ) we need to find where the completeness factor starts to decrease at the faint end . usually this is done with artificial - star tests , injecting stars in the cmd and analyzing the amount and magnitudes of the ones recovered @xcite .
but our large sample and large cluster star density gradients within individual fields make artificial star tests highly inefficient , so we explore another method that is well - suited to our specific fields , and also much more efficient to implement . from the besanon model
we can calculate the number of non - member stars per area in the field that we have in a given magnitude range , and from our data we can also find the number of non - member stars per area in the field in a given magnitude range just by fitting @xmath63 in that magnitude range .
so we iteratively calculate both numbers making the magnitude range smaller taking away stars in the 0.1 mag fainter end every time we do the calculation , repeating the process until we are able to get the cumulative luminosity functions ( lf ) for the seven fainter magnitudes for the non - member stars from every one of our pointings .
then we can compare the modeled and observed lfs to see where the functions deviate one from the other .
but this approximation proves to be not very accurate , especially in cases in which the number of non - member stars in the field is small .
instead we explore yet another approximation .
we notice that the lf in the besanon model is always a concave function , while the observed lf of the non - member stars in the field becomes a convex function at the faint end .
we can then identify the limit where the completeness factor starts to decrease as the inflection point where the observational cumulative lf changes from concave to convex .
the inflection point is the point where the derivative of the lf has a extreme , a minimum in our cases . and the derivative is easily found using _ locfit _ to calculate non - parametrically the lf and its derivative ( see appendix ) .
the completeness limit is usually located in our observations at @xmath64 magnitudes from the faint end in @xmath65 ( see table [ tabclimit ] ) . in the next steps of our dereddening method
we only use stars brighter than this limit .
* the completeness factor also depends on the distance to the center of the cluster , strongly so for the more crowded cases .
close to the cluster center we are not only missing stars at the faint end of the magnitude distribution , but at all magnitudes , since the lower completeness factor in this region is mostly caused by masking saturated regions by dophot @xcite , the program used to do the photometric analysis of the images . from the calculation of the @xmath16 , we can see that this reduced completeness factor reveals itself as a deviation from the model fit shown in @xmath66 at distances close to the center of the cluster ( see figure [ figprobdis ] ) .
if we assume that we are missing an equal percentage of stars from the cluster and from the field , we can calculate its effect as the ratio between the observed and modeled @xmath29 . in the cmd
, this lower completeness factor can lead to a miscalculation of @xmath17 if the fov is small .
however , since the area of the regions whose extinction we are able to map are much bigger than the areas where the reduced completeness factor is an issue , the comparison of our observations with the besanon model is not significantly affected .
the only appreciable consequence is that the calculated @xmath48 is a little higher than the actual @xmath48 . but
this high value for @xmath48 can be corrected if when we calculate the number of stars , we take into account the deviation of the observation from the model fit as a function of the distance to the center of the cluster , as we mentioned before .
* @xmath16 is obtained using the values for the core and tidal radii provided in the harris catalog @xcite , for every cluster in our sample .
notice that these values are not all of equal accuracy .
small variations in the initially adopted @xmath31 and @xmath32 could produce small changes in the individual @xmath16 values , but they will not alter the general trend ( stars closer to the center of the cluster have higher @xmath67 than stars located further away ) , and therefore they will not significantly affect the result of our dereddening technique .
* @xmath17 is obtained comparing a real observation with a model , where photometric errors are incorporated by mimicking the real photometric errors by the exponential function described above . in general
, we observe that stars from the models tend to be a little more concentrated along the different field evolutionary sequences in the cmds than stars from real observations .
therefore , for a given magnitude , @xmath49 peaks higher and decreases faster in the model than in the real observation , producing two main errors in our calculations of @xmath17 .
the first error is that we mistakenly assign a probability of being in the field which is too low , and therefore a probability of being cluster members which is too high , to stars in the regions where the observed color is an extreme for a given magnitude ( see figure [ figprobdcm ] ) .
but since there are not many stars in these regions and they usually have higher photometric errors in color than our functions describe , they are downweighted or removed from our calculations later on ( see section [ sectmap ] ) .
more important is the second error , the case were @xmath49 peaks higher in the model than in reality , making the probability of the star to be a non - member in the field too high , and @xmath17 too low , lower than 0 . in order to avoid these non - physical cases , and even too extreme cases that can give too low a weight value to the information coming from those stars
, we put the hard limit @xmath68 ( see figure [ figprobdcm ] ) .
now we are in conditions to calculate @xmath15 , the conditional probability of the stars being members of the cluster given both their position in the sky and in the cmd . according to bayes theorem , @xmath69 and
since the positions in the sky and in the cmd for non - member stars are independent , the variables are not independent , i.e. , the position in the cmd for a star in our observation is related to its spatial position in the sky , due to the mixture of cluster and field populations . hence @xmath70 . ]
@xmath71 and using bayes theorem again ( see @xmath21 ) we can write @xmath72 @xmath73 , @xmath42 , and @xmath44 , the three elements in the numerator have been calculated previously , so now we just need to find the denominator @xmath74 , the probability density of stars in our observation as a function of both position in the sky and in the cmd . to calculate it we feed
_ locfit _ with the position in the sky and in the cmd of the stars in our observations and a smoothing factor with a nearest neighbor ratio of 0.1 .
since the units of the three physical magnitudes @xmath75 are different , we need to feed _ locfit _ with a scale factor ( see appendix ) to calculate the distance between neighbors in the @xmath75 space .
experimentally we found that a scale of 1 magnitude in color to 5 magnitudes in brightness and 10 arcmin in distance to the the gc center provides good results , and it is the one we used as an input parameter of _ locfit _ to calculate @xmath74 .
the results of the @xmath15 calculation are shown in figure [ figprob ] .
we can observe that the same problems that were previously mentioned affecting the calculation of @xmath16 and @xmath17 are present here .
but for most of the stars observed , this formalism allows a more precise way to discriminate between field and cluster stars than looking to just @xmath16 or just @xmath17 .
stars that are still alive in globular clusters evolve along the cmd following a well determined path .
they spend most of their lives in the ms burning hydrogen in their core .
they leave the ms at the turn - off point ( to ) growing in size and burning hydrogen in a shell as they move across the sgb and the rgb .
then they move to the hb where they burn helium in the core and hydrogen in a shell .
they leave the hb moving across the asymptotic giant branch ( agb ) burning helium and hydrogen in a shell , then moving rapidly across the post agb phases to finish their lives as white dwarfs ( wd ) .
the path of this standard evolutionary sequence in the cmd of a globular cluster is shown in figure [ figridge ] .
we try to model the first stages ( ms , sgb and rgb ) of the evolutionary path followed by the stars in a gc building the ridgeline for the cmds of the ggcs , using _
locfit _ to construct a univariate non - parametric regression of the color of the stars as a function of the magnitude .
this process is composed of three iterations . in the first iteration
, we carry out a non - parametric regression in the whole cmd to obtain a first estimation of the cluster ridgeline as a function of magnitude .
we feed _ locfit _ with the magnitudes and colors of the stars and a smoothing parameter with a constant bandwidth of 0.2 magnitudes .
also , we give _ locfit _ preliminary weights for the stars @xmath76 where @xmath15 is the membership probability calculated in the previous subsection and @xmath77 is a photometric weight equal to the inverse of the square of the poisson error of the photometric magnitude @xmath78 .
these preliminary weights @xmath79 are calculated to favor the magnitude and color information provided by stars with more accurate photometry and higher probability of being cluster members according to both their position in the sky and in the cmd .
the smoothing parameter values used have proven experimentally to be adequate for describing the ms and part of the rgb , but the regression fails to describe the sgb , and the stars in the hb create problems when trying to model the whole rgb .
still , they produce a reasonable first approximation ( see upper left plot in figure [ figridge ] ) and let us identify the to point , which we define as the bluest point of our ridgeline .
we should note here that the ms is defined not to the faintest magnitude limit of the data , but to the completeness limit defined in section [ sectprob ] . in the second iteration
we try to get a less noisy whole ridgeline , along with a better defined rgb ridgeline . as a first step we divide the initial ridgeline in three regions : the ms region , where the ridgeline magnitudes are @xmath80 , the sgb region , where the ridgeline magnitudes are @xmath81 , and the rgb region , where the ridgeline magnitudes are @xmath82 . to smooth out the approximation for the ridgeline
, we take the points from the ridgeline from the first iteration for every region and smooth them using _ locfit _ with a nearest neighbor ratio of 0.7 ( _ locfit _ default ) , no preliminary weights or constant bandwidth requirements .
the rgb ridgeline is still not well defined , because of the effects of stars in the hb . to calculate a better ridgeline for the rgb getting rid of the effects of the blue part of the hb , we take only stars in the magnitude range @xmath82 and in the color range @xmath83 .
we feed _ locfit _ with their magnitudes and colors , and a constant bandwidth of 0.2 mag and a nearest neighbor ratio of 0.1 to calculate the smoothing parameter .
we use the nearest neighbor ratio because if not , the upper part of the rgb , usually scarcely populated , can become very noisy .
still , the red part of the hb , whenever present , is going to deviate the calculated rgb ridgeline . to eliminate its effect we have to locate first where the hb is .
we look for a change in the sign of the slope of the rgb .
the bluest point brighter than the point where the slope sign change is , shows where the hb is . and the difference between these two points provides information about the thickness of the hb , @xmath84 .
once the hb is located , we repeat the analysis done before for the rgb , with the same smoothing parameter and weights , but now omitting points in the interval @xmath85 . that way we have a better ridgeline for the upper and lower regions of the rgb . to get a ridgeline for the whole rgb ,
smoothing any noisy part , we follow a similar process to what we do in the beginning of this second iteration .
we take the points in the description of the upper and lower rgb ridgeline , and use _ locfit _ again , but now with a nearest neighbor ratio of 0.7 , no preliminary weights or constant bandwidth requirements .
the ridgeline is much better now , but still we are a little off in the description of the sgb ( see upper right plot in figure [ figridge ] ) . in the final iteration
, we try to calculate a better ridgeline for the sgb . in order to do that we follow @xcite , where they use rotated histograms to recalculate the ridgeline for the gc in their sample , using only stars perpendicular to the ridgeline in the calculation .
although our approximation is similar , the implementation is a little different .
we calculate the slope of the curve at every point of the previously calculated ridgeline using _ locfit _ to get the derivative of the slope at a given point of the ridgeline .
once we have this , we rotate the coordinate system of every star at every magnitude with the angle @xmath86 of the slope of the ridgeline at that magnitude , and centered at the ridgeline .
once this is done , we can get a ridgeline in the new coordinate system using _ locfit _ again .
this new ridgeline should be a straight line along the y coordinate with a value of 0 in the x coordinate .
any deviation from that means that the ridgeline in the original color - magnitude coordinate system requires a more accurate calculation .
notice that in order for the new coordinate system to consistently show these deviations , we need the range of both coordinates to be similar in the old coordinate system .
this is not the case for the range of colors and magnitudes presented in a cmd . to try to get them to a similar scale
we multiply the color by a factor of 5 before doing the rotation . in the calculation of the ridgeline in the new coordinate system the preliminary weights and smoothing factor used for _ locfit
_ are the same as in the first iteration .
we observe that the x coordinate of the ridgeline in the new reference system does not deviate significantly from 0 in the region of the ms stars , but does so in the sgb and upper rgb region . in the rgb region
it is expected due to the scarcity of the population there .
we took care of that in the previous iteration , so we concentrate our attention in the sgb region .
we derotate the ridgeline in that region ( the part where the original magnitudes are in the range @xmath87 ) to get their coordinates in the color - magnitude coordinate system , and after smoothing out any noise using _ locfit _ with a nearest neighbor ratio of 0.7 , we put together the different parts of the ridgeline ( see lower left plot in figure [ figridge ] ) .
the new ridgeline seems to follow accurately the different evolutionary sequences present in the gcs , and serves as the basis of our subsequent analysis . in section [ sectmap ]
we discuss how we can use stars in the hb to test the accuracy of our method . to carry this test , we need to model a ridgeline for this region too . the process to find
the ridgeline here is a little more interactive than for the other regions of the cmd .
first , we have to decide by visual inspection of the cmd if we are dealing with a cluster that has only blue , only red or both sections of the hb .
if the cluster shows only a blue hb , we carry out a non - parametric regression on the stars bluer than the to to obtain an estimation of the cluster hb ridgeline as a function of magnitude .
we feed _ locfit _ with the magnitudes and colors of the hb , and a constant bandwidth of 0.2 mag and a nearest neighbor ratio of 0.1 to calculate the smoothing parameter .
if only a red hb is present instead , the non - parametric regression is performed on stars in the interval @xmath85 to obtain an estimation of the cluster hb ridgeline as a function of color .
the smoothing parameter is calculated by _ locfit _ with the same parameters as for the blue hb . finally ,
if the hb shows red and blue sections , we calculate both independently following the methods previously described , and then we smooth the result as a function of color , feeding _ locfit _ with the magnitudes and colors of the points in the description of the blue and red hb ridgeline , and a nearest neighbor ratio of 0.25 , no preliminary weights or constant bandwidth requirements ( see lower right plot in figure [ figridge ] ) . to calculate an extinction for every star we move the stars along the reddening vector until they intersect the ridgeline at an astrophysically reasonable location ( see figure [ figredvector ] ) .
the reddening vectors are described by the equations @xmath88 and @xmath89 as given in @xcite , which are evaluated using the @xmath90 extinction laws of @xcite and @xcite .
the hb ridgeline , and therefore also stars in the range @xmath91 , are not used in this calculation , since we want to use them as an independent test of the accuracy of our method ( see section [ sectmap ] ) .
we do not use stars dimmer than the completeness limit defined in section [ sectprob ] either .
these stars and stars that do not intersect with the ridgeline are given a weight 0 in the calculation of the extinction map described in next section .
to calculate the error of these shifts we follow the analysis in @xcite : * we created an error ellipse for every star defined by the poisson error in its color , @xmath92 , and magnitude , @xmath93 . *
since the color and magnitude errors are correlated , the error ellipse is tilted .
the tilt angle and the length of the semi - major and semi - minor axes of the error ellipse are functions of the error in color and in magnitude .
@xmath94 @xmath95^{1/2}\ ] ] @xmath96^{1/2}\ ] ] where @xmath97 is the tilt angle of the error ellipse , @xmath98 and @xmath99 are the semi - major and semi - minor axis of the tilted ellipse , and @xmath100 is the covariance of the color and magnitude .
* for every star , we move the error ellipse along the reddening vector .
the point of first contact ( pfc ) , i.e. , the point where the ellipse touches the ridgeline for the first time , and the point of last contact ( plc ) , where the ellipse touches the ridgeline for the last time , represent the one - sigma - deviation points ( see figure [ figredvector ] ) . * although these two contact points are not necessarily symmetric about the reddening value of the star , i.e. , the center of the error ellipse , we are going to define the error in the shift as : @xmath101 we now take all the color excesses for the individual stars and smooth them using _ locfit _ to build a bivariate non - parametric regression of the extinction as a function of the spatial coordinates right ascension and declination , up to a distance from the center of the cluster equal to where @xmath102 or the limit of our observations , whichever comes first ( see table [ tabclimit ] ) .
this limit is adopted after preliminary application of our method to the clusters in our sample with looser restrictions , and not observing an improvement in the dereddened cmds ( see section [ sectcmd ] ) for stars located beyond these ranges .
stars that were given zero weight in the previous subsections ( stars in the hb region , stars dimmer than the completeness limit , stars that do not intersect with the ridgeline after being moved along the reddening vector ) are not taken into account for the calculation of the map , although after the map is built a reddening correction is applied to all of them .
preliminary weights are assigned in _ locfit _ to all stars as they were calculated in previous subsections , but now @xmath103 .
the spatial smoothing kernel in this case has a constant bandwidth of @xmath104 or a nearest neighbor ratio of 0.03 , whichever is bigger .
these values for the smoothing parameter are used after experimentally checking in a few of our clusters which kernel provides us with a tighter hb , in the sense of a smaller standard deviation of the stars from the ridgeline of this region ( see section [ sectridge ] ) , after generating the dereddened cmd ( see section [ sectcmd ] ) .
this is an independent test to assess the quality of our dereddening method , since our technique does not use stars in the hb to calculate the extinction map .
the tests gave similar results for a range of parameters .
since the tests were not highly conclusive on choosing a particular set of parameters , the set used is an average of the best values obtained for the different clusters tested . along with the extinction map ,
an extinction precision map is also calculated .
this is done assuming that the precision of the extinction map is a function of the spatial coordinates .
hence , we can write @xmath105 where @xmath106 is the calculated extinction for the @xmath107-th star and @xmath108 is the value for the extinction provided by the regression function at the coordinates @xmath109 of the @xmath107-th star . therefore
, regressing directly the square residuals @xmath110 provides us with an estimate @xmath111 of the variance , and its square root allows us to plot a map of the standard deviation @xmath112 as a function of the spatial coordinates .
this precision map provides us with a tool to improve our extinction map .
we iteratively repeat the whole process of the calculation of extinction maps , but every time using only stars that have extinctions that are no more than @xmath113 away from the value that our previous extinction map gives for those coordinates .
preliminary weights and smoothing parameters are the same in every calculation .
we repeat the process until it converges , i.e. , until the number of stars used in the calculation of the extinction maps changes by less than @xmath114 with respect to the previous iteration . this way for every cluster we are able to provide an extinction map ( see top plot of figure [ figextinc ] ) , a precision map for this extinction map ( see bottom plot of figure [ figextinc ] ) , and a resolution map with the bandwidth that we have use in a certain region to calculate the extinction map there ( see central plot of figure [ figextinc ] ) .
all the observed stars are given a reddening correction from the extinction map and a dereddened cmd is constructed ( see figure [ figcmd ] ) .
this dereddened photometry is the input on the next iteration to calculate the cluster membership probabilities , and to build a new improved ridgeline ( see figure [ figflowchart ] ) .
notice though that the input on the calculation of the individual reddenings to generate the extinction maps is the original photometry in every iteration ( see figure [ figflowchart ] ) .
we iteratively repeat this process until there is a convergence in the calculated ridgeline ( see figure [ figflowchart ] ) , which results in good convergence of the reddening values of the extinction map . for our dataset
, convergence usually occurs after just two or three iterations .
the method will be extensively applied to a sample of 25 globular clusters in following papers . here , as an example , we applied the method to one of those clusters , m62 .
m 62 ( ngc 6266 ) is a moderately reddened , @xmath115 cluster , located only @xmath116 from the galactic center @xcite , with a patchy extinction @xcite .
the observation , reduction and calibration processes will be explain in detail in paper ii .
suffice it to say here that @xmath117 , @xmath65 , and @xmath118 data for this cluster where obtained with the imacs camera in the magellan 6.5 m telescope , and , in @xmath65 and @xmath118 for the inner part of the cluster with the wfpc2 camera on the hst .
after applying fairly standard reduction and calibration processes to these data , we merged the databases from the two telescopes to obtain accurate astrometry ( @xmath119 dispersion ) and accurate photometry ( @xmath120 magnitude dispersion ) in b , v and i for the stars in a field of @xmath121 centered in the cluster . in figure [ figcmd ]
we can see the resulting cmds of this cluster .
the first step of our dereddening method is to calculate the probabilities @xmath15 for the observed stars to belong to the cluster . to do that
, we use the probability densities of the stars as functions of distance to the center of the cluster and of color and magnitude , a king profile plus constant field model , and a besanon model for the field , as explained in section [ sectprob ] ( see figures [ figprobdis ] , [ figprobdcm ] , and [ figprob ] ) .
the different parameters used to build the king and besanon models for this calculation are shown in tables [ tabking ] and [ tabbesancon ] .
notice that we restrict our analysis to stars in the limits shown in table [ tabclimit ] , because of the reasons mentioned in the previous sections .
once the probabilities to belong to the cluster have been calculated , we need to build the ridgeline .
following the the three step recipe explained in section [ sectridge ] , we get the ridgelines in the two available colors ( see figure [ figridge ] ) . after that , we move the stars along the reddening vector until they intersect the ridgeline ( see figure [ figredvector ] ) and smooth the resulting individual color excesses , as explained in sections [ sectext ] and [ sectmap ] .
we obtain the extinction map shown in figure [ figextinc ] . from this map ,
we take the relative extinction that corresponds to every star observed .
we then plot the cmd of the stars in our observation after having been corrected for the differential reddening and use it as the input for the next iteration .
the process converges for both colors after 3 iterations and we generate the dereddened cmds that we can see in figure [ figcmd ] .
we leave the detailed examination and analysis of the physical characteristics of the cluster to paper iii , but we mention here how much better the definition of the different parts of the cluster is .
the width of the ms narrows by a factor of 2 after being dereddened ( see figure [ figgauss ] ) .
this improvement will help us to obtain better constraints on the cluster age and distance when compared to theoretical isochrone models .
the rgb is also narrowed by a factor of 2 ( see figure [ figgauss ] ) , which will help when trying to better constrain the metallicities . as a reference
, we compare the final reddening map obtained using our technique with maps and extinction values within the covered region available in the literature ( see figure [ figextincomp ] ) .
we found a general agreement in the identification of regions of high and low extinction , although our map shows sharper features than the one presented by @xcite , perhaps the result of their mapping the interstellar extinction in the foreground and also in the background of the cluster ( upper plots in figure [ figextincomp ] ) , and smoother values than those by @xcite , result of our spatial smoothing of the extinction values ( lower plots in figure [ figextincomp ] ) .
as we have already mentioned , the technique explained in the paper does not calculate the absolute extinction toward a target cluster . comparisons like the ones shown in figure [ figextincomp ] will allow to find the extinction of the adopted reddening zero point in our dereddening method .
this work was supported by grants 0206081 from nsf and go10573.01-a from stsci .
stsci is operated by aura under nasa contract nas5 - 26555 .
support for j.a .
was also provided by mideplan s programa inicativa cientfica milenio through grant p07 - 021-f , awarded to the milky way millennium nucleus .
in our work we have extensively used non - parametric statistics to estimate probability density and regression functions directly , without reference to a specific form .
this differs from other approaches in which probability density and regression functions are expressed in a parametric way , where the function used to describe them can be written as a mathematical formula which is fully described by a finite set of parameters that we have to find . in our analysis
we have used _ locfit _ , a local likelihood estimation software implemented in the r statistical programming language . _ locfit _ is extensively explained in @xcite , so here we just want to give a general idea of how it works and the main tuning parameters that we have chosen in our analysis .
_ locfit _ does not constrain the functions globally , i.e. , it is non - parametric , but assumes that locally , around a certain point x , the function can be well approximated by a member of a simple class of parametric functions .
_ locfit _ defines a local window around a point , weighting the observations according to their distance to that point @xmath122 and inside this local window , the function is approximated by a polynomial , using not the usual local least square criterion but a local log likelihood criterion .
we have to choose the weighting function and the order of the polynomial .
for both cases we take the default given by the program : a polynomial of order 2 with a tricube weight function @xmath123 .
still we are left with one last argument to choose , the smoothing parameter that controls the bandwidth @xmath124 .
this smoothing parameter is defined by the maximum of two elements : a bandwidth generated by a nearest neighbor fraction @xmath125 , and a constant bandwidth .
the nearest neighbor bandwidth is computed in two steps , first computing the distances @xmath126 for all the data points and then choosing @xmath124 to be the _
k_th smallest distance , where @xmath127 $ ] .
also the likelihood criterion can be chosen , and we choose for the regression the family _ qrgauss _ , which is equivalent to a local robust least squares criterion where outliers are iteratively identified and downweighted , similar to the _ lowess _ method @xcite .
a scale factor can be applied to the different variables in multivariate fitting , when variables are measured in non - comparable units , or when we want to give more importance to one of them like in the determination of the densities of stars in our cmd , where the range in color for the stars is smaller than the range in magnitudes .
finally , a word is in order about the evaluation structures in _ locfit_. _
locfit _ does not perform local regression directly in every point , but selects a set of evaluation points obtaining the fit there and interpolating later elsewhere .
this is done for efficiency : it is much faster to evaluate the structure in a small number of points and then interpolate for the rest .
but we should make sure that we do not lose information on the process . in order to achieve this , _ locfit _ uses , by default , a growing adaptive tree , which is the evaluation structure that we use in our analysis .
a growing adaptive tree is a grid of points .
one begins by bounding the data in a rectangular box and evaluating the fit at the vertices of the box .
one then recursively splits the box into two pieces , then each subbox into two pieces and so on . for this kind of structure ,
an edge is always split at the midpoint and the decision to split an edge is based solely on the bandwidths at the two ends of the edge , depending on the score @xmath128 where @xmath129 is the distance between the two vertices of an edge , and @xmath130 and @xmath131 are the bandwidths used at the vertices .
any edge whose score exceeds a critical value c ( @xmath132 by default ) is split .
bedin , l. r. , piotto , g. , anderson , j. , cassisi , s. , king , i. r. , momany , y. , & carraro , g. 2004 , 605 , l125 calamida , a. , et al .
2005 , , 634 , 69 cardelli , j. a. , clayton , g. c. & mathis , j. s. 1989 , , 345 , 245 cleveland , w. s. 1979 , j. amer . statist .
, 74 , 829 contreras , r. , catelan , m. , smith , h. a. , pritzl , b. j. , & borissova , j. 2005 , , 623 , l117 contreras , r. , catelan , m. , smith , h. a. , pritzl , b. j. , borissova , j. & kuehn , c. a. 2010 , , 140 , 1766 dantona , f. , & caloi , v. 2008 , , 390 , 693 grillmair , c. j. , freeman , k. c. , irwin , m. , & quinn , p. j. 1995 , , 109 , 2553 harris , w. e. 1996 , , 112 , 1487 heitsch , f. , & richtler , t. 1999 , , 347 , 455 hughes , j. , & wallerstein , g. 2000 , , 119 , 1225 hughes , j. , wallerstein , g. , covarrubias , r. , & hays , n. 2007 , , 134 , 229 kaluzny , j. , & krzeminski , w. 1993 , , 264 , 785 king , i. 1962 , , 67 , 471 law , d. r. , majewski , s. r. , skrutskie , m. f. , carpenter , j. m. , & ayub , h. f. 2003 , , 126 , 1871 loader , c. 1999 , local regression and likelihood , ( new york : springer ) marn - franch , a. , et al .
2009 , , 694 , 1498 melbourne , j. & guhathakurta , p. 2004
, , 128 , 271 mighell , k. j. , sarajedini , a. , & french , r. s. 1998 , , 116,2395 odenkirchen , m. , et al .
2001 , , 548 , 165 odonnell , j. e. 1994 , , 422 , 1580 piersimoni , a. m. , bono , g. , & ripepi , v. 2002 , , 124 , 1528 piotto , g. , zoccali , m. , king , i. r. , djorgovski , s. g. , sosin , c. , dorman , b. , rich , r. m. & meylan , g. 1999 , , 117 , 264 piotto , g. , zoccali , m. , king , i. r. , djorgovski , s. g. , sosin , c. , rich , r. m. & meylan , g. 1999 , , 118 , 1727 piotto , g. , bedin , l. r. , anderson , j. , king , i. r. , cassisi , s. , milone , a. p. , villanova , s. , pietrinferni , a. , & renzini , a. 2007 , , 661 , l53 robin , a. c. , reyl , c. , derrire , s. , & picaud , s. 2003 , , 409 , 523 schechter , p. , mateo , m. , & saha , a. 1993 , , 105 , 1342 schlegel , d. , finkbeiner , d. , & davis , m. 1998 , , 500 , 525 valenti , e. , ferraro , f. r. , & origlia , l. 2007 , , 133 , 1287 von braun , k. & mateo , m. 2001 , 121 , 1522 cc core radius ( according to most updated harris catalog)&0.18 arcmin + tidal radius ( according to most updated harris catalog)&8.97 arcmin + @xmath38 constant from the king profile ( see @xmath41)&12264.7 stars arcmin@xmath133 + surface density of non - member field stars from the fit ( see @xmath41)&226.4 stars arcmin@xmath133 + surface density of non - member field stars according to besanon model&216.6 stars arcmin@xmath133 + cc field&small field + & l=@xmath134 ; b=@xmath135 + & solid angle=0.190 square degree + diffuse extinction & 0.0 mag / kpc + discrete clouds & @xmath136=1.36 ; distance=0pc + intervals of magnitude & @xmath137 + & @xmath138 + & @xmath139 + photometric errors & error function : exponential + & band=@xmath117 ; a=0.006 , b=22.68 , c=0.866 + & band=@xmath65 ; a=0.005 , b=23.22 , c=0.899 + & band=@xmath118 ; a=0.009 , b=30.12 , c=1.238 + @xmath65 magnitude where the completeness limit is reached&21.51 + @xmath140 between the completeness limit and the dimmest star in our ground observation&2.2 + distance where the ratio of gc stars to total number of stars drops to 0.1&6.20 arcmin + | a serious limitation in the study of many globular clusters especially those located near the galactic center has been the existence of large and differential extinction by foreground dust . in a series of papers we intend to map the differential extinction and
remove its effects , using a new dereddening technique , in a sample of clusters in the direction of the inner galaxy , observed using the magellan 6.5 m telescope and the _ hubble space telescope_. these observations and their analysis will let us produce high quality color - magnitude diagrams of these poorly studied clusters that will allow us to determine these clusters relative ages , distances and chemistry and to address important questions about the formation and the evolution of the inner galaxy .
we also intend to use the maps of the differential extinction to sample and characterize the interstellar medium along the numerous low latitude lines of sight where the clusters in our sample lie . in this first paper
we describe in detail our dereddening method along with the powerful statistics tools that allow us to apply it , and we show the kind of results that we can expect , applying the method to m62 , one of the clusters in our sample .
the width of the main sequence and lower red giant branch narrows by a factor of 2 after applying our dereddening technique , which will significantly help to constrain the age , distance , and metallicity of the cluster . |
the flows through porous medium are very much prevalent in nature and therefore , the study of such flows has become of principal interest in many scientific and engineering applications @xcite .
this type of flows has shown their great importance in petroleum engineering to study the movements of natural gas , oil and water through the oil reservoirs ; in chemical engineering for the filtration and water purification processes .
further , to study the underground water resources and seepage of water in river beds one need the knowledge of the fluid flow through porous medium
. therefore , there are number of practical uses of the fluid flow through porous media .
the porous medium is in fact a non - homogeneous medium but for the sake of analysis , it may be possible to replace it with a homogeneous fluid which has dynamical properties equal to those of non - homogeneous continuum . thus one can study the flow of hypothetical fluid under the action of the properly averaged external flow and the complicated problem of the flow through a porous medium reduces to the flow problem of homogeneous fluid with some additional resistance@xcite . in this work , we considered incompressible viscous fluid flow in porous medium between two porous plates .
we assumed that the plates are parallel , horizontal and stationary .
we applied injection at the lower plate and suction at upper see ( hinvi et al.@xcite ) .
the objective is to study the effect of reynolds number @xmath0 and permeability parameter @xmath1 introduced in navier - stokes equations by the small suction velocity and permeability of medium , on the stability of the poiseuille flow in porous medium as in ( hinvi et al.(@xciteand @xcite ) ) . for this as in previous paper , we derive fourth - order equations named modified fourth - order orr - sommerfeld equations governing the stability analysis of poiseuille flow in porous medium .
this allowed us to see the simultanous effect of @xmath2 or @xmath0 and medium permeability parameter @xmath1 on the stability in the boundary layer .
thus , we solve numerically the corresponding eigenvalues problems .
we employ matlab in all our numerical computations to find eigenvalues .
such attempt has been made earlier by monwanou et al .
@xcite the case of the non - porous plate boundary layers ( a flat plate - law ) without wall suction by normalizing all the components of the velocity with the free stream velocity @xmath3 .
the paper is organized as follows . in the second section the mathematical formulation of the problem
is made . in the third section the modified orr - sommerfeld equation governing the stability analysis of poiseuille flow in porous medium
is deduced . in the fourth section
, the effects of small injection / suction parameter @xmath0 , wave number @xmath4 and permeability parameter @xmath5 on linear stability of poiseuille flow in porous medium will be investigated with the help of figures .
the conclusions will be presented in the final section .
we considered a poiseuille viscous incompressible fluid flow between a porous me - dium .
we assumed the medium isotrope see figure [ fig:1 ] .
the two porous parallel plates delimiters are assumed of infinite lengh , distant @xmath6 .
the @xmath7 axis is normal to the planes of the plates .
we considered the simple case where , the permeability parameter @xmath1 and velocities suction and injection are constants .
we work at constant temperature , the heat transfer aspect of the flow is not studied .
the small constant injection @xmath8 is applied at the lower plate and a same small constant suction @xmath8 , at the upper plate .
the two plates are fixed .
we choose the origine on the plane @xmath9 such as @xmath10 and @xmath11 parallel to the direction of the mean fluid flow .
initially , @xmath12 , the fluid is assumed to be at rest . when @xmath13 , the fluid starts moving .
the porous medium is infact a non - homogeneous medium but for the sake of analysis , it may be possible to replace it with a homogeneous continuum .
thus , one can study the flow of hypothetical homogeneous fluid under the action of the properly averaged external forces .
thus , a complicated problem of the flow through a porous medium reduces to the flow problem of a homogeneous fluid with some resistance .
then the equations of fluid flow in porous medium are simply the equations of fluid flow in homogeneous medium see ( [ fig:2 ] ) added the permeability term .
extra force terms as compared to non - porous medium fluid are added on @xmath14 of navie - stokes equations for homogeneous medium , due to the permeability medium .
we can treat porous medium fluid as a single fluid under assumption that there is a permeability term to added .
the equations of continuity , motion for the viscous incompressible fluid in porous medium are @xmath15 we introduced the following non - dimensional quantities as in ( hinvi et al.(@xcite ) @xmath16 @xmath17 @xmath18 @xmath19 then the eqs.([1 ] ) and ( [ 2 ] ) become in cartesien cordinates : @xmath20 @xmath21 @xmath22 @xmath23
the flow is decomposed into the mean flow and the deviation from it ( the disturbance ) according to @xmath24 @xmath25 where @xmath26 is the spatial coordinate vector . inserting ( [ a1 ] ) and ( [ a2 ] ) into the navier - stokes equations ( [ 3])- ( [ 6 ] ) we obtain the equations following .
@xmath27 @xmath28 @xmath29 @xmath30 we take the dimensional base flow for small suction and injection see ( hinvi et al.@xcite ) : @xmath31 to obtain the stability equations for the spatial evolution of three - dimensional , we take the dependent on time disturbances @xmath32 which are scaled in the same way as above . with the equations ( [ a6]-[a9 ] ) ,
the equations ( [ 7]-[10 ] ) become @xmath33 @xmath34 @xmath35 @xmath36 the equations ( [ 11]-[14 ] ) above are the linear disturbance equations that model the poi - seuille flow of the incompressible viscous fluid in porous medium with the assumption that there is a constant small suction at upper plate and a constant small injection at the lower plate in cartesien coordinates .
in this section the modified orr - sommerfeld equation governing the stability analysis of poiseuille flow in porous medium is deduced by using navier - stokes equations with the same strategy as ( see hinvi et al.@xcite ) . by using continuity equation ,
the pressure terms can be eliminated from ( [ 11])-([14 ] ) , by reducing the system to two equations for two unknown quantities . for a mean profile ( [ a6]-[a8 ] ) ,
the divergence of ( [ 11])-([14 ] ) and continuity gives @xmath37 we applied @xmath38 to the equation ( [ 12 ] ) , unsing ( [ 15 ] ) we found @xmath39 by neglecting the quadratic derivation ( because of linear stability analysis ) the equation ( [ 16 ] ) becomes @xmath40 we transform the eq.([16 ] ) in an eigenvalue problem .
thus , the disturbances are taken to be periodic in time in the streamwise , spanwise directions , which allow us to assume solutions of the form @xmath41 where @xmath42 represents either one of the disturbances @xmath43 , @xmath44 , @xmath45 or @xmath46 and @xmath47 the amplitude function , @xmath4 , @xmath48 and @xmath49 are the wave numbers , @xmath50 is the frequency of the wave . with @xmath51 , @xmath52 , @xmath53 wave velocity which is taken to be complex , @xmath54 and @xmath55 are real because of temporal stability analysis considered . then with the equation ( [ 18 ] ) the equation ( [ 17 ] ) becomes @xmath56 where @xmath57 .
we have a same boundary conditions @xmath58 as in ( hinvi et al.@xcite ) .
to transform these boundary conditions to @xmath59 we make the change , by puting @xmath60 the equation ( [ 19 ] ) becomes then @xmath61\hat{v}_{p } & = & \nonumber\\ c \left ( d^{2}-k^{2}\right)\hat{v}_{p}\label{21 } \end{aligned}\ ] ] with boundary conditions for all @xmath62 : @xmath63 the eq.([21 ] ) is a flow equation modified by suction reynolds number @xmath64 ( or the speed of suction and injection ) and permeability parameter @xmath65 , which differentiated it , from the model study by ( hinvi et al.@xcite ) by permeability term , rewritten as an eigenvalue problem , where @xmath66 is the eigenvalue and @xmath67 the eigenfunction .
@xmath68 and @xmath69 are the operators .
we consider three - dimensional disturbances .
we use a temporal stability analysis as mentioned above . with @xmath66 complex as we have defined above , when @xmath70 we have stability , @xmath71 we have neutral stability and elsewhere we have instability .
we employ matlab @xmath72 in all our numerical computations to find the eigenvalues .
the mean flow profile for poiseuille flow in porous medium is @xmath73 for @xmath0 small ( i.e. small suction ) is considered ( see @xcite ) the eigenvalue problem ( [ 21])-[22 ] ) is solved numerically with the suitable boundary conditions .
the solutions are found in a layer bounded at @xmath74 .
the results of calculations are presented in the figures below .
we present the figures related to the eigenvalue problem . for all figure we ploted @xmath75 by fixing the other parameters .
the yellow color is for @xmath71 . vs. reynolds number for @xmath76 , @xmath77 , @xmath78 , @xmath79 and @xmath80.,width=453,height=302 ] in the figure [ fig : s01 ]
, we note for all small values of permeability parameter that , @xmath81 .
we conclude that the flow is unstable anywhere .
these curves also tell us that an increase in the permeability parameter promotes flow stability . vs. reynolds number @xmath82 , for @xmath76 , @xmath83 , @xmath84 , @xmath85 .
the right frame is zoom of the left frame.,width=453,height=302 ] through the curves of the figure [ fig : s1zo ] , we remark that for @xmath86 , large values of @xmath1 stabilize the flow and for the small value @xmath87 the flow is unstable .
when @xmath88 for the all values of @xmath1 the flow is unstable and after ( @xmath89 ) it becomes stable elsewhere .
the flow stabilizes for small values of @xmath90 , becomes unstable and stabilizes after for @xmath91 i.e we have two transitions of the fluid flow .
we conclude that @xmath1 s increasing stabilizes the poiseuille fluid flow in porous medium . vs. reynolds number for @xmath76 , @xmath77 , @xmath78 , @xmath92 and @xmath80.,width=453,height=302 ]
this figure [ fig : l1 ] shows the behavior of the curve @xmath93 vs. reynolds number @xmath82 for @xmath76 , @xmath77 , @xmath94 , when @xmath1 increases considerably ( strong values of @xmath1 ) . with large values of @xmath1 ( @xmath95 )
the curves have the same behavior and the limit critical reynolds numbers are @xmath96 and @xmath97 respectively for the first and second transitions . in the equation ( [ 21 ] ) , we remark that , when @xmath98 the permeability term @xmath99 vanishes and the model becomes which has been studed in ( hinvi et al.@xcite ) where the first critical reynolds number is @xmath100 . vs. injection / suction reynolds number @xmath0 for @xmath76 , @xmath101 , @xmath84 .
the right frame ( @xmath87 ) and the left frame ( @xmath102).,width=453,height=302 ] the figure [ fig : s2w ] show growth rate @xmath93 vs. injection / suction reynolds number @xmath0 for @xmath76 , @xmath101 , @xmath84 , @xmath85 .
the both curves are downward , in the left frame we note @xmath103 i.e the flow is unstable . in the right frame
@xmath104 then the flow is completely stable for this value of @xmath102 .
we conclude that the suction / injection reynold s number affects the flow stability and it kept stable for large permeability parameter value . vs. wave number @xmath4 for @xmath105 , @xmath84 , @xmath106 , @xmath107 and @xmath108.,width=453,height=302 ] the curves in figure [ fig : s03k ]
show the appearance of @xmath109 as a function of wavenumber @xmath4 , for two different values of @xmath1 .
we note that for high values of the permeability parameter , the flow is completely stable for any order of @xmath4 .
thus for small @xmath1 s values of the order @xmath80 , the flow remains unstable at the beginning ie for @xmath110 and stabilizes after .
we also note that after @xmath111 the two curves merge .
we conclude that there is no special value for @xmath1 to induce any unstability . vs. @xmath112 for @xmath76 , @xmath101 , @xmath108 .
@xmath87 for the left frame and @xmath102 for the right frame.,width=453,height=302 ] through figure [ fig : s4 t ] we have @xmath93 vs. @xmath112 .
we note that for @xmath87 , @xmath103 and the curve is ascendent for all value of @xmath113 and for @xmath114 the curve is downward but we note unstability in beginning and stability at the end .
in this paper , the study of the effect of the reynolds number @xmath0 and particular the permeability parameter @xmath1 introduced in orr - sommerfeld s equation by the small suction velocity and permeability of medium , on the stability of the poiseuille flow in porous medium has been made .
the modified orr - sommerfeld equation modeling poiseuille flow in porous medium is deduced .
the growth rates @xmath93 vs. the parameters of system have been plotted .
the previous results analysis allowed us to conclude that all parameters of the system affect the poiseuille flow stability in porous medium and have a stabilizing effect on the flow .
the authors thank imsp - uac and benin gorvernment for financial support .
chand k. , kumar r. and sharma s.(@xmath115 ) . _ hydromagnetic oscillatory flow through a porous medium bounded by two vertical porous plates with heat source and soret effect _
( advances in applied science research , @xmath115 , @xmath116 ) .
l. a. hinvi , a. v. monwanou and j. b. chabi orou .
_ linear stability analysis of hydrodynamic couette flow in porous medium_. sumetted in ( annals of faculty engineering hunedoara ) international journal engineering .
a. v. monwanou , c. h. miwadinou and jean b. chabi orou,_stability analysisof boundary layer in poiseuille flow through a modifed orr - sommerfeld equation _ , applied physics research , vol@xmath118 , n@xmath119,issn @xmath120 , e - issn @xmath121 , pp.@xmath122 , doi:@xmath123/apr.v4n@xmath124 .
akkari a. ( 2012 - 2013 ) _ mcanique des fluides _ ( notes )
h. e. huntley , _ dimensional analysis _ , rinehart , new york , 1951 .
r. esnault - pelterie , _
l analyse dimensionelle _ , f. rouge , lausanne , switzerland , 1946 .
robert a. greenkarm ( book ) _ flow phenomena in porous media _ isbn 8 - 8247 - 1862 - 5 .
kharagpur _ the science of surface and ground water ; principles of ground water flow_. | we investigate the effect of small suction reynolds number and permeability parameter on the stability of poiseuille fluid flow in a porous medium between two parallel horizontal stationary porous plates .
we have shown that the perturbed flow is governed by an equation named modified orr - sommerfeld equation .
we find also that the normalization of the wall - normal velocity with characteristic small suction ( or small injection ) velocity is important for a perfect command of porous medium fluid flow stability analysis .
the stabilizing effect of the parameters in general and small suction reynolds number and permeability parameters in particular on the linear stability are found . |
we consider that it is impossible to solve the singularity issue of space - time and the issue of the cosmological constant in the frame of the conventional theory , so that new conjectures are necessary . on the basis of a new essential conjecture
, we can solve the two issues , well explain evolution of the universe and the characters of huge voids , and give new predictions and some guesses .
as is now well known , there is space - time singularity under certain conditions@xmath1}$ ] .
these conditions fall into three categories .
first , there is the requirement that gravity shall be attractive .
secondly , there is the requirement that there is enough matter present in some region to prevent anything escaping from that region .
the third requirement is that there should be no causality violations .
because of the theorems , there must be space - time singularity in the conventional theory .
on the other hand , there should be no space - time singularity in physics .
hence this problem must be solved .
recent astronomical observations show that the universe expanded with a deceleration earlier while is expanding with an acceleration now .
this implies that there is dark energy .
@xmath2 of the total energy density of the universe is dark energy density@xmath3}.$]what is dark energy ?
many possible answers have been given .
one possible interpretation is in terms of the effective cosmological constant @xmath4 @xmath5 here @xmath6 and @xmath7 are respectively the einstein s cosmological constant and the gravitational mass density of the vacuum state .
according to the equivalent principle , @xmath8 , @xmath9 is the mass density of the vacuum state , hence @xmath10 may written as @xmath11 @xmath12 can not be derived from basic theories@xmath13}$ ] and @xmath14 .
hence the interpretation is unsatisfactory . alternatively ,
dark energy is associated with the dynamics of scalar field @xmath15 that is uniform in space@xmath16}$ ] .
this is a seesaw cosmology@xmath17}$ ] .
thus , discussion about the universe expansion with an acceleration is still open to the public .
@xmath18 originates from the conventional quantum field theory and the equivalent principle .
@xmath19 and the singularity issue imply that the conventional theory is not self - consistent .
@xmath20 is a necessary result of our quantum field theory without divergence@xmath21}.$ ] in this theory , there is no divergence of loop corrections , and dumpling dark matter is predicted@xmath22}.$ ] it is different from the supersymmetric quantum field theory in which @xmath20 can be obtained in only some models but is not necessary .
thus , issue of the cosmological constant is open as well .
huge voids in the cosmos have been observed@xmath23}.$ ] the model in which the hot dark matter ( e.g. neutrinos ) is major can explain the phenomenon , however , it can not explain the structure with middle and small scales . hence this is an open problem as well .
we consider that all important existing forms of matter , dark matter and dark energy , to be presented .
hence the basic problems , e.g. divergence problem in quantum field theory , cosmological constant and space - time singularity problems in cosmology , should be solved based on the important existing forms of matter .
we have constructed a quantum field theory without divergence by dark matter .
we construct a model of the universe without singularity in the paper by dark energy .
we consider the following conditions 1 and 2 to be necessary in order to solve the singularity and the cosmological constant problems in the basis of the classical cosmology and the frame of the conventional quantum field theory .
the third condition is necessary to cause the space inflation .
condition 1 : : there are two sorts of matter which are symmetric , whose gravitational masses are opposite to each other and whose energies are all positive .
the two sorts of matter are called @xmath24 and @xmath25 , respectively .
the condition implies that if @xmath26 @xmath27 condition 2 : : when temperature is high enough , the thermal equilibrium between the two sorts of matter can come into being , @xmath28 and @xmath29 can transform from one into another so that @xmath30 .
condition 3 : : @xmath31 and @xmath32when @xmath33 here @xmath34 is the critical temperature . the conditions 1 and 2 can not be realized in the conventional theory . in order to uniformly solve the above four problems on one basis , we present new conjectures ( see section 2 ) and construct a cosmological model on the basis of the conjectures@xmath35}$ ] . the basic idea of the present model is the conjecture 1 , which realizes the conditions consequently there is no singularity of space - time and @xmath36 is proven , although @xmath9 is still very large .
thus , there is no the fine tuning problem , even if @xmath37 .
there are two sorts of breaking modes which are called @xmath38 and @xmath39 due to the conjecture 1 .
@xmath40 @xmath41 or @xmath42 are all possible due to the conjecture 1 . hence the curvature factor @xmath43 in the @xmath44 metric should be changeable .
we consider @xmath43 to be a function of @xmath45 @xmath46 is the gravitational mass density in comoving coordinates , i.e. , @xmath47 .
the evolving equations corresponding to @xmath48 have been derived@xmath35}$ ] .
an important basis of the present model is the temperature effect on expectation values of higgs fields .
the present model have the following results . according to the present model ,
the evolving process of space is as follows . in the @xmath49 space contracts @xmath50temperature @xmath51 rises to the highest temperature @xmath52 and the highest symmetry comes into being @xmath50 space inflation @xmath50 reheating process @xmath53 the state with the highest symmetry transits to the state with the @xmath39 and space expands with a deceleration
@xmath50 then comes to static @xmath50 finally expands with an acceleration up to now .
@xmath54 can change from @xmath55 to @xmath56 .
there are the critical temperature @xmath34 , the highest temperature @xmath57 , the least scale @xmath58 and the largest energy density @xmath59 in the universe . @xmath58 and @xmath34 are two new important constants , @xmath52 and @xmath60 are determined by @xmath61 .
a formula comes out which well describes the relation between a luminosity distance and its redshift .
the new predictions of the present model are as follows .
@xmath62 voids in the @xmath39 are not empty , but there is @xmath24 with its density @xmath63 in them .
their effects are similar to those of huge concave lenses for @xmath64 . in huge voids , both matter and dark matter are even shorter in the huge void .
thus the density of hydrogen in the huge voids must be more less that that predicted by the conventional theory .
right or mistake of the predict can be confirmed by the observation of distribution of hydrogen .
when the distance between two @xmath65 is very large , the gravitation between both will be less than that predicted by the conventional theory , because there must be @xmath24 between both .
we generalize equations governing nonrelativistic fluid motion to present model .
the equations of structure formation have been derived . according to the equations
, galaxies can form earlier and easier than that in the conventional theory . on the basis of this model ,
we have three guesses . the universe is composed of infinite @xmath66 islands and @xmath67 islands .
some huge redshifts ( e.g. the big redshifts of quasi - stellar objects ) are explained as the mass redshifts which is caused by less mass @xmath68 of an electron than given @xmath69 .
it is possible that a @xmath70 hole with its big enough mass and density can transform into a huge white hole by its self - gravitation .
of course , the effects of quantum mechanics , e.g. hawking radiation , must be considered .
but there is no contradiction between the transformation and quantum mechanics .
it seems that there are some difficulties in the present model , e.g. , it seems that this model is not consistent with primordial nucleosynthesis and the @xmath71 data .
it is a misunderstanding .
in fact , in a broad scope of parameters , the primordial nucleosynthesis and @xmath71 can be explained based on the @xmath72 dark matter model ( or the mirror dark matter model ) and this cosmological model .
the first peak of the @xmath73 power spectra is the evidence of existence of the elementary wave .
the elementary wave began at reheating and ended at recombination after @xmath74 years according to the conventional theory . but according to the present model , it is necessary that @xmath75 because the sound speed @xmath76 and @xmath77 here @xmath78 is the period in which @xmath79 descends into @xmath80 @xmath81 @xmath82 and @xmath83 are the physical quantities in the conventional theory . consequently , space - time is open or @xmath84 according to the present model . in section 2 , action ,
energy - momentum tensor and field equations are presented ; in section 3 , spontaneous breaking of symmetry and the gravitational mass density of the vacuum state are discussed ; in section 4 , evolving equations of space are derived out ; in section 5 , temperature effects are considered ; in section 6 , the inflation process and change of @xmath85 and @xmath86 are discussed ; in section 7 , contraction of space , the highest temperature and inflation of space are considered ; in section 8 , expansion of space after inflation is discussed ; in section 9 , existing and distributive forms of @xmath87 color single states in the @xmath39 are discussed ; in section 10 , dynamics of @xmath88 formation and the distributive form of the @xmath87 color single states are got ; in section 11 , some guesses , new predictions and an inference are given ; in section 12 , the primordial nucleosynthesis is discussed ; in section 14 , cosmic microwave background radiation is explained ; section 14 is the conclusions ; section 15 is discussion about conjecture 1 .
in order to solve the above problems , we propose the following conjectures . there are two sorts of matter which are called @xmath89 ( @xmath24 ) and @xmath90 ( @xmath25 ) , respectively . both are symmetric and their contributions to the einstein tensor
are opposite each other@xmath0 there is no other interaction between both except the interaction described by @xmath91 between @xmath92 @xmath93 and @xmath94 @xmath93 . * * * * @xmath95 where @xmath10 is the effective cosmological constant . * * * * the curvature factor @xmath96 in the robertson -walker metric is a monotone and finite function of @xmath45 @xmath97 and @xmath98 for @xmath99 here @xmath100 is the gravitational mass density in the comoving coordinates . * * * * when @xmath101 symmetry holds water and temperature is low , all particles in free states must exist in @xmath101 color single states .
the other premise of the present model is the conventional @xmath101 grand unified theory @xmath102 but it is easily seen that the present model does not rely on the special @xmath103 . provided the conjecture 1 and such a coupling as @xmath104 are kept in a @xmath105 the @xmath103 can be accepted .
in fact , only the conjecture 1 is essential .
the other conjectures are obviously consistent with the conventional theory .
all the following inferences hold water when @xmath106 and @xmath107 due to the conjecture 1 .
@xmath108 @xmath109 and @xmath25 are asymmetric because of the symmetry spontaneously breaking .
matter determines properties of space - time .
different breaking modes of higgs fields correspond to different ground states .
there are two sorts of breaking modes which are called @xmath38 and @xmath39 . in the @xmath110
the expectation values of @xmath92 fields are not zero and the expectation values of all @xmath94 fields are zero .
consequently , the @xmath111 symmetry finally breaks into @xmath112 @xmath113 can get their masses and form @xmath114 @xmath115 and @xmath116 , and the @xmath117 symmetry is still strictly kept , all @xmath118 and @xmath119 bosons are massless and must form @xmath117 color - single states when temperature is low .
there is no electroweak gauge interaction among the @xmath120 color - single states because the @xmath117 is a simple group .
consequently the @xmath117 color - single states can not form @xmath121 @xmath122 and @xmath65 , and must loosely distribute in space as the so - called dark energy .
thus , in the @xmath38 @xmath24 is identified with the conventional matter forming the given world , and @xmath25 can cause space to expand with an acceleration as dark energy and can not be observed except by the repulsion .
in contrast with the dark energy , the gravitational masses of @xmath25 is negative in the @xmath123 @xmath124 .
there are only the repulsion between @xmath24 and @xmath25 when temperature is low@xmath0 the interaction @xmath104 between the @xmath94 fields and the @xmath125 fields is repulsive , the masses of higgs particles are all very large and the higgs particles must decay fast at low temperature .
hence the interaction may be ignored when temperature is low .
thus there are only the repulsion between @xmath24 and @xmath25 when @xmath126 consequently , any bound state is composed of only @xmath113 or only @xmath127 , there is no the transformation of @xmath113 and @xmath127 from one into another when @xmath128 , and if @xmath129 is very large , @xmath28 must be very little in the same region .
thus , in the @xmath130 there must be @xmath131 in a @xmath132 so that @xmath133 may be ignored .
@xmath134 the equivalence principle still holds for the ordinary particles . in the @xmath135 @xmath136
are identified as the ordinary particles to form the given world and there are only the @xmath137 and there is no @xmath138 hence the gravitational masses of @xmath25 must be positive , i.e. @xmath139 and the gravitational masses of @xmath24 must be negative relatively to the @xmath137 i.e. @xmath140 because of the conjecture 1 .
thus the equivalence principle still holds for @xmath141 ( given matter)@xmath142 but is violated for @xmath24 . thus a @xmath143 falling in a gravitational field must have redshift , but a @xmath144 ( there is no @xmath145 and there are only @xmath87 color single states ) will have purple shift
. this result does not contradict the experiments and observations up to now , because of the above reasons .
in fact , it is too difficult that a @xmath146 observes @xmath113 , because @xmath147 in a @xmath132 , there is only the repulsion between @xmath148 and @xmath25 and the @xmath87 color single states can only loosely distribute in space . in the other hand , there is no reason to demands demand unknown matter to satisfy the equivalence principle .
@xmath149 @xmath28 and @xmath129 can transform from one into another when temperature is high enough , i.e. , @xmath150 ( see later).@xmath0 when @xmath151 the expectation values of all higgs fields and the masses of all particles are zero and the interaction @xmath104 between the @xmath92 fields and the @xmath94 fields is important . consequently , @xmath28 and @xmath129 can transform from one into another by @xmath104 so that @xmath152 , @xmath153 and the symmetry @xmath154 holds in this case .
this is a new case which is different from any given experiment and observation .
in addition , conjecture 1 is necessary in order to solve the following issues .
@xmath108 the cosmological constant issue .
@xmath155 the space - time singularity issue .
@xmath157 is a necessary inference because we can explain evolution of space without @xmath10 . on the basis , the cosmological constant problem
is easily solved .
@xmath158 in contrast with the conventional theory , all @xmath159 @xmath41 and @xmath42 are possible , hence the curvature factor @xmath160 @xmath41 and @xmath42 are all possible as well .
consequently @xmath48 is regarded as a function of the gravitational mass density @xmath161 in the comoving coordinates , i.e. @xmath162 . the evolving equations corresponding to @xmath48 have been derived@xmath35}$ ] .
@xmath163as is well known , the @xmath164 theory has proven that there can be the @xmath164 glue - balls whose masses are not zero .
the @xmath101 color single states can be regarded as generalization of the @xmath164 glue - balls .
in contrast with the @xmath164 glue - balls , there is no the interaction similar with @xmath165 gauge interaction among the @xmath101 color single states because @xmath101 is a single group .
sum up , in the @xmath166 the@xmath167 have only the cosmological effects and can not be observed .
consequently there is no contradiction between conjectures and experiments and observations up to now .
based on the conjecture and the @xmath72 dark matter model ( or mirror dark matter model ) , the following inferences are consistent with the cosmological observations . @xmath108 the evolution of the universe and the relation between distance and redshift ; @xmath155 large - scale structure formation ; @xmath134 the features of huge voids ; @xmath149 primordial nucleosynthesis ( it is necessary to consider simultaneously the conjecture and the @xmath72 dark matter model ) ; @xmath168 cosmic microwave background radiation ( it is necessary to consider simultaneously the conjecture and the @xmath72 dark matter model ) .
@xmath169 @xmath54 can change from @xmath55 to @xmath170 based on the conjectures , there are predicts to be confirmed .
the breaking mode of the symmetry is only one of the @xmath38 and the @xmath171 due to @xmath104 . in the @xmath49
there are only the @xmath172 in the @xmath135 there are only the @xmath173 .
hence the actions should be written as two sorts of form , @xmath174 in the @xmath175 and @xmath176 in the @xmath39 .
because of the conjecture 1 , the structures of @xmath174 and @xmath176 are the same , i.e. @xmath177 when @xmath106 and @xmath178 .
thus , at the zero - temperature we have @xmath179 @xmath180 @xmath181 @xmath182 @xmath183 @xmath184 @xmath185 @xmath186 where the meanings of the symbols are as follows .
@xmath187 , @xmath188 in flat space .
@xmath189 is the scalar curvature . here
@xmath190 and @xmath191 are two parameters and we finally take @xmath192 .
@xmath193 is a parameter which is so taken that @xmath194 in the @xmath38 or @xmath195 in the v - breaking at the @xmath196 @xmath197 .
@xmath198 ( @xmath199 ) is the lagrangian density of all @xmath200 ( @xmath201 ) and their couplings of the @xmath101 @xmath103 except the higgs potentials @xmath202 @xmath203 and @xmath204 . @xmath205 and @xmath206 represents all @xmath200 and all @xmath201 , respectively . @xmath207 and @xmath207 do not contain the contribution of the gravitational and repulsive fields .
it is seen that the set of equation @xmath208 is unchanged when thesubscripts@xmath178 and @xmath106 .
this shows the symmetry between @xmath24 and @xmath209 gibbons and hawking pointed out@xmath210}$ ] that in order to get the einstein field equations , it is necessary @xmath211 this is because it is not necessary that @xmath212 in @xmath213 on the boundary @xmath214 hence @xmath215 is replaced by @xmath216 in @xmath158 @xmath217 is a manifold with four dimensions .
@xmath218 is the boundary of @xmath219 @xmath220 @xmath221 is the outer curvature on @xmath222 @xmath223 is the vertical vector on @xmath214 @xmath224 and @xmath225 is the induced outer metric on @xmath218 .
when @xmath218 is space - like , @xmath226 takes positive sign .
when @xmath218 is time - like , @xmath226 takes negative sign .
the higgs potentials in @xmath227 is the following.@xmath228 @xmath229 @xmath230 where @xmath231 , @xmath232 and @xmath233 are respectively @xmath234 , @xmath235 and @xmath236 dimensional representations of the @xmath237 group , @xmath238 , here the couplings of @xmath239 and @xmath233 are ignored for short@xmath240}.$ ] we do not consider the terms coupling to curvature scalar , e.g. @xmath241 , for a time .
in fact , @xmath242 when temperature @xmath51 is high enough due to the symmetry between @xmath24 and @xmath141 . for short , we take @xmath243 in the paper .
we will see in the following paper that when @xmath244 the duration @xmath245 of inflation can be long enough without the slow approximation . by the conventional method , from @xmath246 we can get @xmath247 considering @xmath192 , from @xmath248 we have @xmath249 \delta g^{\mu\nu}\sqrt { -g}d^{4}x \notag \\ & = \dint \left\ { \frac{1}{\sqrt{-g}}\left
[ \frac{\partial\mathcal{l}_{sm}% \sqrt{-g}}{\partial g^{\mu\nu}}-\left ( \frac{\partial\mathcal{l}_{sm}\sqrt{-g% } } { \partial g_{,\sigma}^{\mu\nu}}\right ) _ { , \sigma}\right ] -\frac{1}{2}% g_{\mu\nu}\left ( v_{s}+v_{0}\right ) \right\ } \delta g^{\mu\nu}\sqrt { -g}% d^{4}x \notag \\ & -\dint \left\ { \frac{1}{\sqrt{-g}}\left [ \frac{\partial\mathcal{l}_{vm}% \sqrt{-g}}{\partial g^{\mu\nu}}-\left ( \frac{\partial\mathcal{l}_{vm}\sqrt{-g% } } { \partial g_{,\sigma}^{\mu\nu}}\right ) _ { , \sigma}\right ] -\frac{1}{2}% g_{\mu\nu}v_{v}\right\ } \delta g^{\mu\nu}\sqrt{-g}d^{4}x \notag \\ & = \dint \frac{1}{2}\left ( t_{s\mu\nu}-g_{\mu\nu}v_{0}-t_{v\mu\nu}\right ) \delta g^{\mu\nu}\sqrt{-g}d^{4}x , \tag{11b}\end{aligned}\ ] ] where@xmath250 -g_{\mu\nu}v_{s } , \tag{12a}\end{aligned}\]]@xmath251 -g_{\mu\nu}v_{v}. \tag{12b}\end{aligned}\ ] ] from @xmath252 we obtain @xmath253 in the @xmath123 it is seen from @xmath254 that @xmath255 is independent of @xmath256 in fact , @xmath257 due @xmath258 in the @xmath38 , hence @xmath259
@xmath260 @xmath261 and @xmath255 are the gravitational energy - momentum tensor density , the gravitational energy - momentum tensor density without the higgs potential and the gravitational potential density of the higgs fields in the @xmath38 , respectively .
analogously , from @xmath262 and @xmath263 we obtain@xmath264 in the @xmath265 from @xmath266 the energy - momentum tensor density which does not contain the energy - momentum tensor of gravitational and repulsive fields can be defined as@xmath267 \notag \\ & \equiv t_{s\mu\nu}+t_{v\mu\nu}-g_{\mu\nu}\left ( v_{sv}+v_{0}\right ) = t_{sm\mu\nu}-g_{\mu\nu}v_{s}=t_{v\mu\nu}\equiv t_{\mu\nu } , \tag{15a}\end{aligned}\]]@xmath268 in fact , @xmath269 due @xmath258 in the @xmath175 , hence @xmath270 it should be pointed out that only @xmath271 is applicable in the @xmath49 and only @xmath272 applicable in the @xmath265 it should be noticed from @xmath273 that the potential energy @xmath204 is different from other energies in essence .
there is no contribution of @xmath274 to @xmath275 i.e. , there is no gravitation and repulsion of the potential energy @xmath256 this does not satisfy the equivalence principle .
but this does not cause any contradiction with all given experiments and astronomical observations because @xmath276 in either of breaking mode .
it is proved that the necessary and sufficient condition of @xmath277 is @xmath278 to be a scalar quantity@xmath279}$ ] .
@xmath174 and @xmath176 are all scalar quantities , hence @xmath280 the geodesic equations of the present model are the same as the conventional equations , i.e.,@xmath281 the field equation can be rewritten as @xmath282 , \tag{18}\ ] ] in the @xmath38 . in order to compare the equation of motion of a @xmath144 and the equation of motion of a v - particle in @xmath38 .
we take the newtonian approximation . under the newtonian limit , from @xmath283 we have@xmath284}$].@xmath285@xmath286 from @xmath287 we have @xmath288
on the other hand , the approximate value of @xmath289 can be found from expression of the ricci tensor , after neglecting the nonlinear terms and the terms that are time derivative , one finds@xmath290 from @xmath291 and @xmath292 we have@xmath293@xmath294@xmath295 consequently , from @xmath296 and @xmath297 we have @xmath298@xmath299 @xmath300 is the same as @xmath301 in ref .
@xmath302 , $ ] and the equation @xmath303 of motion of a @xmath144 is different from that of a @xmath304 in the same gravitational field .
@xmath300 and @xmath303 are consistent with observed data .
the reasons are as follows .
@xmath305 . in the @xmath306 in a @xmath307 @xmath129 must be very small and @xmath133 must be large .
hence the equivalent principle still holds for @xmath308 and the gravitational field , the equation @xmath300 and trajectories of motion of @xmath113 are still the same as those of the conventional theory in observed precision .
@xmath309 in the @xmath306 the equation @xmath303 and trajectories of motion of @xmath136 must be different from those of the conventional theory .
but it is impossible to observe the @xmath117 color single states by a @xmath310 in practice , because @xmath117 color single states can not form dumpling and must loosely distribute in space , and there is only the repulsion between @xmath24 and @xmath25 .
in fact , @xmath303 has only theoretical meanings .
in fact , only the cosmological effects of @xmath25 are important and are consistent with the observed data up to now .
ignoring the couplings of @xmath312 and @xmath313and suitably choosing the parameters of the higgs potential@xmath142 analogously to ref.@xmath314 } $ ] .
we can prove from @xmath315 that there are the following vacuum expectation values at the zero - temperature and the tree - level approximation , @xmath316 @xmath317 @xmath318 @xmath319 @xmath320 ignoring the contributions of @xmath312 and @xmath321 to @xmath322 at the zero - temperature we get @xmath323 @xmath324 @xmath325 we take @xmath326 from @xmath327-@xmath104 and @xmath328-@xmath329 it can be proved that all @xmath94 bosons can get their masses big enough .
the masses of the higgs particles exclusive of the @xmath330 and the @xmath331are respectively in the @xmath175 @xmath332 @xmath333 @xmath334 @xmath335 we can choose such parameters so that @xmath336 e.g. , @xmath337 @xmath338 and @xmath339 .
it is easily seen from @xmath340 that all real components of @xmath341 have the same mass @xmath342 , all real components of @xmath343 have the same mass @xmath344 in the @xmath123 the @xmath38 and the @xmath39 are symmetric because @xmath24 and @xmath141 are symmetric .
hence when @xmath345 and @xmath346 in @xmath347 the formulas are still kept .
the characters of the vacuum state are as follows .
taking @xmath348 at the zero - temperature , considering @xmath349 and @xmath350 in the @xmath38 and the symmetry between @xmath24 and @xmath351 in the vacuum state we have@xmath352@xmath353 applying the conventional quantum field theory to the present model , we have @xmath354 because of the conjecture 1 , @xmath355 and @xmath356 consequently , we have@xmath357 here @xmath9 and @xmath7 are the mass density and the gravitational mass density of the vacuum state , respectively@xmath0 according to the our quantum field theory@xmath21}$],@xmath358 it is seen that in any case , @xmath359 is necessary provided the conjecture 1 is valid .
provided the cosmological principle is valid , the metric tensor is the robertson - walker metric which can be written @xmath360 where @xmath43 is the curvature factor .
@xmath43 is regarded an arbitrary real constant in the friedmann model . when @xmath361 @xmath362 and @xmath363 , @xmath364 is unchanged .
thus , without losing generality , @xmath43 may be taken as @xmath365 , @xmath366 or @xmath367 . here
@xmath368 must be a positive number , hence @xmath43 can not change from @xmath365 into @xmath366 or @xmath367 by altering @xmath369 matter in the universe may approximately be regarded as ideal gas evenly distributed in the whole space when temperature is not very high .
the energy - momentum tensor densities of the ideal gas are @xmath370@xmath371 where @xmath372 is a 4-velocity , @xmath238 .
considering the potential energy densities in @xmath373 , we can rewrite @xmath374 and @xmath375 as @xmath376 u_{\mu } u_{\nu}+\widetilde{p}_{s}g_{\mu\nu } , \tag{43}\]]@xmath377@xmath378 u_{\mu } u_{\nu}+\widetilde{p}_{v}g_{\mu\nu } , \tag{45}\]]@xmath379@xmath380 in fact , @xmath381 due @xmath258 in the @xmath38 , hence @xmath382 substituting @xmath383 and the rw metric into @xmath384 and considering @xmath385 and @xmath386 which implies that a medium moves only as expanding of space , we get the friedmann equations in the @xmath387 @xmath388 r^{2}=\eta\left [ \rho_{sg}+v_{sg}% \right ] r^{2}\equiv\eta\widetilde{\rho}_{sg}r^{2 } , \tag{48a } \\
\rho_{sg } & \equiv\rho_{s}-\rho_{v},\text{\ } v_{sg}\equiv v_{s}+v_{0}-v_{v},% \text { \ } \widetilde{\rho}_{sg}=\rho_{sg}+v_{sg},\text { \ } \eta \equiv8\pi g/3 , \tag{48b}\end{aligned}\ ] ] @xmath389 r.\equiv-\frac{1}{2}\eta\left ( \widetilde{\rho% } _ { sg}+3\widetilde{p}_{sg}\right )
r \notag \\ & = -\frac{1}{2}\eta\left [ \rho_{sg}+3p_{sg}-2v_{sg}\right ] , \text { \ \ } % p_{sg}\equiv p_{s}-p_{v},\text { \ \ } \widetilde{p}_{sg}=p_{sg}-v_{sg}. \tag{49}\end{aligned}\ ] ] analogously , from @xmath390 , @xmath391 and @xmath392 , we get @xmath393 r^{2}=\eta\widetilde{\rho}_{vg}r^{2 } , \tag{50}\]]@xmath394 r=-% \frac{1}{2}\eta\left ( \widetilde{\rho}_{vg}+3\widetilde{p}_{vg}\right ) r , \tag{51}\ ] ] in the @xmath265 in the @xmath49 only @xmath395 is applicable , and in the @xmath135 only @xmath396 is applicable .
it is impossible when temperature is low that the @xmath38 transforms into the @xmath39 because of the coupling @xmath104 .
but it is possible that @xmath397 @xmath258 when temperature is high enough .
thus the transformation @xmath398 is possible , i.e. , the @xmath38 can transform into the @xmath39 via the highest temperature ( see the following ) . from @xmath395
we obtain @xmath399 /dt+r^{3}\left ( dv_{sg}/dt\right ) = -\left ( p_{s}-p_{v}\right ) dr^{3}/dt+r\left ( dk/\eta dt\right ) .
\tag{53}\ ] ] let @xmath400 and @xmath401 denote the mass density of particles whose static masses are not zero and the mass density of photon - like particles , respectively , we have @xmath402 and @xmath403 @xmath404 and @xmath405 may be ignored when temperature is low .
when @xmath400 is major and @xmath406 @xmath407 ( see the following section ) and @xmath408 may be ignored , from @xmath409 we have@xmath410 in the @xmath38 , here @xmath411 is a constant .
in contrast to the conventional theory , it is possible @xmath412 @xmath413 or @xmath414 when @xmath415 both @xmath416 and @xmath417 are all very big , hence @xmath418 can not transform into @xmath419 by @xmath420 consequently one has@xmath421 analogously , we have @xmath422 in the @xmath39 .
when photon - like gases are major and @xmath407 and @xmath423 may be ignored , after thermal equilibrium , @xmath424 we have@xmath425 dt & = -%
\overset{\cdot}{v}_{sg}r^{4}\sim0 , \notag \\ \left ( \rho_{s\gamma}-\rho_{v\gamma}\right ) r^{4 } & = d_{s } , \tag{56a}\end{aligned}\ ] ] in the @xmath123 when the transformation @xmath426 into @xmath427 may be ignored , we have@xmath428 similarly , one has @xmath429@xmath430 in the @xmath39 . provided the cosmological principle is valid , the metric tensor is the robertson - walker metric .
@xmath43 in @xmath432 is regarded an arbitrary real constant in the friedmann model . when @xmath361 @xmath362 and @xmath363 , @xmath364 is unchanged .
thus , without losing generality , @xmath43 may be taken as @xmath365 , @xmath366 or @xmath367 . here
@xmath368 must be a positive number , hence @xmath43 can not change from @xmath365 into @xmath366 or @xmath367 by altering @xmath369 the r - w metric is undoubtedly right when all gravitation masses are positive , i.e. @xmath433 in contrast with the conventional theory , according to the present model , all @xmath40 @xmath41 and @xmath42 are possible .
hence @xmath43 in @xmath434 should be changeable from @xmath435 to @xmath367 corresponding to change of @xmath436 from @xmath437 to @xmath438 .
we consider @xmath43 to be a function of @xmath45 @xmath100 is the gravitational mass density in comoving coordinates@xmath0 @xmath439 can change as @xmath189 because @xmath440 and @xmath441 on the other hand , although @xmath442 @xmath46 can also change ( see section 8.7 ) .
this is because the repulsive potential energy is chiefly transformed into the kinetic energy of color single states .
hence we have @xmath443@xmath444@xmath445 r-\frac{1}{3}\frac{\overset{\cdot}{r}\overset{\cdot\cdot}{k}}{% \overset{\cdot}{k}}. \tag{59}\]]@xmath446 we discuss the evolving equations as follows@xmath35}$ ] .
@xmath447 when @xmath448 from @xmath449 we have @xmath450 substituting @xmath451 into @xmath452 we get the friedmann equations @xmath453 or @xmath454 anew . hence the equations are self - consistent .
thus , when @xmath448 we can still determine @xmath455 by @xmath453 or @xmath456 @xmath457 is possible . because @xmath458 is a monotone and finite function of @xmath100 and @xmath459
it is necessary when @xmath460 or @xmath461 @xmath462 slowly changes so that @xmath457 .
in fact , considering @xmath463 to be the gravitational mass density in the comoving coordinates , we have@xmath464 when @xmath465 @xmath466 so that @xmath467 from the conjecture 3 , we can also determine only when @xmath468 so that @xmath469 @xmath470 is important .
because @xmath471 is a monotone and finite function of @xmath45 @xmath472 and @xmath459 it is necessary @xmath473 when @xmath474 and @xmath475 when @xmath476 when @xmath477 or @xmath461 we will regard @xmath478 as a constant , e.g. @xmath479 for @xmath477 and @xmath480 for @xmath481 @xmath482 when @xmath483 , from @xmath484 we have @xmath485 this is similar to the conventional theory only when @xmath486 according to present model , @xmath487 and @xmath488 or @xmath42 are possible as well .
thus , the present model is different from the conventional theory .
@xmath489 when @xmath490 @xmath491 becomes @xmath492@xmath493 this is because only when @xmath494 , it is possible that @xmath495 hence @xmath496 .
it is seen that although @xmath497 it is still possible @xmath498 and @xmath499 in the case space expands with a deceleration .
this is different from the conventional theory in which when @xmath500 @xmath501 is necessary .
it is seen that when @xmath502 and @xmath503 @xmath470 must be considered . @xmath504 .
although @xmath48 is a function of @xmath505 when @xmath503 @xmath487 and @xmath189 are still independent of @xmath506}.$ ] @xmath507 . in the @xmath38 , @xmath28 can transform into @xmath129 because of @xmath508 when @xmath509 after reheating , in fact , @xmath510 it is necessary that @xmath511 at some a time , because @xmath512 @xmath513 , and @xmath514 when @xmath515 @xmath516 @xmath508 is marked and the universe is matter - dominated so that @xmath517 may be ignored .
consequently from @xmath518 we see@xmath399 /dt
= r\left ( dk/\eta dt\right ) < 0 .
\tag{65c}\ ] ] this implies @xmath28 can transform into @xmath129 because of @xmath519 when @xmath509 in fact , in this stage , @xmath520 can be fast formed and @xmath521 holds .
the thermal equilibrium between the @xmath136 and the @xmath113 can be realized by only @xmath420 the higgs bosons are hardly produced because the masses of the higgs particles are all very big at low temperature .
consequently , the interaction between the @xmath136 and the @xmath113 may be ignored so that there is no thermal equilibrium between the @xmath136 and the @xmath522 thus , when temperature is low , we should use two sorts of temperature @xmath523 and @xmath524 respectively to describe the thermal equilibrium of @xmath25 and the thermal equilibrium of @xmath525 generally speaking , @xmath526 .
when temperature @xmath527 is so high that @xmath528 in @xmath110 the masses of @xmath92 bosons will tend to zero ( see below ) . on the other hand , because @xmath529 in the @xmath49 @xmath530 @xmath342 and @xmath531 will tend to zero as @xmath532 tends to zero as well .
consequently @xmath533 @xmath312 and @xmath534 can be enormously produced and easily transformed into @xmath535 @xmath341 and @xmath343 by the couplings in @xmath536 other @xmath136 can be easily produced by the couplings of @xmath117 as well@xmath0 consequently thermal equilibrium between the @xmath136 and the @xmath113 will appear provided @xmath524 is high enough .. in the case , contraction of space will stop and inflation must occur . thus there must be the highest temperature @xmath537 .
the influence of finite temperature on the higgs potential in the present model are consistent with the conventional theory . for short ,
we consider only @xmath231 and @xmath538 @xmath238 .
when @xmath233 is considered as well , the following inferences are still qualitatively valid .
for @xmath539 to ignore the terms proportional @xmath540 @xmath541 , the finite - temperature effective potential approximate to 1-loop in flat space is@xmath542}$ ] @xmath543 considering the influence of the expectation values @xmath544 @xmath545 and @xmath546 and ignoring the terms irrelevant to @xmath533 we have @xmath547@xmath548 in the @xmath38 , @xmath549 @xmath550 and @xmath551 from @xmath552 we find @xmath553 similarly , for @xmath554 we have@xmath555@xmath556@xmath557 in the @xmath38 , because of @xmath558 it will be proved ( see section 5.3)@xmath559 for @xmath560 ignoring the contributions of the higgs fields and the fermion fields to one loop correction and only considering the contribution of the gauge fields to one - loop correction , when @xmath561 , @xmath43 is the boltzmann constant , the finite - temperature effective potential approximate to 1-loop in flat space is@xmath562}$]@xmath563 where @xmath564 @xmath565 in general , @xmath566 and @xmath567 we take @xmath568 for simplicity@xmath0 @xmath569 is a parameter at which the renormalized coupling - constant @xmath570 is defined , @xmath571 only considering the contribution of the expectation values of @xmath572 and @xmath554 to @xmath573 taking @xmath574 , and ignoring the term @xmath575 unconnected with @xmath576 from @xmath577 @xmath578 we have @xmath579 similarly , we have@xmath580 the critical temperature @xmath581 is such a temperature at which the minima are degenerate , i.e. , @xmath582 . in other words , @xmath583 when @xmath584and @xmath585 when @xmath586 @xmath581 and @xmath587 can be determined from @xmath588 by that when @xmath589 @xmath590@xmath591 /c .
\tag{73b}\end{aligned}\ ] ] in the @xmath38 , @xmath592 considering @xmath585 when @xmath593 and @xmath594 and taking @xmath595 we find@xmath596 /\left
( c+\frac{15}{4}% \alpha\right ) < \frac{\mu^{2}}{\lambda}. \tag{73c}\ ] ] when @xmath597 @xmath598 , \tag{74a}\end{aligned}\ ] ] where @xmath599 is called the effective mass of @xmath600 which implies that the temperature effect is considered .
when @xmath601@xmath602 similarly , from @xmath603 we have @xmath604 /c . \tag{75b}\end{aligned}\ ] ] in the @xmath38 , @xmath605 considering @xmath594 and taking @xmath606 we find@xmath607 /c , \tag{75c }
\\ t_{v,\varphi cr\max}^{2}\left ( t_{s},t_{v}\right ) & = b\sigma^{2}e^{-1/2}/c < b\sigma^{2}e^{-1/2}/\lambda<\mu^{2}/\lambda .
\tag{75d}\end{aligned}\ ] ] @xmath608 implies @xmath609 @xmath610 will increase from @xmath366 to @xmath611 as @xmath612 decreases from @xmath613 to @xmath366 due to @xmath614 in the case , it seems @xmath615 so that @xmath616 to be possible@xmath0 in fact , this is impossible for the space - contraction process . in the case , it is necessary @xmath617 due to @xmath618 hence @xmath572 and @xmath554 or @xmath619 and @xmath600 and @xmath554 can transform from one into another by @xmath620 and @xmath621 .
consequently @xmath622 and @xmath623 so that @xmath624 ( see below ) .
hence in the @xmath175 , @xmath625 when @xmath626@xmath627 .
\tag{76b}\end{aligned}\ ] ] in fact , only when @xmath628 @xmath629 due to @xmath420hence when @xmath630@xmath631 in fact , in general , the case can not emerge , because when @xmath632 @xmath524 is very large .
hence @xmath28 and @xmath129 can transform from one into another so that @xmath633 thus @xmath634 can not emerge@xmath0 if @xmath635 @xmath636 it is obvious @xmath637 in the @xmath38 , when @xmath638 @xmath639 thus the critical temperature @xmath640 is determined by @xmath641 @xmath642 thus , when @xmath643@xmath644 when @xmath645@xmath646 when @xmath647 @xmath648 and when @xmath649 @xmath650 considering @xmath651 when @xmath652 from @xmath653 we find@xmath654 when @xmath655,@xmath656 in the @xmath38 , when @xmath657 it is necessary @xmath658 in the case , @xmath28 and @xmath129 can transform from one into another and @xmath659 hence @xmath660 consequently there still is @xmath661 hence in the @xmath175 , when @xmath662 there still are@xmath663 in the @xmath668
we consider the space contraction process . in low temperatures , @xmath669 , @xmath670 and the masses of all @xmath671 single states
are not zero , i.e. , @xmath672 thus @xmath673 and @xmath419 are major .
@xmath674 so that space can contract .
according to the conjecture 3 and the discussion about @xmath451 , @xmath675 and @xmath676 when@xmath677 so that @xmath678-@xmath679 are still applicable . when @xmath680 the transformation @xmath28 into @xmath129 may be ignored .
from @xmath681 one can rewrite @xmath678-@xmath682 as@xmath683 where @xmath684 @xmath685 @xmath238 and @xmath686 it is seen that space will monotonously contract faster and faster .
@xmath524 must go up high as @xmath189 decreases , because the non - zero momentum of a free particle @xmath687 @xmath688 and @xmath689 when @xmath524 and @xmath523 are high enough so that masses of particles may be ignored , if the transformation @xmath28 into @xmath129 may still be ignored , from @xmath690 and @xmath691 we have @xmath692 where @xmath693 @xmath694 @xmath695 is the total number of the spin states of @xmath696 @xmath697 considering @xmath24 and @xmath25 are symmetric , we have @xmath698 it is seen that space contraction will cause @xmath524 and @xmath523 to rise
. if @xmath524 and @xmath523 are high enough and the transformation @xmath28 into @xmath699 can not be ignored , there will be @xmath700 and @xmath701 this is a expected result .
if there is only one sort of matter as the conventional theory or @xmath28 and @xmath129 can not transform from one into other , space will continue to contract and @xmath524 and @xmath523 will continue to rise provided @xmath702 in fact , in this case , space will contract to a singular point and @xmath524 and @xmath523 tend to infinite .
in contrast with the conventional theory , when @xmath524 and @xmath523 are high enough , @xmath28 and @xmath129 can transform from one into another by @xmath620 i.e. , @xmath703 so that@xmath704 the expectation values @xmath705 , @xmath706 , @xmath707 and @xmath708 will tend to zero when @xmath709 because of @xmath710 @xmath711 @xmath712 and @xmath713 consequently , the masses of all particles which originating from the couplings with @xmath705 , @xmath706 , @xmath714 and @xmath715 are zero .
thus , the @xmath716 particles and the @xmath717 or the @xmath718 particles can transform from one into another by the @xmath719 couplings so that the @xmath716 particles can enormously emerge @xmath720 . the @xmath92 particles and the @xmath94 particles can transform from one into another by the coupling @xmath721 so that the thermal equilibrium between @xmath24 and @xmath25 comes into being .
after thermal equilibrium , the number and the energy of every sort of particles will satisfy statistical distribution determined by their spins .
thus we prove that @xmath722 is a necessary result of space contraction because of @xmath723 it is seen that when @xmath724 i.e. , @xmath725 @xmath24 and @xmath25 are completely symmetric , and both @xmath726 and @xmath117 strictly hold water . it is seen the end of space contraction is in the most symmetric state in which @xmath727 .
@xmath728 is the sufficient but is not the necessary condition for @xmath722 .
in fact , provided the following conditions can be realized , @xmath722 can come into being when @xmath524 and @xmath729 as well .
@xmath730 or @xmath731 or temperature is high enough so that @xmath732 can enormously emerge . here
@xmath733 and @xmath734 denote fermions and gauge bosons , respectively . @xmath735 and @xmath736 are small .
thus @xmath231 or @xmath737 and @xmath738 or @xmath734 can transform from one into other by the @xmath101 couplings .
@xmath739 @xmath740 thus , @xmath741 and @xmath742 can transform from one into other by @xmath104 the two conditions can be realized as well when @xmath743 from @xmath744 and @xmath745 , in the @xmath38 we can rewritten @xmath746 as @xmath747 when @xmath748 @xmath749 @xmath750 as @xmath751 and@xmath752 in the @xmath49 @xmath753 and @xmath754 when @xmath755 and space contracts@xmath0 hence when @xmath756 it is necessary @xmath757 in fact , if @xmath758 in @xmath759 is very large , @xmath760 is definitely very large , e.g. @xmath761 consequently , because of @xmath762 and @xmath763 @xmath764 is definitely very small and @xmath765 .
this case satisfies the two conditions @xmath305 and @xmath739 above .
if @xmath766 is very small , @xmath767 must be less than @xmath758 provided @xmath768 when @xmath769 @xmath770 because @xmath771 hence @xmath772 when @xmath773 it is seen provided @xmath774 @xmath775 is definitely very small and the conditions @xmath305 and @xmath739 above are definitely satisfied . thus , @xmath572 and @xmath776 can transform into @xmath554 by @xmath104 so that @xmath523 and @xmath129 increase . hence @xmath722 can come into being , and @xmath777 in the @xmath38 when @xmath524 and @xmath778 as well@xmath0 thus we prove @xmath779 to be valid . when @xmath780 continues to decrease to zero as @xmath524 rises to @xmath34 due to space contraction , @xmath781 and @xmath782 even @xmath321 and @xmath343
are considered@xmath142 the above conclusions still hold water qualitatively .
if we only discuss space inflation of the present model , the higgs potentials @xmath783 can be simplified as follows.@xmath784@xmath785@xmath786 because of @xmath787 the breaking mode can only be the @xmath175 @xmath788 or the @xmath39 @xmath789 for simplicity , we only consider the @xmath39 and ignore @xmath343 and @xmath321 for a time@xmath790@xmath791 the critical temperatures and the effective masses are determined by @xmath792 substituting @xmath793 or @xmath794 into @xmath795 we obtain the critical temperatures and the effective masses in the @xmath39 or in the @xmath123 there is no influence of @xmath796 to @xmath797 because @xmath798 in the @xmath265 hence the potential @xmath799 , in fact , is equivalent to the @xmath800 potential discussed by ref .
@xmath801.$ ] when temperature descends due to space inflation so that @xmath802 @xmath803 @xmath804 in the case , there is space inflation in the slow roll approximation .
the duration @xmath805 ( @xmath806 ) of inflation has been estimated by the ref .
@xmath807 $ ] .
the difference between @xmath799 and @xmath808 is the following .
the minimum @xmath809 is the absolute minimum at the zero - temperature in the @xmath171 according to @xmath799 .
in contrast to @xmath810 according to @xmath811 @xmath812 is not absolute minimum at the zero - temperature in the @xmath39 , but @xmath813 @xmath814 is the absolute minimum at the zero - temperature .
we will see @xmath783 has more important significance . in the inflation period ( see section 7.1 , @xmath817 )
, @xmath818 and @xmath819 @xmath820 can be rewritten as@xmath821 after reheating , @xmath822 in the @xmath39 , @xmath823 \notag \\ & = -\frac{\eta}{2}\left [ \rho+3p\right ] , \tag{90}\end{aligned}\ ] ] where@xmath824 @xmath825 is the conventional positive mass density , and @xmath826 is the effective pressure density relative to @xmath825 .
when the evolution equation is in the form @xmath827 @xmath85 is defined as @xmath828 thus , in the inflation period , from @xmath829 we find @xmath830 for @xmath831 we have @xmath832 the static masses of all color - single states are non - zero , hence @xmath833 in the early period after reheating , temperature is very high so that the masses of particles may be neglected , @xmath834 @xmath835 and @xmath836 .
consequently,@xmath837 from @xmath521 and section 8.7 , we know @xmath838 is changeable .
if after some a time,@xmath839 considering @xmath840 and @xmath841 when @xmath189 is large enough , @xmath842 and temperature is low so that @xmath843 and @xmath844 from @xmath845 and @xmath846 we have@xmath847 it is seen that @xmath85 can change from @xmath55 to @xmath848 according to the present model.@xmath0 according to the simplified higgs potential @xmath849 in the slow rolling approximation , we can get the results similar to @xmath850 and @xmath851 in fact , space inflation or expansion with an acceleration is independent of the slow approximation , and @xmath85 is not important for the present model . although there is no slow approximation ,
space inflation can be explained because @xmath852 in the inflation process and the parameters @xmath368 and @xmath85 in @xmath783 are so chosen that @xmath853 .
although @xmath854 after reheating , space expansion with a deceleration or an acceleration can still be explained by the present model .
@xmath86 is defined as@xmath855 when @xmath856 ( or @xmath857 according to the simplified higgs potential @xmath858 ) , @xmath859 @xmath860 @xmath861 , @xmath862 @xmath863 to be finite and @xmath864 consequently,@xmath865 after the reheating , @xmath866 @xmath867 and @xmath868 when @xmath869 @xmath870 and @xmath871@xmath872 where @xmath873 due to @xmath437 . when @xmath874 @xmath875 from @xmath876 we have@xmath877 when @xmath878 @xmath879 .
considering @xmath880 @xmath881 @xmath882 @xmath883 may be neglected and @xmath884 when @xmath189 is large enough , we have@xmath885 in the case , space expands with an acceleration , although @xmath822
on the basis of the cosmological principle , if there is the space - time singularity , it is a result of space contraction .
thus , we discuss the contracting process and find the condition of space inflation . from the contracting process
we will see that there is no space - time singularity in present model we do not consider the couplings of the higgs fields with the ricci scalar @xmath886 for a time .
we will see in the following paper that the following conclusions still hold water when such couplings as @xmath887 are considered . in fact , @xmath888 because there is the strict symmetry between @xmath148 and @xmath25 when @xmath889 .
we chiefly discuss change of @xmath890 and @xmath891 as temperature@xmath142 @xmath892 in the contracting process of space for short .
when @xmath893 is considered as well , the inferences are still valid qualitatively . there no singularity of space - time in the present model .
@xmath894 the end of space contraction is in the most symmetric state in which @xmath895 @xmath896 and @xmath897 so that @xmath898 this has been proved in section @xmath482 there is no singularity in the present model on the basis of the cosmological principle . if space does not contract because @xmath498 or @xmath899 it is necessary that there is no space - time singularity . provided space contract because @xmath900 @xmath524 and @xmath901 must rise .
when @xmath902 @xmath903 the masses of all particles which originate from the couplings with @xmath904 and @xmath905 are zero .
consequently @xmath28 and @xmath129 can transform from one into another .
thus @xmath152 , @xmath896 and @xmath906 i.e. , the most symmetric state comes into being . in the state
, both the @xmath87 and the @xmath117 are strictly kept . in the case , from the conjecture 3 and the discussion about @xmath907 we may take @xmath479 . thus @xmath453 is reduced to @xmath908@xmath909 consequently space inflation must occur and temperature will fast descend .
let @xmath910 if @xmath911 @xmath189 can continue to decrease with a deceleration or stop contracting . hence there must be the least scale @xmath912 the critical temperature @xmath34 , the highest temperature @xmath52 and the largest energy density @xmath913 @xmath914 thus , when @xmath189 decreases to @xmath915 and space inflation must occur,@xmath916 where @xmath917 is the hubble constant at @xmath918 .
@xmath919 * * * * is just the last moment of the world in the @xmath387 and the initial moment of the world in the @xmath265 @xmath920 and @xmath34 are two new important constants , and @xmath52and @xmath60 are determined by @xmath921 .
it is seen that all @xmath189 , @xmath51 and @xmath825 must be finite in the case .
@xmath690 is considered in @xmath922 the meanings of the parameters are that when @xmath923 @xmath924 and @xmath925 and when @xmath926 , @xmath927 or @xmath928 and @xmath929 we know that the duration of inflation @xmath245 may be long enough for inflation ( see below @xmath930 ) .
after @xmath931 @xmath189 has a large enough increase .
@xmath932 is the condition of space inflation . because the masses of all particles which originate from the couplings with @xmath933 or @xmath934 are zero and @xmath897 when @xmath889 , considering @xmath935 we have@xmath936 thus , when @xmath150 , from @xmath937 we can rewrite the condition of inflation @xmath932 as @xmath938 here @xmath939 and @xmath479 is considered .
thus , when @xmath940 there must be space inflation . if @xmath941 or @xmath942 this implies that @xmath943 already occurs before @xmath189 contracts to @xmath944 or @xmath524 rises to @xmath945 i.e. , @xmath946 and @xmath947 consequently @xmath34 and @xmath948 can not be arrived and there still are @xmath949 and @xmath950 . in the case , all @xmath58 , @xmath951 @xmath952 @xmath28 and @xmath129 must still be finite because of the cosmological principle , i.e. there is no space - time singularity . in the case , it is necessary @xmath953 because @xmath58 is the end of contracting @xmath189 . in the case ,
when @xmath954 the evolving equations are still @xmath955 and space will expand still in the @xmath38 mode , but space inflation can not occur . to sum up , we see that in any case of the contracting process , there must be @xmath956 and the finite @xmath957 because of the cosmological principle , all @xmath958 @xmath959 @xmath960 and @xmath961 are finite because of the cosmological principle . hence @xmath962 @xmath963 and @xmath964
must be finite due to @xmath965 substituting the finite @xmath966 into the einstein field equation @xmath967 we see that @xmath968 and @xmath969 must be finite .
thus , we have proved that there is no singularity in present model .
in fact , when @xmath970 @xmath817 is consistent with the lemaitre model without singularity in which @xmath971 , @xmath435 and the cosmological constant @xmath972}.$ ] we first intuitively explain the reasons that there is no space - time singularity . it has been proved that there is space - time singularity under certain conditions@xmath1}$ ] .
these conditions fall into three categories .
first , there is the requirement that gravity shall be attractive .
secondly , there is the requirement that there is enough matter present in some region to prevent anything escaping from that region .
the third requirement is that there should be no causality violations .
hawking considers it is a reasonable conjecture that @xmath437 and @xmath973}.$ ] but this conjecture is not valid .
the gravitational mass density @xmath974 @xmath41 or @xmath42 are all possible in the present model@xmath0 it is necessary @xmath31 because @xmath28 and @xmath699 can transform from one to another when @xmath889 .
it is seen that @xmath487 does not increase not only , but also decreases to zero when @xmath975 hence the second condition is violated .
the key of non - singularity is that there are @xmath24 and @xmath25with opposite gravitational masses and both can transform from one to another when @xmath889 we explain the reasons that there is no space - time singularity from the hawking theorem as follows .
hawking has proven the following theorem@xmath976}$ ] .
the following three conditions can not all hold : @xmath977 every inextendible non - spacelike geodesic contains a pair of conjugate point ; @xmath978 the chronology condition holds on @xmath979 @xmath980 there is an achronal set @xmath981 such that @xmath982 or @xmath983 is compact .
the alternative version of the theorem can obtained by the following two propositions .
proposition @xmath984}$ ] : if @xmath985 and if at some point @xmath986 the tidal force @xmath987 is non - zero , there will be values @xmath988 and @xmath989 such that @xmath990 and @xmath991 will be conjugate along @xmath992 , providing that @xmath992 can be extended to these values .
proposition @xmath993}$ ] : if @xmath985 everywhere and if at @xmath994 @xmath995cd[e}k_{f]}$ ] is non - zero , there will be @xmath996 and @xmath997 such that @xmath998 and @xmath999 will be conjugate along @xmath1000 provided that @xmath1000 can be extended to these values . an alternative version of the above theorem is the following .
space - time ( @xmath1001 ) is not timelike and null geodesically complete if : @xmath266 @xmath1002 for every non - spacelike vector @xmath1003 @xmath1004 the generic condition is satisfied , i.e. every non - spacelike geodesic contains a point at which @xmath1005cd[e}k_{f]}k^{c}k^{d}\neq0 $ ] , where @xmath1006is the tangent vector to the geodesic .
@xmath1007 the chronology condition holds on @xmath1008 ( i.e. there are no closed timelike curves ) .
@xmath1009 there exists at least one of the following : @xmath1010 a compact achronal set without edge , @xmath1011 a closed trapped surface , @xmath1012 a point @xmath826 such that on every past ( or every future ) null geodesic from @xmath826 the divergence @xmath1013 of the null geodesics from @xmath1014 becomes negative ( i.e. the null geodesics from @xmath826 are focussed by the matter or curvature and start to reconverge ) .
this theorem is an alternative version of the above theorem .
this is because if @xmath1008 is timelike and null geodesically complete , @xmath266 and @xmath1004 would imply @xmath1015 by above propositions @xmath365 and @xmath1016 @xmath266 and @xmath1017 would imply @xmath1018 , and @xmath248 is the same as @xmath1019 in fact , @xmath1020 is determined by the gravitational energy - momentum tensor @xmath1021 .
according to the conventional theory , @xmath1022 so that the above theorem holds@xmath0 in contrast with the conventional theory , according to conjecture @xmath365 , in the @xmath1023 , @xmath1024 .
\tag{109a}\end{aligned}\ ] ] consequently , @xmath1025 @xmath41 and @xmath42 are all possible .
thus , although the strong energy condition still holds , i.e. @xmath1026 k_{a}k_{b}\geq0 , \tag{109b}\ ] ] the conditions of propositions @xmath365 and @xmath1027 and condition @xmath156 do no longer hold , because the gravitational mass density @xmath487 determines @xmath968 and @xmath1028 hence @xmath1015 and @xmath1018 do not hold , but @xmath1029 still holds , and @xmath1008 is timelike and null geodesically complete.@xmath0 as mentioned before , the duration of space inflation is finite ( see below @xmath1030 ) .
supposing @xmath1031 , @xmath1032 for @xmath1033 and considering @xmath1034 ( see @xmath1035 ) , from @xmath1036 we can estimate @xmath52 , @xmath1037 the temperature will strikingly decrease in the process of inflation , but the potential energy @xmath1038 can not decrease to @xmath1039 at once , because this is a super - cooling process .
* * we can get the expecting results by suitably choosing the parameters in @xmath1040 in order to estimate @xmath1041 taking @xmath1042 from @xmath1043 we have @xmath1044 we can take @xmath52 to be the temperature corresponding to @xmath103 because the @xmath101 symmetry strictly holds water at @xmath52 . taking @xmath1045 and @xmath1046
we have @xmath1047 if the duration of the super - cooling state * * * * is @xmath1048 @xmath58 will increase @xmath1049 times . as mentioned before ( see below @xmath930 ) , the duration @xmath245 of inflation may be long enough , @xmath1050}.$ ] taking @xmath1051 and @xmath1052 we have @xmath1053 and @xmath1054 .
the result is consistent with the guth s inflation model@xmath1055}$ ] . before inflation ,
the world in the @xmath38 is in thermal equilibrating state .
if there is no @xmath351 because of contraction by gravitation , the world would become a thermal - equilibrating singular point , i.e. , the world would be in the hot death state .
as seen , it is necessary that there are both @xmath24 and @xmath25 and both the @xmath38 and the @xmath39 .
the parameters in the higgs potential can be so suitalbly chosen that @xmath245 is suitable .
we discuss the problem in the following paper .
after inflation , the temperature must sharply descend . in the case
, it is easily seen that the state with @xmath1056 is no longer stable and must decay into such a state with @xmath1057 either of the @xmath38 and the @xmath39 can come into being , because @xmath24 and @xmath25 are completely symmetric at @xmath1058 .
let the @xmath39 comes into being@xmath142 then @xmath1059 and @xmath87 symmetry is kept . in this case , @xmath919 can be regarded as the initial moment for the world in the @xmath265 thus we take @xmath1060 . ignoring the effect of temperature @xmath1061 , from @xmath1062
we see the phase transition of the vacuum is as follows , @xmath1063 after transition , @xmath1064 must firstly transform into a lot of the @xmath94 particles .
the @xmath94 particles can decay fast into the @xmath119 bosons and the @xmath118 by the @xmath117 couplings . on the other hand , because of the coupling @xmath1065 the @xmath94 particles can transform into the @xmath125 particles as well@xmath0 the @xmath92 particles @xmath533 @xmath600 and @xmath321 can fast decay into the @xmath1066 bosons and the @xmath1067 by @xmath87 couplings@xmath0 let @xmath1068 transform the @xmath1069 , then @xmath1070 transforms the @xmath1071 from the decaying process we see it is necessary @xmath1072 .
let @xmath1073 before the transition , it is necessary after transition that@xmath1074 this is the reheating process , after which , the @xmath136 get their masses , but the @xmath1066 bosons and the elementary @xmath1067 are still massless since @xmath87 is not broken . both @xmath25 and @xmath24 can exist in the form of plasma for a short period , because @xmath524 and @xmath523 must be very high in the initial period after reheating .
in the @xmath135 the space evolving equations are @xmath1075 after the reheating process , @xmath1076 thus , @xmath1077 reduce@xmath1078@xmath1079 r=-\frac{\eta}{2}\left ( \rho_{vg}+3p_{vg}\right ) r. \tag{114b}\ ] ] after @xmath524 and @xmath523 decline further , the @xmath136 exist in forms of nucleons , leptons and photons , and the @xmath113 will form @xmath1080 color single states whose masses are non - zero .
we take the order of time to be @xmath1081 after reheating process , all higgs particles have got their very big masses due to @xmath1082 , hence @xmath24 and @xmath25 can not transform into each other .
@xmath1083 because @xmath1084 all @xmath1085 must be the @xmath1086 single states and there is no @xmath1087 particle so that @xmath1088 so that @xmath1089 , because @xmath87 is not broken in the @xmath265 hence @xmath1090 must decrease as @xmath189 increases because @xmath1091 and @xmath840 . in the case , @xmath1092 and @xmath1093 reduces to @xmath1094@xmath1095 r , \tag{116a } \\ & \simeq-\frac{\eta}{2}\left ( \rho_{vm}+2\rho_{v\gamma}-\rho_{sm}\right ) r=-% \frac{\eta}{2}\left ( \rho_{vmg}+2\rho_{v\gamma g}\right ) r , \tag{116b}\end{aligned}\ ] ] where @xmath1096 and @xmath1097 are neglected and @xmath1098 is considered .
suppose when @xmath1099 ( e.g. @xmath1100 ) , @xmath1101 have formed , @xmath64 have decoupled , @xmath87 color single states have formed , @xmath1102 @xmath1103 and@xmath1104 after the photons decoupled , the number @xmath1105 of the @xmath1106 and the number @xmath1107 of the @xmath64 are invariant and @xmath1108 when space expansion@xmath0 the number @xmath1109 of the @xmath1086 single states is invariant as well .
let @xmath1110 then @xmath1111 @xmath1112 and @xmath1113 thus the gravitational mass density @xmath1114 in comoving coordinates is changeable,@xmath1115 where @xmath1116 and @xmath1117 is considered .
let @xmath1118 then considering the conjecture 3 , we have @xmath1119 thus when @xmath1120 @xmath1121 and @xmath1122 .
@xmath1123 is invariant and @xmath1124 is changeable .
for example , we may take @xmath1125 and @xmath1126 .
@xmath1127 will also change because of @xmath1128 and the reason in section 8.7 . from @xmath1129
we see when @xmath1130 @xmath1131 so that @xmath1132 from @xmath1133 @xmath1134 and @xmath1135 we have@xmath1136 , \tag{120a } \\
\overset{\cdot\cdot}{r } & = -\frac{\eta}{2}\left ( \rho_{vg}+\rho_{v\gamma } \right ) \notag \\ & = -\frac{\eta}{2}\frac{r^{3}\left ( t_{3}\right ) } { r^{2}}\rho_{vm}\left ( t_{3}\right ) \left [ \left ( 1-x\right ) + 2\frac{y_{3}r\left ( t_{3}\right ) } { r}% \right ] . \tag{120b}\end{aligned}\ ] ] in the case , @xmath1137 hence @xmath1138 so that @xmath1139 i.e. , space must expand with a deceleration in the period .
we analyze the process in detail as follows .
@xmath894 it is possible that there is @xmath1140 to satisfy @xmath1141 because @xmath498 and @xmath1142 when @xmath1143 when @xmath1144 @xmath1145 reduces to @xmath1146 in which @xmath1147 and @xmath1148 ( because in the @xmath39 now)@xmath0 there are two sorts of possibility when @xmath1149 .
a. @xmath1150 when @xmath1151 and @xmath1152 hence @xmath1153 ( @xmath496 as well)@xmath0 thus @xmath1154 will decrease as the conventional theory ( space contracts ) .
b. @xmath1155 decreases because space expands .
it is possible that there is @xmath1156to satisfy @xmath1157 and @xmath1158 @xmath1159 because of the conjecture 3 . in the case , @xmath1160 must be considered and @xmath491 reduces to @xmath876 in which @xmath1161 from @xmath876 we see that there are two sorts of possibility when @xmath1162 \a . if @xmath1163 @xmath943 and @xmath1164 hence space will contract with an acceleration . in the case , @xmath1165 \b . if @xmath1166 @xmath498 and @xmath1167 so that space continues to expands with a deceleration . in the case , @xmath1168 @xmath1169 @xmath1170 and @xmath1171 because of @xmath1172 and @xmath1173 there are the following three sorts of possibility , because of @xmath1174 .
@xmath1175 @xmath1176 so @xmath1177 thus @xmath1178 will decrease due to @xmath1179 , i.e. space will contract with an acceleration .
@xmath1180 @xmath1181 so @xmath1182thus @xmath1178 will continue to increase due to @xmath1183 , i.e. space will expand with an acceleration .
so @xmath1186thus @xmath1187 i.e. , space will be static . in the case
@xmath1018 , in fact , @xmath1188 which can still change although @xmath1189 .
this is because the major part of the repulsive potential energy between @xmath24 and @xmath25 will transform into @xmath1190 when @xmath1191 form ( see below ) .
consequently , after long time , @xmath1192 will strikingly decrease so that space will continue to expand with an acceleration .
the cases are different from the conventional theory . according to the conventional theory ,
when @xmath1193 there must be @xmath1194 there is no the case @xmath1195 this is because if @xmath1196 there must be @xmath1197 this is contradictory to the premise @xmath1198 @xmath1199 the other possibility is when @xmath1200 @xmath1201 have changed from @xmath1142 in to @xmath1202 in the case , space will continue to expand with an acceleration , and there is no stage in which @xmath1203 to sum up , we see that after inflation , space first expands with a deceleration , then it is possible that space is static for a period and finally expands with an acceleration up to now .
we consider the expanding process of space . after reheating , @xmath1204 based on the discussion about @xmath451
we may conclude when @xmath1205 say @xmath1206 so that @xmath1207@xmath1208 @xmath448 @xmath84 and @xmath48 can be taken as @xmath367 .
thus , in the @xmath171 , from @xmath1209 we have @xmath1210 , \tag{121a } \\
\overset{\cdot\cdot}{r } & = \frac{\eta}{2}\frac{c_{0}}{r^{2}}\left ( x-1 - 2% \frac{y_{0}r_{0}}{r}\right ) , \text { \ } \overset{\cdot\cdot}{a}=\frac{% h_{0}^{2}}{2}\left ( \frac{\omega_{gm0}}{a^{2}}-2\frac{\omega _ { v\gamma0}}{% a^{3}}\right ) , \tag{121b}\end{aligned}\ ] ] where @xmath1147 for simplicity , @xmath1211 @xmath1212 @xmath1213 @xmath1214 @xmath1215 @xmath1216 @xmath1217 it is obvious that when @xmath1218 @xmath498 and @xmath1219 , i.e. , space will expand with an acceleration .
it is difficult to uniformly describe the three evolving stages space since @xmath1221 is difficultly determined .
considering @xmath480 is applicable to all cases @xmath40 @xmath971 , and @xmath878 as a crude approximation for @xmath437 @xmath1222thereby @xmath1223 and @xmath971 @xmath1222thereby @xmath1224 we describe expansion of space by @xmath1225 in which @xmath1226 in fact , when @xmath1227 @xmath48 may be ignored . because when @xmath437 and @xmath160 space expands with a deceleration , and when @xmath438 and @xmath1228 space expands with an acceleration , the period of @xmath1229 can be approximately solved .
the result is qualitatively consistent with @xmath1230 as mentioned before , when @xmath1231 @xmath1232 so that space expands with a deceleration . from @xmath1233
we have @xmath1234 where @xmath1235 @xmath1236 taking @xmath1237 @xmath1238 @xmath1239}$ ] and @xmath1240 we get @xmath1241 @xmath1220 is shown by the curve @xmath739 in the figure 1 and describes evolution of the cosmos from @xmath1242 ago to now .
taking @xmath1243 @xmath1244 we get the @xmath1220 which is shown by the curve @xmath305 in the figure 1 and describes evolution of the cosmos from @xmath1245 ago to now .
provided @xmath1246 ( the condition @xmath1247 we can get a curve of @xmath1220 which describes evolution of the cosmological scale . from the two curves we see that the cosmos must undergo a period in which the cosmos expands with a deceleration in the past and present period in which the cosmos expands with an acceleration .
ignoring @xmath1248 @xmath1249 and taking @xmath1250 ,
we can reduce @xmath1251 to the corresponding formula @xmath1252 in ref . @xmath1253 $ ] considering @xmath1254 from @xmath1255 and @xmath1256 we have @xmath1257 @xmath1258 \sqrt{1-\left ( \omega_{gm0}-2\omega_{v\gamma0}\right )
z+\omega_{v\gamma 0}^{2}z^{2}}\ } , \tag{124}\end{aligned}\ ] ] where @xmath1259 is the redshift caused by @xmath189 increasing . provided @xmath1260 @xmath1261 is consistent with the corresponding formula @xmath1262 in ref .
$ ] ignoring @xmath1248 @xmath1249 we reduce @xmath1261 to@xmath1263 \sqrt{1-\omega_{gm0}z}\right\ } , \tag{125a}\ ] ] which is consistent with @xmath1264 in ref .
@xmath1253 $ ] .
approximating to @xmath1265 and @xmath1266 , we obtain @xmath1267 taking @xmath1268 @xmath1269 and @xmath1270 } $ ] and @xmath1271 from @xmath1272 we get the @xmath1273 relation which is shown by the curve @xmath305 in the figure 2 ; taking @xmath1274 @xmath1275 we get the @xmath1273 relation which is shown by the curve @xmath739 in the figure 2 .
the repulsion potential energy between @xmath25 and @xmath24 is determined by the distributing mode of @xmath24 and @xmath209 in @xmath171 @xmath136 with their masses can form @xmath1276 bodies , but @xmath24 which is @xmath87 color single states can not form any dumpling and must loosely distribute in space@xmath0 consequently , the huge repulsion potential energy must chiefly transform into the kinetic energy of @xmath87 color single states when the @xmath1276 bodies form or space expands .
in fact , when flat space expands @xmath1277 times , i.e. , @xmath1278 the repulsive - potential energy density @xmath1279 becomes @xmath1280 and@xmath1281 consider a system in flat space which is composed of a @xmath1282 with its mass @xmath1283 and a @xmath1086 single state with its mass @xmath1284 it is easy to get the rate @xmath1285 for static @xmath1283 and @xmath1286 at the initial moment ..
@xmath1287 because @xmath1288 , @xmath1289 space expansion is not the necessary condition to transform repulsive potential energy into kinetic energy .
supposing @xmath943 and some @xmath25 gather to a region and forms a galaxy , @xmath24 which is initially in the region must be repulsed away from the region and increases its kinetic energy .
the repulsive potential energy chiefly transforms into the kinetic energy of the @xmath24 in this case as well .
when @xmath524 is low , the @xmath87 color single states must loosely distribute in space or form @xmath1290 , i.e. , huge voids relative to @xmath209 @xmath798 and @xmath1291 in the @xmath39 . thus the @xmath87 symmetry still holds water and @xmath1292 hence all @xmath1066 particles and @xmath1067 are massless ( if the masses originate from only the couplings of the @xmath1067 with the @xmath92 fields ) .
when the temperature @xmath524 is high enough , all @xmath1066 particles and @xmath1067 must exist in plasma form .
when @xmath524 is low , all @xmath113 will exist in @xmath87 color - single state form ( conjecture 4 ) .
let @xmath305 , @xmath739 , @xmath311 , @xmath1293 , @xmath1294 denote the 5 sorts of colors .
a component of representation carries color @xmath1295 , @xmath1296 @xmath1297 , @xmath739 , @xmath311 , @xmath1293 , @xmath1298 @xmath1299 a component of representation carries color @xmath1300 a gauge boson carries colors @xmath1301 there are the following sorts of the @xmath87 color - single states .
2-fermion states : @xmath1302 or @xmath1303 , @xmath1299 when the spin of @xmath1302 or @xmath1304 is zero , we denote @xmath1305 or @xmath1306 by @xmath1307 . when the spin of @xmath1308 or @xmath1309 is 1 , we denote @xmath1310 or @xmath1311 by @xmath1312 .
analogously to @xmath1313 we have @xmath1314 ( because of color - magnetic interaction ) and can suppose @xmath1315 to be minimum in all @xmath87 color - single states .
in contrast with a given @xmath1316 @xmath1307 must be stable , because there is no the electroweak interaction in the case @xmath798 .
3-fermion states : @xmath1317 or @xmath1318 when the spin of @xmath1319 states is @xmath1320 , we denote them by @xmath1321 .
@xmath1322 is stable , and we can suppose @xmath1323 @xmath1277 denotes a given nucleon .
4-fermion states : @xmath1324 .
5-fermion states : @xmath1325 or @xmath1326 gauge boson single - states : @xmath1327 or @xmath1328 etc.@xmath142 @xmath1329 fermion - gauge boson single - states .
@xmath1330 , @xmath1331 etc .. of course , there are the @xmath1332 corresponding to the @xmath1086 single states above as well .
we can only qualitatively discuss the @xmath101-color single states by comparing the @xmath101-color single states with the @xmath164-color single states , because there is no mature theory of strict @xmath101 symmetry up till now .
the following inferences are obtained on the basis of the sameness of the @xmath101 and @xmath164 groups , hence they are reliable .
the characters of the @xmath87 color single states in the @xmath39 are as follows .
if the masses of the @xmath87 fermions only originate from the coupling of the @xmath87 fermions with the @xmath87 higgs fields , they all must be zero because @xmath1333 in the @xmath39 .
if such mass terms @xmath1334 are added to @xmath1335 ( @xmath1336 and @xmath1337)@xmath142 the masses of the @xmath87 fermions are non - zero . here
@xmath1338 and @xmath1339 are respectively the representations @xmath236 and .
the masses of all @xmath1340 color single states ( include gluon balls ) are non - zero .
consequently , we can determine that all @xmath101-color single states can get their suitable masses .
in contrast with @xmath164 color single states among which there are the electromagnetic and weak interactions due to the gauge group is @xmath1341 , there is no electroweak interaction among the @xmath87 color single states because @xmath87 is a simple group .
hence the @xmath87 color single states with the minimum masses must be stable .
. the interaction radius of the @xmath87 color single states must be finite and much smaller ( @xmath1342 ) than the radius of hydrogen atoms ( @xmath1343 ) .
this is because the interactions can come into being only by exchanging the @xmath87 color single states , and the masses of all @xmath87 color single states are non - zero .
thus we can approximately regard the @xmath87 color single states as ideal gas without collision . in a word ,
the @xmath87 color single states are analogous to @xmath1344 , but in contrast with the @xmath1344 , there is the repulsion between the @xmath111 color single states and @xmath209 from the characters of the @xmath87 color single states mentioned above , we will see that the @xmath87 color single states in the @xmath39 can not form clusters and must loosely distribute in space , but it is possible they form @xmath1290 as the neutrinos .
generalizing equations governing nonrelativistic fluid motion@xmath23}$ ] to present model , we have @xmath1345 @xmath1346 @xmath1347 @xmath1348 @xmath1349 in the @xmath135 where @xmath1350 is call the convective derivative .
we can produce the linearized equations of motion by collecting terms of first order in perturbations about a homogeneous background @xmath1351 etc .
let @xmath1352 considering @xmath1353 we get @xmath1354 @xmath1355 @xmath1356 @xmath1357 @xmath1358 defining the comoving spatial coordinates@xmath1359 we have @xmath1360 let @xmath1361 from @xmath1362 can get @xmath1363@xmath1364 it is necessary that @xmath1365 when @xmath1366 because there is only repulsion between @xmath24 and @xmath209 consequently , @xmath1367 according to the present model , there must be @xmath1368 because @xmath1369 @xmath1370 @xmath1371 @xmath1372 and @xmath1373 so that @xmath1374 and @xmath1375 there possibly is @xmath1376 when @xmath1377 , because @xmath87 color single states in the @xmath39 can be regarded as ideal gas without collision .
the ideal gas has the effect of free flux damping for clustering . ignoring @xmath1378 in @xmath1379 for @xmath1380 and @xmath1381 from @xmath1382 we get @xmath1383 where @xmath1384 @xmath1385 we see that @xmath1386 will exponentially grow for long - wavelength .
we can not get the result as @xmath1387 for @xmath1378 from @xmath1388 , because the velocity @xmath1389 of a @xmath87 color single state is invariant because there is no collision and is very big .
let the duration in which @xmath1390 be @xmath1391 the distance @xmath1392 to be damped out is @xmath1393 the perturbation whose size is less than @xmath1392 can not form .
thus @xmath87 color single states can only form superclusterings .
when @xmath1394 @xmath1395 is possible . consequently , the @xmath87 color single states possibly form superclusterings , because when @xmath1396 is large , the perturbation will slowly grow in power rules .
from the above mentioned , we see that the @xmath87 color single states must loosely distribute in space or form @xmath1397 , i.e. , huge @xmath1398 , in which @xmath1399 to sum up , because of the following two reasons , the perturbation @xmath1400 will grow faster or earlier than that determined by the conventional theory .
there is such a stage in which @xmath1401 and @xmath1402 so that @xmath1403 because the gravitational masses of @xmath24 and @xmath25 are opposite and @xmath1404 is changeable@xmath0 consequently , from @xmath1405 and @xmath1387 we see that @xmath1386 will exponentially grow for long - wavelength in the stage .
\2 . from @xmath1379 @xmath1406 and @xmath1387
we see that @xmath1407 can grow faster than that determined by @xmath1387 because @xmath1408 the origin is the repulsion between @xmath24 and @xmath25 .
if the whole universe is the world in the @xmath38 or the the world the -@xmath39 , the analysis above is correct . but according to the present model , as mentioned before , there are the two sorts of symmetry breaking which are different in essence . thus it is possible that there are different breaking forms in different regions of the universe .
as mentioned above , @xmath25 and @xmath24 are symmetric and mutually repulsive , @xmath25 in the @xmath39 can form @xmath65 and @xmath1409 in the @xmath38 can form @xmath116 .
from this we present a new cosmic model as follows .
the universe is composed of infinite @xmath1410 islands and and @xmath1411 islands . @xmath258 and @xmath1412 in @xmath1410 islands , and @xmath1413 and @xmath1414 in the the @xmath1411 islands .
a @xmath1410 island or a @xmath1411 island must be finite .
there must be a transitional region ( @xmath1415 ) between a @xmath1410 island and a @xmath1411 island . in the @xmath1416 it is necessary
@xmath1417 and @xmath1418 change from @xmath1419 and @xmath258 into @xmath1420 and @xmath1421 respectively .
consequently , the expectation values @xmath1422 and @xmath1423 inside the @xmath1415 must satisfy @xmath1424 there must be only @xmath1411 islands neighboring a @xmath1410 island .
this is because that if two @xmath1410 islands are neighboring , they must form one new larger @xmath1410 island .
it is obvious that if there is only one sort of breaking , it is impossible that such cosmic islands exist .
based on the following reasons , the probability is very little that a @xmath1425 accepts messages from a @xmath1410 island .
1**. * * the probability must be very small that a @xmath144 ( a quark , a lepton or a photon ) in the @xmath1410 island comes into the @xmath1411 island , because a @xmath144 in a @xmath1410 island is @xmath87 non - color single state .
if a @xmath144 comes into the @xmath1411 island , it would still be non - color single state and would get very big mass .
this is impossible due to color confinement .
but a bound state of the @xmath1426 e.g. @xmath1427 which is a color single state in both @xmath39 and @xmath49 possibly comes into the @xmath1411 island .
it is hardly funded by a @xmath1428 because the boson @xmath1427 is a @xmath1429 single state in the @xmath1411 island which is a particle of dark energy . .
. the probability must be very small that a @xmath304 in the @xmath1410 island come into the @xmath1411 island as well@xmath0 the @xmath304 ( a fermion or a gauge boson ) in the @xmath1410 island must be massless . if the massless @xmath304 comes into the @xmath1411 island , it would get its mass .
thus its static mass will change from @xmath1430 to @xmath1431 so that it must suffer a strong - repulsive interaction , hence it hardly comes into the @xmath1411 island .
higgs particles in the @xmath1410 island must decay fast into fermions or gauge bosons , hence they can not come to the @xmath1411 island .
@xmath1432 is so big that the probability through which a particle passes is very small .. the probability must be very small that particles leave the @xmath1411 island because of the same reasons . a @xmath1411 islands and a @xmath1410 island can influence on each other by the higgs potential in the @xmath1415 between both .
as a consequence a @xmath1428 in the @xmath1411 island can regard the @xmath1433 island as the whole cosmos .
it is possible that some cosmic islands are forming or expanding , and the other cosmic islands are contracting .
thus , according to the present model the cosmos as a whole is infinite and its properties are always unchanging , and there is no starting point or end of time . hydrogen spectrum is @xmath1434 where @xmath1286 is the mass of an electron , and @xmath1283 is the mass of a proton . according the unified model , @xmath1435 , the mass of a quark @xmath1436 where @xmath1437 and @xmath1438 are the expectation values of the higgs fields coupling with the electron and the quark , respectively .
@xmath1439 if there are some galaxies inside a @xmath1416 from @xmath1440 we see that the mass @xmath1441 of an electron and the mass @xmath1442 of a proton inside the @xmath1415 must be @xmath1443 thus we have @xmath1444 the sort of red - shifts is called @xmath1445 @xmath1446 . the mass redshift is essentially different from the cosmological red - shift mentioned before .
thus , the photons coming from the star in a @xmath1415 must have larger red - shift than that determined by the hubble formula at the same distance .
thereby we guess that some quasars are just the galaxies in the @xmath1415 of our cosmos island and they have the mass redshifts .
the fine - structure constant is considered to change based on the redshifts of some quasars according to the conventional theory .
in contrast with the conventional , we consider that the mass of electrons possibly changes , but is not the fine - structure constant to change .
an ordinary @xmath307 and a @xmath1447 can be neighboring , because a @xmath1448 must be neighboring to an ordinary region .
hot dark matter , e.g. the neutrinos , can form huge voids , but can not explain evolution of structure with middle and small scales .
cold dark matter can explain evolution of structure with the middle and the small scales , but can not explain the huge voids .
the problem of huge voids remains unsettled .
according the present model , huge @xmath1398 in the @xmath39 are , in fact , superclusterings of @xmath113 .
the huge @xmath1398 not empty , and in which there is @xmath24 with a bigger density , @xmath1449 @xmath1450 and @xmath1451 here @xmath1452 and @xmath1453 denote the densities in the huge @xmath1398 . because there is the repulsion between @xmath148 and @xmath25 and
there is the gravity among @xmath1426 a huge @xmath1454 can form .
the forming process is analogous to that of the neutrino - superclusterings .
the characters of such a huge @xmath1454 are as follows .
a @xmath1454 must be huge , because there is no other interaction among the @xmath87 color single states except the gravity and the masses of the @xmath111 color single states are very small .
when @xmath64 pass through such a huge @xmath1454 , the @xmath64 must suffer repulsion from @xmath1452 and are scattered by @xmath1455 as they pass through a huge concave lens .
consequently , the galaxies behind the huge @xmath1454 seem to be darker and more remote .
both density of matter and density of dark matter in huge voids must be more lower than those predicted by the conventional theory .
specifically , the density of hydrogen in the huge voids must be more less than that predicted by the conventional theory .
right or mistake of the predict can be confirmed by the observation of distribution of hydrogen .
it is seen that the present model can well explain the characters of some huge voids .
this is a decisive prediction which distinguishes the present model from other models .
there must be @xmath24 between two @xmath65 distant enough , hence the gravitation between the two @xmath65 must be less than that predicted by the conventional theory due to the repulsion between @xmath24 and @xmath1456 .
when the distance between the two @xmath65 is small , the gravitation is not influenced by @xmath1457 because @xmath28 must be small when @xmath129 is big .
let there be a @xmath70 hole with its mass and density to be big enough so that the critical temperature @xmath34 can be reached in the @xmath265 if its mass is so big that its temperature @xmath1458 since the black hole contracts by its self - gravitation , then the expectation values of the higgs fields inside the @xmath70 hole will change from @xmath1459 and @xmath798 into @xmath1460 .
consequently , inflation must occur .
after inflation , the highest symmetry will transit into the @xmath39 or the @xmath38 .
no matter which breaking appears , the energy of the black hole must transform into both @xmath1069 and @xmath1190 .
thus , a @xmath1428 will find that the black hole disappears and a white hole appears . in the process ,
part of @xmath1461 transforms into @xmath1071 a @xmath1428 will consider the energy not to be conservational because he can not detect @xmath148 except by repulsion . * * * * the * * * * transformation * * * * of black holes is different from the hawking radiation .
this is the transformation of the vacuum expectation values of the higgs fields .
there is no contradiction between the transformation and the hawking radiation or another quantum effect , because both describe different processes and based on different conditions ( the density and mass of the black hole must be large enough so that its temperature @xmath1462 ) . according to the present model , there still are the hawking radiation or other quantum effects of black holes .
in fact , the universe is just a huge black hold .
the universe can transform from the @xmath38 into the @xmath171 because of its contraction or expansion .
this transformation is not quantum effects .
let there be a @xmath1411 island neighboring on a @xmath1410 island .
it is possible that the @xmath1411 island transforms into a @xmath1410 island after it contracted . in this case
, a @xmath1463 in the @xmath1410 island must observe a very huge white hole .
as mentioned before , in the @xmath39 , when @xmath1464 , space will expand with an acceleration .
it seems that the universe will always expand with an acceleration .
this is impossible . in the expanding process of the universe
, galaxies can be continue to exist and @xmath25 will gather so that a huge @xmath70 hole with its mass and density big enough can be formed after long enough time , because the repulsion between s - matter and v - matter and the gravitation among @xmath136 .
after the temperature of the black hole reaches the critical temperature @xmath34 , the expectation values of all higgs fields will tend to zero and the local space will expand . in the case , both @xmath39 and @xmath1023 are possibly realized .
if the density of @xmath25 around the huge @xmath70 hole is little enough , the @xmath38 will is realized .
consequently the @xmath171 transforms into the @xmath38 and the @xmath1465 transforms the @xmath1466 . if the density of @xmath25 around the huge @xmath70 hole is very large , the @xmath39 will is realized .
this is because the transformation of symmetry breaking must cause the transformation of existing form of matter .
@xmath1467 whose temperature is low around the huge @xmath70 hole will prevent the transformation of the @xmath39 into the @xmath123 after the average density of v - matter is little because space expands and a huge @xmath70 hole with its mass and density big enough formed , the transformation of the @xmath39 into the @xmath38 can occur .
it is seen that in the both cases that the universe is little enough because space contracts or is large enough because space expands , it will occur that one sort of breaking transforms into the other and the world transforms the other world .
the effective cosmological constant @xmath1470 the conventional theory can explain evolution with a small @xmath1471 such a small @xmath10 can not be derived from a elementary theory . according to the conventional quantum field theory , @xmath1472 @xmath1473 in fact , @xmath9 is divergence . according to the conventional gravitational theory , @xmath1474 ( @xmath1475 ) .
consequently , the issue of the cosmological constant appears .
@xmath20 can be obtained by some supersymmetric model , but it is not a necessary result .
on the other hand , the particles predicted by the supersymmetric theory have not been found , although their masses are not large .
@xmath20 is a necessary result of our quantum field theory without divergence@xmath21}.$ ] in this theory , @xmath20 is naturally obtained without normal order of operators , there is no divergence of loop corrections , and dumpling dark matter is predicted@xmath22}.$ ] according to the present model , @xmath1476 although @xmath1477 is still very big according to the conventional quantum field theory .
proof : @xmath1476 although @xmath1478 applying the conventional quantum field theory to the present model , we have @xmath1479 both @xmath1480 and @xmath1481 must be two constants . according to the conjecture 1 , @xmath113 and @xmath136
are strictly symmetric in essence .
hence @xmath1482 according to the conjecture 1 , the gravitational mass of @xmath24 is opposite to that of @xmath25 .
i.e. , @xmath1483 when @xmath1484 .
hence we have@xmath1485 thus , there is no the fine tuning problem , even if @xmath37 .
@xmath157 is a necessary inference , because evolution of the cosmos can be explained by the present model without @xmath10 .
consequently , although @xmath1486 we have still @xmath1487 this is an direct inference of the present model , and independent of a quantum field theory .
thus , the cosmological constant issue has been solved . for the vacuum state in the @xmath38 or the @xmath39 ,
the einstein field equation is reduced to @xmath1488 this is a reasonable result .
the @xmath72 dark matter model@xmath22}$ ] is a necessary inference of the quantum field theory without divergence .
the @xmath72 dark matter model is similar with the mirror dark matter model . according to the mirror dark matter model
, it is impossible that the density of matter is equal to that of dark matter in order to explain the primordial nucleosynthesis and @xmath71 .
this is too difficultly understood , because matter and mirror matter are symmetric and both can transform from one into another when temperature is high enough .
in contrast with the mirror dark matter model , according to the @xmath72 matter model , @xmath1489 ( ordinary matter ) and @xmath1490 ( dark matter ) are not only symmetric , but also @xmath1491 if the total density of matter and dark energy is @xmath1492 and the ratio of the density of dark energy @xmath1493 to @xmath1494 is @xmath2 , @xmath1495 considering only @xmath1496 matter is visible and the radio of the density @xmath1497 to @xmath1492 is @xmath1498 dark matter can be classified into the following three sorts : invisible @xmath1489 for a time whose density is @xmath1499 invisible @xmath1500 matter whose density is @xmath1501 and invisible @xmath1502 matter whose density is @xmath1503 @xmath1504 can form dark galaxies and can be observed , @xmath1505 and @xmath1506 can not form any dumpling .
the @xmath1507 is possibly observed in future . @xmath1508 and @xmath1509 can transform from one into another when temperature is high enough . according to the present cosmological model , there are @xmath24 and @xmath141 which are symmetric in principle .
after symmetry spontaneously breaking , @xmath24 and @xmath25 are no longer symmetric . in the @xmath171
, @xmath25 corresponds visible matter and dark matter , @xmath1409 corresponds to so - called dark energy and must exist in the form of the @xmath87 color single states .
the masses of all color single states are non - zero .
but the masses are different from each other .
some masses of color single states ( their density are denoted by @xmath1510 ) are possibly less than @xmath1511 so that they may be ignored when temperature @xmath1512 and the others ( their density are denoted by @xmath418 ) are larger than @xmath1513 .
@xmath418 and @xmath1510 can be determined based on observation . when the @xmath72 dark matter model and the present cosmological model are simultaneously considered , the primordial nucleosynthesis and cmbr can be explained .
according to the @xmath72 matter model@xmath22}$ ] which is similar with the mirror dark matter model , the mechanism of primordial nucleosynthesis is the same as the conventional theory .
but the mechanism of space expansion of the present model is different from that of the conventional theory . for short ,
we consider only influence of space expansion on the primordial nucleosynthesis and @xmath1514 the primordial helium abundance @xmath1515 is determined by @xmath1516},$ ] @xmath1517 , \tag{151a } \\
n_{n}/n_{p } & = exp\left ( -\bigtriangleup m / kt_{1}\right ) , \text { } % \bigtriangleup m = m_{n}-m_{p } , \tag{151b}\end{aligned}\ ] ] where @xmath1518 is the neutron - proton ratio in the unit comoving volume at the freeze - out temperature @xmath1519 @xmath1520 is determined by @xmath1521 and @xmath1522 here .@xmath1523 is the interaction rate experienced by a particle . as mentioned above , @xmath1524 and in the @xmath135 @xmath1525 and in general @xmath1526 since the masses of all @xmath1527 single states are non - zero , it is possible there are some @xmath1527 single states with their masses @xmath1528 and the others with their masses @xmath1529 and @xmath1530 thus , when @xmath1531 @xmath1529 may be ignored so that @xmath1532 and when @xmath1533 , @xmath1534 considering @xmath1535 @xmath1536 @xmath1537 and @xmath1538 @xmath1539 @xmath1540 and @xmath245 , are regarded as zero , @xmath1541 @xmath1542 and @xmath1543 so that @xmath1544 when @xmath1545 @xmath69 may be ignored and the universe is dominated by @xmath1510 and @xmath1546 we have @xmath1547 \notag \\ & = \eta\left [ \left ( \rho_{vfm}+\rho_{vwm}+\rho_{vf\gamma}+\rho_{vw\gamma } \left ( t_{v1}\right ) \right ) -\left ( \rho_{sm}^{\prime}+\rho_{sl}\left ( t_{s1}\right ) \right ) \right ] \notag \\ & \simeq\eta\left [ \rho_{vf\gamma}+\left ( \rho_{vw\gamma}\left ( t_{v1}\right ) -\rho_{sl}\left ( t_{s1}\right )
\right ) \right ] \notag \\ & = \eta\left [ \frac{\pi^{2}}{30}g^{\ast}t_{v1}^{4}+\left ( \frac{\pi^{2}}{30}% g^{\ast}t_{v1}^{4}-\rho_{sl}\left ( t_{s1}\right ) \right ) \right ] .
\tag{152}\end{aligned}\]]@xmath1548 where @xmath1549 is the @xmath143 decoupling temperature ( see the following ) . @xmath1510 and @xmath1550 are two parameters which should be determined by observations .
@xmath1551 can be so chosen that @xmath1552 for the freeze - out temperature @xmath1553 is determined by @xmath1554 and @xmath1555 .
@xmath1554 in the present model is the same as that in the conventional theory .
@xmath1556 is the same as that in the conventional theory as well .
hence the present model can explain the primordial nucleosynthesis and @xmath1515 as the conventional theory .
for example , taking the rough approximation @xmath1557 considering @xmath1558 we get the equation to determine @xmath1553 @xmath1559 ^{1/2}\simeq\left ( \eta\frac{\pi^{2}}{30}g_{v}^{% \ast}t_{v1}^{4}\right ) ^{1/2}. \tag{155}\ ] ] this result and @xmath1518 and @xmath1515 corresponding to this are the same as those of the conventional theory@xmath1560}$ ] .
it is seen that although ordinary matter and dark matter are completely symmetric so that @xmath1561 , we can still obtained the result of conventional theory , provided the f - w dark matter model and the present model are simultaneously considered .
this is different from the mirror dark matter model .
it is the same as the conventional theory that there are the inflation and big bang processes in the present model .
hence there must be the cosmic microwave background radiation @xmath1564 the recombination temperature of the present model is the same as that of the conventional theory , because it is independent of s - matter in the v - breaking . from the following formulas@xmath1560}$ ] we can determine the recombination temperature @xmath1565@xmath1566 where @xmath1567 @xmath1568 to be number density and @xmath1569 is the ionization potential of hydrogen .
taking @xmath1570 and @xmath1571 considering @xmath1572 , @xmath1573 @xmath1574 @xmath1575 and @xmath1576 we have@xmath1560}$ ] @xmath1577 in contrast with the conventional theory , according to the @xmath72 dark matter model , @xmath1489 and @xmath1490 are completely symmetric so that not only there are @xmath1579 ( ordinary photons ) , but also @xmath1580 ( dark - matter photons ) .
@xmath1581 @xmath1582 from this we can estimate the temperature @xmath1578 of matter - radiation equality as follows . when only the photons and the three species of neutrinos are considered ( here the three species of neutrinos are regarded as massless ) .
, we have@xmath1583}$]@xmath1584@xmath1585 thus , considering @xmath1586 we get@xmath1587 \notag \\ & = 6.7425\times10^{-51}gev^{4}. \tag{158}\end{aligned}\ ] ] observation shows that the total density of matter and dark matter is @xmath1588 according to the @xmath72 model , this implies @xmath1589 where @xmath1590 @xmath1591 and@xmath1592 from @xmath1593 @xmath1594 can be determined as follows @xmath1595 ^{3}=\rho_{v\gamma0}\left [ r_{0}/r_{eq}\left ( t_{veq}\right ) \right ] ^{4}=\rho_{v\gamma eq}\left ( t_{veq}\right ) , \text { } \tag{160a } \\ \frac{r_{0}}{r_{eq } } & = \frac{t_{veq}}{t_{v0}}=\frac{\rho_{vm0}}{% \rho_{v\gamma0}}=\frac{\omega_{0}\rho_{c}}{2\rho_{vfr0}}=\left ( 1+z_{eq}\right ) = 1370,\text { } \notag \\
t_{veq } & = 0.32ev,\text { when } h=0.65 , \tag{160b }
\\ \frac{r_{0}}{r_{eq } } & = \frac{t_{veq}}{t_{v0}}=\frac{\rho_{vm0}}{% \rho_{v\gamma0}}=1272\text { , } \notag \\ t_{veq } & = 0.295ev = t_{rec},\text { when } h=0.624 .
\tag{160c}\end{aligned}\ ] ] according to the present model , @xmath1596 is a crude approximation , @xmath1597 is changeable not only because of @xmath1598 @xmath1599 and @xmath512 but also because of @xmath1600 according to the conventional theory , @xmath1601 , @xmath1602 hence if @xmath1603 and @xmath1604,@xmath1605 it is seen that @xmath1606 is remarkably different from @xmath1607 when @xmath1608 the universe is not matter - dominated .
let when @xmath1609 @xmath1610 @xmath1611 and the masses of all s - color single states can not be ignored , i.e. @xmath1612 considering @xmath1613 ignoring the transformation @xmath129 into @xmath958 we can rewrite @xmath1614 @xmath1615 and @xmath418 as follows .
@xmath1616 ignoring @xmath1617 considering @xmath1618 @xmath1619 and @xmath1620we get@xmath1621 this is the only deference between the present model and the conventional theory in order to determine the decoupling temperature .
we have the same interaction rate experienced by one photon and saha equation as the conventional theory@xmath1560}$ ] @xmath1622@xmath1623 where @xmath1624 is considered . only for comparison of the present model with the conventional theory
, we use the same equation @xmath1625 and the same crude approximation @xmath1626 to evaluate the decoupling temperature .
taking @xmath1627 , and @xmath1628 from @xmath1629 we have@xmath1630 substituting @xmath1631 into @xmath1625 , we get@xmath1632 taking @xmath1633 and @xmath1634 , we get @xmath1635 taking @xmath1633 and @xmath1636 , we get@xmath1637 taking place of @xmath1638 by the equation in the conventional theory@xmath1639 we have@xmath1640@xmath1641 @xmath1642 is not susceptible for change of @xmath1643 in the scope @xmath1644 considering @xmath1645 we have @xmath1646 it is seen from @xmath1647 that in the decoupling stage , @xmath1648 and the universe is not matter - dominated .
this is different from the conventional theory .
the first peak of the @xmath73 power spectra is the evidence of existence of the elementary wave .
the elementary wave began at reheating ( @xmath1649 ) and ended at recombination after @xmath1650 years ( @xmath1651 ) according to the conventional theory .
let the temperature of reheating is @xmath1652
. in the period @xmath79 descends into @xmath1565 baryons exist in plasma .
the sound speed of plasma is @xmath1653 let the period in which @xmath79 descends into @xmath1563 is @xmath1654 according to the present model and that according to the conventional theory is @xmath1655 there must be @xmath1656 the reasons are as follows . that @xmath1657 holds implies @xmath1658 where @xmath1659 and @xmath1660 are the mass density in the conventional theory @xmath1661 and @xmath1662 etc are considered .
it is necessary that when @xmath1663 @xmath1657 or @xmath1664 still holds@xmath0 this is because @xmath1665 @xmath1666 and @xmath1667 and @xmath1668 when @xmath1669 when @xmath1670 it is necessary that @xmath1671 .
this is because @xmath1672 still holds , but @xmath1673 ( due to @xmath1674 ) when @xmath1675 hence when @xmath1670 @xmath1676 this implies @xmath1677 to hold . on the other hand ,
the sound speed @xmath1678 is determined by only @xmath1679 and @xmath129 or @xmath523 and @xmath1680 and is independent of @xmath1681 hence @xmath1682 when @xmath79 descends into @xmath1563 , here @xmath1683 is the sound speed in the conventional theory when @xmath1684 .
thus when temperature descends from @xmath79 descends into @xmath1565 the propagating distance of the elementary sound wave must be longer according to the present model than that according to the conventional theory .
based on @xmath1685 space - time is flat or @xmath98 as the conventional theory , but based on @xmath1686 and @xmath1687 space - time is open or @xmath84 as the the present model .
this is consistent with the present model according which @xmath84 when @xmath1688 in present stage .
we will discuss the issue in detail in the following paper .
the new conjectures are proposed that there are @xmath24 and @xmath25 which are symmetric , whose gravitational mass densities are opposite to each other and whose energies are all positive
. both can transform from one to another when temperature @xmath889 .
consequently there is no singularity in the model and the cosmological constant @xmath1689 is determined although the energy density of the vacuum state is still large and there is no the fine tuning problem , even if @xmath1690 .
there are two sorts of breaking modes , i.e. the @xmath38 and the @xmath171 . in the @xmath39 @xmath117 symmetry
is broken into @xmath1691 for @xmath136 and @xmath87 symmetry is still strictly kept for @xmath522 consequently @xmath136 get their masses determined by the @xmath101 @xmath103 and form the @xmath65 etc . , while @xmath113 are massless and form @xmath87 color - single states which loosely distribute in space and can cause space to expand with an acceleration@xmath0 in contrast with the dark energy , the gravitational masses of @xmath24 is negative in the @xmath39 .
based on the present model@xmath142 the space evolving process is as follows .
firstly , in the @xmath38 , @xmath1692 and @xmath160 hence space contracts and @xmath524 rises@xmath0 when @xmath1693 , @xmath1694 both @xmath117 and @xmath111 symmetries is strictly kept ( this state have the highest symmetry ) , and the masses of all particles originating from the couplings with the higgs fields are zero so that @xmath28 and @xmath129 can transform from one into another . as a consequence @xmath26 @xmath1695 and @xmath1696 so that inflation must occur . after the inflation , the phase transition of the vacuum , i.e. the reheating process , occurs . after the reheating process , this state with the highest symmetry transits to the state with the @xmath39 .
space in the @xmath171 have three evolving stages .
space firstly expands with a deceleration because @xmath1697 and @xmath675 ; then comes to static because @xmath1698 and @xmath98 ; and finally expands with an acceleration up to now because @xmath1699 and @xmath84 .
the results above is still valid when @xmath1700 and @xmath1701 it is seen that the world in the @xmath38 and the world in the @xmath39 can transform from one into another .
the evolving process of the cosmos is different from that determined by the conventional theory .
@xmath54 changes from @xmath55 into @xmath56 is obtained .
there are the critical temperature @xmath34 , the highest temperature @xmath57 , the least scale @xmath58 and the largest energy density @xmath59 in the universe . @xmath58 and @xmath34 are two new important constants , @xmath52 and @xmath60 are determined by @xmath61 .
generalizing equations governing nonrelativistic fluid motion to the present model , the equations of @xmath88 formation have been derived . according to the equations , galaxies formed earlier and more easily than the conventional theory .
two guesses have been presented .
the universe is composed of infinite @xmath1702 islands and @xmath1411 islands .
some huge redshifts ( e.g. the big redshifts of quasi - stellar objects ) are explained as the mass redshifts which is caused by @xmath1703 here @xmath1704 is the mass of an electron in a transiting region and @xmath69 is the given mass of an electron in the normal region .
three new predicts have been given .
huge @xmath1398 in the @xmath39 are not empty , but are superclusterings of @xmath113 , there must be @xmath1409 with its density @xmath63 in the @xmath1398 .
it is possible that huge voids is equivalent to a huge concave lens .
the density of hydrogen and the density of helium in the huge voids predicted by the present model must be more less than that predicted by the conventional theory .
it is possible that a @xmath70 hole with its big enough mass and density can transform into a huge white hole by its self - gravitation .
the first peak of the @xmath73 power spectra is the evidence of existence of the elementary wave
. the elementary wave began at reheating and ended at recombination after @xmath74 years according to the conventional theory . but according to the present model , it is necessary that @xmath75 because the sound speed @xmath76 and @xmath1705 consequently , space - time is open or @xmath84 according to the present model s - h .
chen , ` quantum field theory without divergence a ' , ( 2002a ) hep - th/0203220 ; s - h , chen , ` significance of negative energy state in quantum field theory a ' ( 2002b ) hep - th/0203230 ; s - h , chen , ` quantum field theory : new research ' , o. kovras editor , nova science publishers , inc .
( 2005a ) p103 - 170 .
s - h , chen , ` a possible candidate for dark matter ' , ( 2001 ) hep - th/0103234 ; s - h , chen , ` progress in dark matter research ' editor : j. val blain , ( 2005b ) pp.65 - 72 .
nova science publishers , inc .
arxiv : 1001.4221 .
s - h chen , ` discussion of a possible universal model without singularity ' , ( 2009 ) arxiv .
0908.1495 ; s - h , chen , ` a possible universal model without singularity and its explanation for evolution of the universe ' , ( 2006 ) hep - th/0611283 .
m chaichian and n.f.nelipa , ` introduction to gauge field theories ' springer - verlag berlin heidelberg , ( 1984 ) p269 ; g. g. ross , ` grand unified theories ' , the benjamin / cummings publishing company , inc , ( 1984 ) p177 - 183 . | the new conjectures are proposed that there are s - matter and v - matter which are symmetric and whose gravitational masses are opposite to each other .
there are two sorts of symmetry breaking modes called v - breaking and s - breaking , respectively . in the v - breaking ,
v - particles get their masses and form v - galaxies etc . , while s - fermions and s - gauge bosons are still massless and form s - su(5 ) color - single states which loosely distribute in space and cause space to expand with an acceleration@xmath0 the curvature factor k in the rw metric is regarded as a function of gravitational mass density in the comoving coordinates . in the s - breaking
, space can contract and causes temperature to rise .
when it reaches the critical temperature , masses of all particles are zero so that s - particles and v - particles transform from one to another so that the gravitational mass density is zero .
consequently space inflation occurs . after the reheating process
, the state with the highest symmetry transits into the v - breaking@xmath0 in the v - breaking , space first expands with a deceleration ; then comes to static , and finally expands with an acceleration up to now .
the cosmological constant is determined to be zero , although the energy density of the vacuum state is still large .
there is no space - time singularity in the present model .
there are the critical temperature , the highest temperature and the least scale in the universe .
a formula is obtained which well describes the relation between a luminous distance and its redshift .
the equations of structure formation have been derived , on which , galaxies can form earlier than the conventional theory .
the universe is composed of infinite s - cosmic islands and v - cosmic islands .
a huge v - voids is not empty , in which there must be s - matter with its bigger density , and which is equivalent to a huge concave lens .
the densities of hydrogen and helium in the huge voids must be more less than that predicted by the conventional theory .
the gravitation between two galaxies distant enough will be less than that predicted by the conventional theory .
it is possible that a v - black hole with its big enough mass and density can transform into a huge white hole by its self - gravitation .
nucleosynthesis and cmbr are explained .
space - time is open or k0 according to the present model .
w can change from w 0 to w -1 . |
the idea of quantum annealing ( qa ) is a late offspring of the celebrated _ simulated thermal annealing _ by kirkpatrick _
et al . _ @xcite . in simulated annealing
, the problem of minimizing a certain cost ( or energy ) function in a large configuration space is tackled by the introduction of a fictitious temperature , which is slowly lowered in the course of a monte carlo or molecular dynamics simulation @xcite .
this device allows an exploration of the configuration space of the problem at hand , effectively avoiding trapping at unfavorable local minima through thermal hopping above energy barriers .
it makes for a very robust and effective minimization tool , often much more effective than standard , gradient - based , minimization methods .
an elegant and fascinating alternative to such a classical simulated annealing ( ca ) consists in helping the system escape the local minima through _ quantum mechanics _
, by tunneling through the barriers rather than thermally overcoming them @xcite .
experimental evidence in disordered ising ferromagnets subject to transverse magnetic fields showed that this strategy is not only feasible but presumably winning in certain cases @xcite .
in essence , in quantum annealing one supplements the classical energy function , let us denote it by @xmath0 , with a suitable _ time - dependent _ quantum kinetic term , @xmath1 , which is initially very large , for @xmath2 , then gradually reduced to zero in a time @xmath3 .
the quantum state of the system @xmath4 , initially prepared in the fully quantum ground state @xmath5 of @xmath6 , evolves according to the scrdinger equation @xmath7 |\psi(t)\rangle \;,\ ] ] to reach a final state @xmath8 .
a crucial basic question is then how the residual energy @xmath9 , decreases for increasing @xmath3 . here
@xmath10 is the absolute minimum of @xmath0 , and @xmath11 is the average energy attained by the system after evolving for a time @xmath3 , @xmath12 , generally speaking , this question has to do with the _ adiabaticity _ of the quantum evolution , i.e. , whether the system is able , for sufficiently slow annealing ( sufficiently long @xmath3 ) , to follow the instantaneous ground state of @xmath13 , for a judiciously chosen @xmath1 .
the fictitious kinetic energy @xmath1 can be chosen quite freely , with the only requirement of being reasonably easy to implement .
for this reason , this approach has also been called quantum adiabatic evolution @xcite . at the level of practical implementations on an ordinary ( classical ) computer , the task of following the time - dependent schrdinger evolution in eq .
[ schrodinger : eqn ] is clearly feasible only for toy models with a sufficiently manageable hilbert space @xcite .
actual optimization problems of practical interest usually involve astronomically large hilbert spaces , a fact that calls for alternative stochastic ( quantum monte carlo ) approaches .
these qmc techniques , in turn , are usually suitable to using _ imaginary time _ quantum evolution , where the @xmath14 in eq .
[ schrodinger : eqn ] is replaced by @xmath15 .
one of the questions we will try to address in the present paper will be , therefore , if working with an imaginary - time schrdinger evolution changes the quantum adiabatic evolution approach in any essential way .
we will argue that , as far as annealing is concerned , imaginary - time is essentially equivalent to real - time , and , as a matter of fact , can be quantitatively better .
alternatively , a number of recent theoretical papers have applied path - integral monte carlo strategies to qa .
a certain success has been obtained in several optimization problems , such as the folding of off - lattice polymer models @xcite , the random ising model ground state problem @xcite , and the traveling salesman problem @xcite .
it is fair to say , however , that there is no general theory predicting the performance of a qa algorithm , in particular correlating it with the energy landscape of the given optimization problem .
this is a quite unpleasant situation , in view of the fact that it is a - priori not obvious or guaranteed that a qa approach should do better than , for instance , ca .
indeed , for the interesting case of boolean satisfiability problems more precisely , a prototypical np - complete problem such as 3-sat recent attempts in our group showed that path - integral monte carlo annealing may perform definitely _ worse _ than simple ca @xcite .
evidently , the performance of qa over ca depends in detail on the energy landscape of the problem at hand , in particular on the nature and type of barriers separating the different local minima , a problem about which very little is known in many practical interesting cases @xcite .
it makes sense therefore to move one step back and concentrate attention on the simplest textbook problems where the energy landscape is well under control : essentially , one - dimensional potentials , starting from a double well potential , the simplest form of barrier . on these well controlled landscapes we can carry out a detailed and exhaustive comparison between quantum adiabatic schrdinger evolution , both in real and in imaginary time , and its classical deterministic counterpart , i.e. , fokker - planck evolution .
the rest of the paper is organized as follows . in sec .
[ annealing_def : sec ] we define the problem we want to tackle , i.e. , comparing fokker - planck annealing to schrdinger annealing . in sec .
[ tls : sec ] we consider in detail the case of a double - well barrier . in sec .
[ washboard : sec ] we move to the case of a potential with many minima , but no disorder , were the behavior of classical and quantum annealing is remarkably different . in sec .
[ disorder : sec ] the crucial role of disorder is discussed .
finally , sec .
[ summary : sec ] contains a summary of the results found and a few concluding remarks .
details of the calculations are contained in the appendices .
suppose we are given a potential @xmath16 , ( with @xmath17 a cartesian vector of arbitrary dimension ) , of which we need to determine the absolute minimum ( @xmath18,@xmath19 ) .
assume generally a situation in which a steepest - descent approach , i.e. , the strategy of following the gradient of @xmath20 , would lead to trapping into one of the many local minima of @xmath20 , and would thus not work .
classically , as an obvious generalization of a steepest - descent approach , one could imagine of performing a stochastic ( markov ) dynamics in x - space according to a langevin equation @xmath21 where the strength of the noise term @xmath22 is controlled by the squared correlations @xmath23 , with @xmath24 .
both @xmath25 and @xmath26 with dimensions of a diffusion constant and of a friction coefficient and related , respectively , to fluctuations and dissipation in the system are temperature dependent quantities which can be chosen , for the present optimization purpose , with a certain freedom .
the only obvious constraint is in fact that the correct thermodynamical averages should be recovered from the langevin dynamics only if @xmath27 , an equality known as einstein s relation @xcite .
physically , @xmath25 should be an increasing function of @xmath28 , so as to lead to increasing random forces as @xmath28 increases , with @xmath29 , since noise is turned off at @xmath30 .
classical annealing can in principle be performed through this langevin dynamics , by slowly decreasing the temperature @xmath31 as a function of time , from some initially large value @xmath32 down to zero . instead of working with the langevin equation a stochastic differential equation one might equivalently address the problem by studying the probability density @xmath33 of finding a particle at position @xmath17 at time @xmath34 .
the probability density is well known to obey a _ deterministic _ time - evolution equation given by the _ fokker - planck _
( fp ) equation @xcite : @xmath35 here , the second term in the right - hand side represents the well known _ diffusion term _ , proportional to the diffusion coefficient @xmath25 , whereas the first term represents the effect of the _ drift _ force @xmath36 , inversely proportional to the friction coefficient @xmath37 @xcite .
annealing can now be performed , in principle , by keeping the system for a long enough equilibration time at a large temperature @xmath32 , and then gradually decreasing @xmath28 to zero as a function of time , @xmath31 , in a given annealing time @xmath3 .
we can model this by assuming @xmath38 where @xmath39 is some assigned monotonically decreasing function for @xmath40 $ ] , with @xmath41 and @xmath42 . in this manner the diffusion constant @xmath43 in eq .
( [ fp : eqn ] ) becomes a time - dependent quantity , @xmath44 .
the fp equation should then be solved with an initial condition given by the equilibrium boltzmann distribution at temperature @xmath45 , i.e. , @xmath46 . the final average potential energy after annealing in excess of the true minimum value , will then be simply given by : @xmath47 where @xmath10 is the actual absolute minimum of the potential @xmath20 . in a completely analogous manner
, we can conceive using schrdinger s equation to perform a deterministic quantum annealing ( qa ) evolution of the system , by studying : @xmath48 \psi(x , t ) \;,\ ] ] where @xmath49 for a _ real - time _ ( rt ) evolution , while @xmath50 for an _ imaginary - time _ ( it )
evolution . here
@xmath51 will be our annealing parameter , playing the role that the temperature @xmath31 had in classical annealing .
once again we may take @xmath52 varying from some large value @xmath53 at @xmath2 corresponding to a small mass of the particle , hence to large quantum fluctuations down to @xmath54 , corresponding to a particle of infinite mass , hence without quantum fluctuations .
again , we can model this with @xmath55 , where @xmath56 is a preassigned monotonically decreasing function
. a convenient initial condition here will be @xmath57 , where @xmath58 is the _ ground state _ of the system at @xmath2 , corresponding to the large value @xmath59 and hence to large quantum fluctuations .
for such a large @xmath60 the ground state will be isolated , and separated by a large energy gap from all excited states .
the residual energy after annealing will be similarly given by eq .
( [ residual : eqn ] ) , where now , however , the probability @xmath61 should be interpreted as @xmath62 in general , the residual energy will be different for a rt or an it schrdinger evolution .
we will discuss in some detail rt versus it schrdinger evolution in sec .
[ rt_vs_it : sec ] .
the basic question we pose is which annealing scheme is eventually more effective , leading to the smallest final residual energies @xmath63 .
this might seem an ill - posed question , because the time scales involved in the classical and in the quantum evolution are different , and also because , practically , the two approaches might involve different computational costs which would imply different cpu time scales . in other words , it might seem that it only makes sense to ask how well an annealing scheme performs in a given cpu - time @xmath64 , with all the unavoidable uncertainty associated to a cpu - time - related answer ( involving , among other things , the programmer s skills , the algorithmic choices , and the computer architecture ) .
we will show , however , that the behavior of @xmath63 can be so vastly different for the different schemes , obviously in strict relation with the form of the potential , that such time scale concerns are often practically irrelevant @xcite .
preliminary to any further treatment of a potential with barriers , and as a warm - up exercise which will be useful later on , we start here with the simple case of a parabolic potential in one dimension , @xmath65 , which has a trivial minimum in @xmath66 , with @xmath67 , and no barriers whatsoever .
let us consider classical fp annealing first . as detailed in the appendix , it is a matter of simple algebra to show that , for the harmonic potential , one can write a simple closed linear differential equation @xcite for the average potential energy @xmath68 , which has the form : @xmath69 \;,\ ] ] the initial condition being simply given by the equipartition value @xmath70 . as with every one - dimensional linear differential equation , eq .
( [ der_residual_closed : eqn ] ) can be solved by quadrature for any choice of @xmath31 and @xmath44 . assuming the annealing schedule to be parameterized by an exponent @xmath71 , @xmath72 , @xmath3 being the annealing time , and the diffusion constant @xmath25 to behave as a power law of temperature , @xmath73 with @xmath74
, we can easily extract from the analytical solution for @xmath68 the large @xmath3 asymptotic behavior of the final residual energy @xmath75 . that turns out to be : @xmath76 trivial as it is , annealing proceeds here extremely fast , with a power - law exponent @xmath77 that can increase without bounds ( for instance if @xmath78 ) upon increasing the exponent @xmath79 of the annealing schedule . consider now the schrdinger evolution problem for this potential , @xmath80 \psi(x , t ) \\ & & \psi(x , t=0 ) = \psi_{0}(x ) \ ;
, \nonumber \end{aligned}\ ] ] where @xmath81 is the ground state gaussian wavefunction corresponding to the initial value of the laplacian coefficient @xmath82 , and @xmath83 or @xmath84 for a real time ( rt ) or an imaginary time ( it ) evolution , respectively .
this problem is studied in detail in the appendix , where we show that a gaussian _ ansatz _ for @xmath85 , of the form @xmath86 with @xmath87 , satisfies the time - dependent schrdinger equation as long as the inverse variance @xmath88 of the gaussian satisfies the following ordinary non - linear first - order differential equation : @xmath89 contrary to the classical case , there is no simple way of recasting the annealing problem in terms of a closed linear differential equation for the average potential energy @xmath68 .
the final residual energy @xmath75 is still expressed in terms of @xmath90 ( or better , of its real part @xmath91 ) , @xmath92 but the behavior of @xmath88 must be extracted from the study of the non - linear equation ( [ gauss_ode_b : eqn ] ) .
the properties of the solutions of eq .
( [ gauss_ode_b : eqn ] ) are studied in detail in the appendix , where we show that : * @xmath63 can not decrease faster than @xmath93 , for large @xmath3 , i.e. , a power - law exponent @xmath94 is bound to be @xmath95 . * adopting a power - law annealing schedule @xmath96 , the exponent @xmath97 for the it case is @xmath98 increasing towards the upper bound @xmath99 as @xmath100 is increased towards @xmath101 . *
rt quantum annealing proceeds with exactly the same exponent @xmath97 as it quantum annealing although @xmath102 in general , except that the limit @xmath103 ( abrupt switch - off of the laplacian coefficient ) is singular in the rt case . summarizing
, we have learned that , for a single parabolic valley in configuration space , both ca and qa proceed with power - laws , but ca can be much more efficient than qa , with an arbitrarily larger power - law exponent .
we underline however that this is merely an academic matter at this point , steepest descent being much more efficient than both ca and qa in such a simple case .
the power of qa shows up only when potentials with barriers are considered .
let us take now the classical potential to be optimized as a simple double - well potential in one - dimension @xmath104 with @xmath105 , @xmath106 , and @xmath107 real constants
. in absence of the linear term ( @xmath108 ) , the potential has two degenerate minima located at @xmath109 , and separated by a barrier of height @xmath105 .
when a small linear term @xmath110 is introduced , with @xmath111 , the two degenerate minima are split by a quantity @xmath112 , the minimum at @xmath113 becoming slightly favored . for reasons that will be clear in a moment ,
it is useful to slightly generalize the previous potential to a less symmetric situation , where the two wells possess definitely distinct _ curvatures _ at the minimum ( i.e , their widths differ substantially ) .
this is realized easily , with a potential of the form : @xmath114 with @xmath115 , both positive .
( the discontinuity in the second derivative at the origin is of no consequence in our discussion . ) to linear order in the small parameter @xmath107 , the two minima are now located at @xmath116 , the splitting between the two minima is given by @xmath117 , while the second derivative of the potential at the two minima , to lowest order in @xmath107 , is given by : @xmath118 obviously , @xmath119 is recovered if we set @xmath120 in @xmath121 .
we now present the results obtained by the annealing schemes introduced in sec .
[ annealing_def : sec ] above .
the fokker - planck and the schrdinger equation ( both in rt and in it ) were integrated numerically using a fourth - order runge - kutta method , after discretizing the @xmath17 variable in a sufficiently fine real space grid @xcite . for the fp classical annealing ,
the results shown are obtained with a linear temperature schedule , @xmath122 ( i.e. , @xmath123 ) , and a diffusion coefficient simply proportional to @xmath31 , @xmath124 ( i.e. , @xmath78 ) .
consequently , the friction coefficient is kept constant in @xmath34 , @xmath125 .
similarly , for the schrdinger quantum annealing we show results obtained with a coefficient of the laplacian @xmath52 vanishing linearly in a time @xmath3 , @xmath126 ( i.e. , @xmath127 ) .
( a , b ) : the annealed final probability distribution @xmath61 at different values of the annealing time @xmath3 , for both the fokker - planck classical annealing ( ca , panel ( a ) ) , and the imaginary time schrdinger quantum annealing ( it - qa , panel ( b ) ) .
( c ) final residual energy @xmath63 versus annealing time @xmath3 for quantum annealing in real time ( rt ) and imaginary time ( it ) compared to the fokker - planck classical annealing ( ca ) .
the solid line in ( c ) is a fit of the ca data .
the double well potential ( dashed line in ( a , b ) , inset of ( c ) ) is here given by eq .
( [ vasym : eqn ] ) with @xmath128 .
, width=302 ] same as fig . , for the asymmetric potential in eq .
( [ vasym : eqn ] ) with @xmath129 ( dashed line in ( a , b ) , inset of ( c ) ) .
notice the different behavior of rt and it , in the present case .
, width=302 ] instantaneous eigenvalues ( a ) and ground state wavefunctions ( b ) of the schrdinger problem @xmath130 for different values of @xmath60 , for the symmetric potential in eq .
( [ vasym : eqn ] ) with @xmath128 .
, width=302 ] same as fig . , for the asymmetric potential in eq .
( [ vasym : eqn ] ) with @xmath129 .
notice the clear landau - zener avoided crossing in ( a ) , indicated by the arrow and magnified in the inset .
, width=302 ] fig .
[ fig_ann_sym : fig ] shows the results obtained for the final annealed probability distribution @xmath61 at different values of @xmath3 , for both the fokker - planck ( ca , panel ( a ) ) and the scrdinger imaginary - time case ( it , panel ( b ) ) , for a symmetric double well potential @xmath131 , with @xmath132 ( our unit of energy ) , @xmath133 ( unit of length ) , @xmath134 . fig .
[ fig_ann_sym : fig](c ) summarizes the results obtained for the residual energy @xmath63 .
[ fig_ann_asym : fig](a , b , c ) shows the corresponding results for an asymmetric double well potential , eq . [ vasym : eqn ] , with @xmath135 , @xmath136 , and @xmath134 .
we notice immediately that qa wins , in both cases , over ca for large enough value of @xmath3 .
the rt - qa behaves as its it counterpart for the symmetric double well , while it shows a different behavior in the asymmetric case ( see below for comments ) . to go deeper into the details of the different evolutions ,
let us begin discussing the ca data ( panel ( a ) and ( c ) of figs . [ fig_ann_sym : fig ] and [ fig_ann_asym : fig ] ) , which show similar behaviors for both choices of the potential .
starting from an initially broad boltzmann distribution at a high @xmath137 , @xmath138 ( solid lines ) , the system quickly sharpens the distribution @xmath33 into two well - defined and quite narrow peaks located around the two minima @xmath139 of the potential .
this agrees very well with expectations based on the ca in a harmonic potential , which showed that the width of the gaussian should decrease linearly in @xmath3 ( @xmath140 for @xmath141 ) , as is indeed found in our double well case too .
if we denote by @xmath142 the integral of each of the two narrow peaks , with @xmath143 , it is clear that the problem has effectively been reduced to a _
discrete _ two - level system problem .
the time evolution @xmath142 therefore obeys a discrete master equation which involves the thermal promotion of particles over the barrier @xmath105 , of the form : @xmath144 \ , e^{-\frac{\delta_v+b}{k_bt(t ) } } - p_{+}(t ) e^{-\frac{b}{k_bt(t ) } } \;,\ ] ] where @xmath145 is an attempt frequency , while @xmath146 and @xmath147 are the potential barriers seen from a particle in the metastable minimum , @xmath148 , and in the true minimum , @xmath149 , respectively .
( [ hf_1:eqn ] ) was studied by huse and fisher in ref . , where they showed that the asymptotic value of the residual energy @xmath150 is given by : @xmath151 where @xmath152 is a constant .
so , apart from logarithmic corrections , the leading power - law behavior is of the form @xmath153 , where the exponent is controlled by the ratio @xmath154 between the energy splitting of the two minima @xmath155 and the barrier @xmath156 . as shown in figs .
[ fig_ann_sym : fig](c ) and [ fig_ann_asym : fig](c ) ( solid lines through solid circles ) , the asymptotic behavior anticipated by eq .
( [ hf_2:eqn ] ) fits nicely our ca residual energy data ( solid circles ) , as long as the logarithmic corrections are accounted for in the fitting procedure @xcite .
obviously , we can make the exponent as small as we wish by reducing the linear term coefficient @xmath107 , and hence the ratio @xmath154 , leading to an exceedingly slow classical annealing .
the behavior of the qa evolution is remarkably different .
starting from fig .
[ fig_ann_sym : fig ] , we notice that it and rt evolutions give very similar residual energies , definitely faster decaying than the ca data , while the corresponding final wavefunction only slowly narrows around the minimum of the potential .
notice also the asymptotic behavior of the residual energy , @xmath157 , indicated by the dashed line in fig .
[ fig_ann_sym : fig](c ) : this rather strange exponent is simply the appropriate one for the schrdinger annealing with a linear schedule within an harmonic potential ( the lower minimum valley , see sec . [ harmonic : sec ] ) . the asymmetric potential results , shown in fig . [ fig_ann_asym : fig ] , are even more instructive .
the initial wavefunction squared @xmath158 corresponds to a quite small mass ( a large @xmath159 ) , and is broad and delocalized over both minima ( solid line ) . as we start annealing , and if the annealing time @xmath3 is relatively short that is , if @xmath160 , with a characteristic time @xmath161 which depends on which kind of annealing , rt or it , we perform the final wavefunction becomes mostly concentrated on the _ wrong minimum _ , roughly corresponding to the ground state with a still relatively large @xmath162 ( see fig . [ fig_eigen_asym : fig ] ) .
the larger width of the wrong valley is crucial , giving a smaller quantum kinetic energy contribution , so that tunneling to the other ( deeper ) minimum does not yet occur . by increasing @xmath3
, there is a crossover : the system finally recognizes the presence of the other minimum , and effectively tunnels into it , with a residual energy that , once again , decays asymptotically as @xmath163 ( dashed line in fig .
[ fig_ann_asym : fig](c ) ) .
there is a characteristic annealing time @xmath161 different in the two scrdinger cases , rt and it above which tunneling occurs , and this shows up as the clear crossover in the residual energy behavior of both it and rt , shown in fig .
[ fig_ann_asym : fig](c ) .
these findings can be quite easily rationalized by looking at the _ instantaneous _ ( adiabatic ) eigenvalues and eigenstates of the associated time - independent schrdinger problem , which we show in fig .
[ fig_eigen_asym : fig](a , b ) .
looking at the instantaneous eigenvalues shown in fig .
[ fig_eigen_asym : fig](a ) we note a clear avoided - crossing occurring at @xmath164 , corresponding to a resonance condition between the states in the two different valleys of the potential .
for @xmath165 the ground state wavefunction is predominantly concentrated in the wider but metastable valley , while for @xmath166 it is mostly concentrated on the deeper and narrower global minimum valley . in the full time - dependent rt evolution , transfer to the lower valley
is a landau - zener problem @xcite : the characteristic time @xmath161 for the tunneling event is given by @xmath167 , where @xmath168 is the relative slope of of the two crossing branches as a function of @xmath60 , @xmath169 is the gap at the avoided - crossing point , and @xmath53 is the initial value of the annealing parameter .
( for the case shown in fig .
[ fig_eigen_asym : fig ] , we have @xmath170 , @xmath171 , hence @xmath172 , see rightmost arrow in fig .
[ fig_ann_asym : fig](c ) . ) the landau - zener probability of jumping , during the evolution , from the ground state onto the `` wrong '' ( excited ) state upon fast approaching of the avoided level crossing is @xmath173 , so that adiabaticity applies only if the annealing is slow enough , @xmath174 .
the it characteristic time is smaller , in the present case , than the rt one .
we will comment further on this point in sec .
[ rt_vs_it : sec ] . in a nutshell ,
the reason for this is the following .
after the system has jumped into the excited state , which occurs with a probability @xmath173 , the residual it evolution will filter out the excited state ; this relaxation towards the ground state is controlled by the annealing rate as well as by the average gap seen during the residual evolution .
numerically , the characteristic time @xmath175 seen during the it evolution is of the order of @xmath176 , see leftmost arrow in fig .
[ fig_ann_asym : fig](c ) , rather than being proportional to @xmath177 as @xmath178 would imply .
obviously , instantaneous eigenvalues / eigenvectors can be studied for the fokker - planck equation as well ; their properties , however , are remarkably different from the landau - zener scenario just described for the schrdinger case . fig .
[ fig_diag_sym :
fig](c ) shows the first four low - lying eigenvalues of the fp equation as a function of @xmath28 ( for a symmetric choice of the potential ) , while fig .
[ fig_diag_sym : fig](a , b ) show the corresponding eigenstates for two value of the temperature , @xmath179 and @xmath180 .
( the asymmetric potential cases are virtually identical , and are not shown ) .
the lowest eigenvalue of the fp operator is identically @xmath181 and the corresponding eigenvector @xcite is the boltzmann distribution @xmath182 , with roughly symmetric maxima on the two valleys .
the first excited state correspond to distribution peaked on the two valley but with a node at the origin , and is separated from the ground state by an exponential small arrhenius - like gap @xmath183 .
higher excited states are separated by a very large gap , so that , effectively , only the two lowest lying states dominate the dynamics at small temperature .
the reduction of a continuum double - well fp classical dynamics onto a discrete effectively quantum two - level system , previously noticed , is quite evident from this form of the spectrum . on the contrary ,
the true quantum case never allowed for a discrete two - level system description whatsoever , except perhaps for large @xmath60 . for small enough @xmath166 ,
in particular , the tower of oscillator states within the valley at @xmath149 is always very close in energy to the actual ground state , and the quantum annealing evolution reduces effectively to a particle in a single harmonic well .
this explains the rather large width of the final distributions @xmath184 observed in the quantum case .
instantaneous fokker - planck eigenvalues ( panel ( c ) ) as a function of temperature @xmath28 , and the corresponding eigenstates for two values of @xmath28 ( panels ( a ) and ( b ) ) .
the potential is here the symmetric one , @xmath119 in eq .
[ vsym : eqn ] with @xmath132 , @xmath185 , @xmath134 .
similar results ( not shown ) are obtained for the asymmetric double well potential @xmath121 .
, width=302 ] summarizing , we have found that qa and ca proceed in a remarkably different way .
ca is sensitive to the height of the barrier , more precisely to the ratio @xmath154 between the energy offset @xmath155 of the two minima , and the barrier height @xmath156 . on the contrary ,
qa crucially depends on the tunneling probability between the two valleys , which is reflected in a landau - zener ( avoided crossing ) gap : a wide tunneling barrier is obviously bad for qa .
finally , we noticed that rt and it proceed with somewhat different characteristic times : we discuss this issue a bit more in the following section . a schrdinger dynamics in imaginary - time ( it ) is clearly much more convenient than that in real - time ( rt ) for simulations on current classical computers , but it makes a difference in the final results ?
the answer to this question is , we believe : no , it does not make a difference , in essence , although it does give quantitatively better results . to qualify this statement ,
let us denote by @xmath186 the solution of the scrdinger equation @xmath187 |\psi^{(\xi)}(t)\rangle \nonumber \\ assume that @xmath5 is the ground state of the initial hamiltonian at time @xmath188 , @xmath189 , while @xmath83 for rt or @xmath190 for it . by definition ,
the final residual energy after annealing up to time @xmath191 , where the kinetic energy is finally turned off , is given by : @xmath192 we conjecture that the residual energies for the two alternative way of doing a schrdinger evolution verify the following : i ) the it residual energy is _ not larger _ then the rt one , that is @xmath193 and ii ) in many problems , the leading asymptotic behavior , for @xmath194 , might be identical for @xmath195 and @xmath196 . expectation ( i ) seems very reasonable : it is simply inspired by the time - independent case , where it is well known that the it schrdinger dynamics tends to `` filter the ground state '' out of the initial trial wave function , as long as the gap between the gs and the first excited state is non - zero .
however , we have here a time - dependent situation , and the result is a priori not guaranteed .
we do not have a proof of this statement , but we have verified it in all the cases where an explicit integration of the schrdinger equation has been possible ( see , for instance , the results of the previous section ) .
( needless to say , we have no proof of ( ii ) either , but , again , it never failed in all our tests . ) the simplest time - dependent problem where one can test our conjectures , is the discrete two - level system ( tls ) problem . here , in terms of pauli matrices , @xmath197 , while @xmath198 , with @xmath199
the full @xmath200 is therefore @xmath201 the annealing interpretation is very simple : the classical optimal state is @xmath202 , with energy @xmath203 , separated from the excited state @xmath204 by a gap @xmath169 .
the kinetic term induces transitions between the two classical states . starting from the ground state of @xmath205 at time @xmath206 we let the system evolve up to to time @xmath207 , when the hamiltonian is entirely classical , @xmath208 .
the probability of missing the instantaneous final ground state @xmath202 , ending up with the excited state @xmath204 , is : @xmath209 . in principle
, @xmath210 depends , for given @xmath211 , both on the initial @xmath212 and on the annealing time @xmath3 .
the really important parameter , however , turns out to be the ratio @xmath213 between these two quantities , which determines the `` velocity of annealing '' : taking @xmath194 ( i.e. , @xmath214 ) , and @xmath215 with @xmath199 for every @xmath34 , the problem can be solved analytically ( in terms of parabolic cylinder functions , see for instance ref . for the rt case ) for both rt and it .
the probability @xmath216 of ending into the excited state can be expressed in terms of the variable @xmath217 .
the explicit expressions , in terms of gamma functions , are : @xmath218 where @xmath219 and @xmath220 for rt , while @xmath221 and @xmath222 for it . a plot of @xmath210 for both rt and it is shown in fig .
[ zener : fig ] as a function of @xmath217 . the probability @xmath216 of ending up into the excited state , given by eq . ,
for the discrete two - level system problem in eq .
, for both imaginary - time ( it , dashed line ) and real - time ( rt , solid line ) schrdinger annealing .
the large-@xmath145 behavior of @xmath216 is , in both cases , given by @xmath223 .
, width=302 ] comparison between the rt ( solid lines ) and the it ( dashed lines ) evolution of a landau - zener problem , eq .
[ lz_rot : eqn ] , for several values of the tunneling gap @xmath169 ( the values of @xmath211 shown are @xmath224 , while @xmath225 ) .
the inset shows the two instantaneous eigenvalues of the problem , @xmath226 , as a function of @xmath34 .
, width=302 ] note that : i ) the it - result for @xmath210 ( dashed line ) is always below the rt - result , ii ) the difference between the two curves is only quantitative : one can verify analytically that the leading behavior for large @xmath145 is the same in both cases , i.e. , @xmath227 .
similar results are obtained by direct numerical integration of the scrdinger equation for finite @xmath53 and @xmath3 , and with other forms of @xmath52 . with the same toy model
, we can illustrate another point raised in the previous section : what happens to the it evolution _
after _ a landau - zener avoided crossing gap is encountered .
the hamiltonian we consider is essentially that in eq .
[ lz_1:eqn ] , simply rotated in spin space , @xmath228 in absence of the tunneling amplitude @xmath211 , the two energy levels would cross at @xmath229 , while for @xmath230 the two instantaneous eigenvalues are simply @xmath231 ( see inset in fig .
[ comp_it_rt : fig ] ) . starting with the system in the ground state at @xmath232
, we can monitor the probability of getting onto the excited states at any time @xmath34 , which we plot in fig .
[ comp_it_rt : fig ] for both the rt and the it evolution and for several values of @xmath211 ( taking @xmath225 ) .
the rt data provide an illustration of the well - known landau - zener result : after a ( relatively short ) tunneling time , and possibly a few oscillations , the probability of of getting onto the excited state saturates to a value given by @xmath233 . as for the it data ,
the initial ( tunneling ) part and the subsequent plateau of the curves are similar to the rt case : the plateau value attained , call it @xmath234 , is indeed very close to the rt saturation value ( in fact , asymptotically the same for @xmath235 ) ; after that , the it evolutions starts to filter out the ground state component initially present in the state with a small amplitude @xmath236 through the usual mechanism of suppression of excited states , leading to a @xmath237 which is nicely fit by the curve @xmath238 } } { ( 1-p_{ex}^ * ) + p_{ex}^ * e^{-2\int_0^t dt ' [ e_+(t')-e_-(t ' ) ] } } \;,\ ] ] which asymptotically goes to zero as @xmath239 .
this rather trivial effect of filtering , if on one hand explains the discrepancy between the it and the rt evolution observed in the asymmetric double well case of the previous section , is , on the other hand , of no harm at all : on the contrary , it provides a quantitative improvement of it over rt . in summary ,
the essential equivalence of it and rt schrdinger annealing ( with , moreover , a quantitative improvement of it over rt ) justifies practical implementations of quantum annealing based on imaginary - time quantum monte carlo schemes .
after discussing at length the annealing problem in a potential with one minimum and with two minima , we wish to move on to a multi - minima problem , however simple .
there are simple but interesting one - dimensional potentials which allow us to do that .
the first example was proposed and solved by shinomoto and kabashima in ref . , and consists in a parabolically shaped washboard potential .
this example will display a logarithmically slow classical annealing , showing ca may run into trouble even in simple models with no complexity whatsoever , whereas quantum mechanics can do much better in this case .
parabolic washboard potential resulting in a logarithmically slow classical annealing .
the minima are regularly located at positions @xmath240 , and the dashed line shows the parabolic envelope potential . , width=302 ] consider a wiggly one - dimensional potential with barriers of individual height @xmath241 separating different local minima , regularly located a distance @xmath106 apart one from each other , i.e. , at positions @xmath242 .
the ith - local minimum is at energy @xmath243 , so that the resulting envelope is parabolic . in order to study the dynamics of a particle in this potential ,
a good starting point is to write the master equation for the probability @xmath244 that the particle is in the ith - valley at time @xmath34 : @xmath245 \;,\ ] ] where @xmath145 is an attempt frequency , @xmath246 is the effective barrier from @xmath247 to @xmath248 , and @xmath249 .
this is a well justified starting point , in view of the results of the previous sections ( secs .
[ tls : sec ] and [ harmonic : sec ] ) , showing that classical annealing is extremely fast in reducing the width of a probability distribution within each valley to an essentially delta - function like sequence of peaks of strength @xmath250 .
the actual form of the effective barriers depends on the way we model the details of the potential , with the only constraint that detailed balance is satisfied , i.e. , @xmath251 in such a way that the stationary solution , for constant @xmath28 , is simply the boltzmann probability distribution , @xmath252 .
takes @xmath253 , while @xmath254 and @xmath255 , with @xmath256 .
the potential energy of the valleys enters only through the @xmath246 , which control the probability of making transitions between valleys . in order to study eq .
( [ master : eqn ] ) , shinomoto and kabashima introduced a continuum limit , by defining a macroscopic coordinate @xmath17 , such that the mimima are at @xmath240 , and writing the equation governing the probability @xmath33 in the limit @xmath257 .
the derivation involves writing @xmath258 in terms of derivatives of @xmath33 , keeping consistently terms up to order @xmath259 and expanding exponentials with the assumption that @xmath260 .
the continuum limit equation governing the evolution of @xmath33 turns out to be a fokker - planck ( fp ) equation , eq .
( [ fp : eqn ] ) , with an effective diffusion constant of the form @xmath261 @xmath262 , and an effective drift potential @xmath65 given by the macroscopic parabolic envelope potential . in order to study the annealing properties of the system , one can then follow exactly the same steps leading to eq .
( [ der_residual_closed : eqn ] ) , which applies here too , except that now @xmath263 is substituted by @xmath264 , which has an exponential activated behavior , @xmath265 .
this exponentially activated @xmath266 changes the annealing behavior in a drastic way . recall that the ca exponent @xmath77 of sec
( [ harmonic : sec ] ) decreases towards zero as the exponent @xmath267 in the relationship @xmath268 increases . since , close to @xmath30 , @xmath269 for any arbitrarily large @xmath267
, we could suspect that the behavior of @xmath63 will no longer be a power law .
in fact , the surprising result of this exercise @xcite is that the optimal annealing schedule @xmath31 is _ logarithmic _ and that @xmath270 converges to @xmath181 at best as @xcite @xmath271 the reason behind this result is that the time derivative of @xmath270 becomes exponentially small as one anneals @xmath28 towards @xmath181 , due to the presence of @xmath183 in the diffusion constant , and an exponentially small derivative brings not surprisingly a logarithmically slow decrease of the function . to put it more physically , consider solving eq .
( [ der_residual_closed : eqn ] ) for a time independent @xmath28 ; the solution is trivially @xmath272 we observe that the solution converges to the equilibrium ( equipartition ) value @xmath273 exponentially with a characteristic time @xmath274 which itself increases exponentially fast with decreasing @xmath28 . as a result , the system will never be able to follow the decreasing @xmath28 till the end of the annealing , by maintaining roughly the equilibrium value @xmath275 .
indeed , if we assume for instance @xmath122 , the relaxation of the systems will cease to be effective
i.e. , the system will fall _ out of equilibrium _ at a time @xmath276 , and temperature @xmath277 , at which @xmath278 , i.e. , when @xmath279
. the residual energy at this point can not be smaller then the equipartition value @xmath280 , hence @xmath281 as well .
this freezing and falling out of equilibrium for classical systems with barriers seems to provide an ubiquitous source of logarithms in classical annealing @xcite .
how would one tackle this annealing problem quantum mechanically ? as the quantum analog of the master equation eq .
( [ master : eqn ] ) , we propose studying the schrdinger evolution governed by a tight - binding hamiltonian with on - site energies @xmath282 and hopping matrix - elements between adjacent sites @xmath283 @xmath284 the justification and possible limitations of this starting point , over the actual original continuum problem will be discussed at the end of the section .
here it is enough to consider that the hopping matrix - elements @xmath283 depend on tunneling through the barrier separating @xmath247 from @xmath285 .
the precise form of @xmath283 is likely to be inessential , including its energy ( and hence site ) dependence , for which we will assume the semi - classical ( wkb ) form : @xmath286 @xmath287 being simply related to the quantum confinement energy of a particle of mass @xmath288 in a valley of size @xmath289 . here
@xmath290 and @xmath291 are energy parameters related to the details of the potential and of the barrier , which will play little or no role .
if the mass of the particle @xmath288 ( and hence @xmath60 and @xmath283 ) is kept constant the particle will explore the potential due to the kinetic term in the hamiltonian : the correspondence between the quantum and the classical formulation is that @xmath60 plays the role of @xmath28 , @xmath283 plays the role of the classical transition probabilities , the ground state wavefunction @xmath292 at a given value of @xmath60 ( or , equivalently , of the hopping term @xmath293 ) plays the role of the classical equilibrium boltzmann distribution @xmath294 .
the question , once again , is how to anneal @xmath60 , by reducing it as a function of time , @xmath52 , in such a way as to squeeze the wavefunction of the system so that the average potential energy @xmath295 is minimal .
as it turns out , the continuum limit is once again useful .
one goes to the continuum exactly as in the fp case by using @xmath242 , writing @xmath296 and expanding everything to order @xmath259 . when written in first quantized form , the hamiltonian for the quantum particle in the macroscopic continuum coordinate @xmath17 is simply @xmath297 where the coefficient of the laplacian @xmath298 is the quantum counterpart of the classical effective diffusion constant @xmath299 @xmath300 and @xmath65 , as in the classical case .
the continuum limit quantum problem has therefore exactly the form we have considered in sec .
( [ harmonic : sec ] ) , except that @xmath52 in eq .
( [ gauss_ode_b : eqn ] ) is now substituted by an effective laplacian coefficient @xmath298 which has a highly non - linear , in fact exponential , dependence on the annealing parameter @xmath52 . we know , however , from sec .
( [ harmonic : sec ] ) , that a non - linear behavior of the type @xmath301 for the laplacian coefficient leads to a power - law decrease of @xmath63 with an exponent @xmath97 which is , remarkably , an _ increasing function of @xmath100 _ , approaching @xmath99 as @xmath103 .
therefore , contrary to the classical case , where an exponentially activated behavior of the diffusion constant @xmath299 is _ strongly detrimental _ to the annealing ( turning a power law into a logarithm ) , here the exponential wkb - like behavior of @xmath298 will do no harm at all .
indeed , we numerically integrated the relevant equation for @xmath88 , eq .
( [ gauss_ode_b : eqn ] ) with @xmath298 in place of @xmath60 , using the exponential wkb expression eq .
( [ gamma_eff : eqn ] ) for @xmath298 while annealing @xmath52 with a linear schedule , @xmath126 .
the integration was performed , as usual , with a fourth - order runge - kutta method , and was carried on up to time @xmath302 , when the kinetic term in the hamiltonian ceases to exist .
the numerical results ( not shown ) have a clear power - law behavior for the quantum annealed ( qa ) final residual energy @xmath303 , with a power law exponent @xmath97 which is compatible with @xmath99 .
once again , the exponent appears to be insensitive to the choice of the type of quantum evolution ( rt versus it ) , although the numerical values of residual energies always respect the inequality @xmath102 . before ending this section
, we would like to discuss briefly the reason for treating by tight - binding , eq .
[ h_tb : eqn ] , what was originally a continuum problem with a well defined potential landscape . as we learned from the double - well case , there is never a clear - cut discrete model ( a discrete two - level system , in that case ) describing in a complete way the continuum schrdinger problem , in all stages of the annealing . obviously ,
when the mass of the particle is very small , the tight - binding approximation contained in eq .
[ h_tb : eqn ] is not particularly good , since more than one state per valley is generally important to describe the wavefunction accurately . as the mass of the particle increases , however , the tight - binding approach gets more and more appropriate , until a further limitation appears : when the mass if very large , it is not legitimate to neglect excited states within , say , the central valley compared to the lowest states localized in metastable valleys .
we can imagine that the ultimate behavior of @xmath63 , in the quantum case , will be actually dominated by the rather trivial problem of squeezing the wavefunction in the lowest central minimum , with its characteristic power - law exponent ( @xmath304 , for instance , for a linear schedule ) .
there is , however , an intermediate region , between the very short @xmath3 scale , where the full details of the potential are important , and the very long @xmath3 scale , where the trivial squeezing mentioned above sets in , and where the tight - binding approximation reasonably predicts a power - law exponent ( of order @xmath99 ) for @xmath63 .
we believe that one of the important points that makes qa so different from ca in the present case is that the spectrum of the instantaneous eigenvalues of the quantum problem does not show any dangerous landau - zener avoided - crossing , and , correspondingly , the ground state wavefunction is always more peaked in the central valley than elsewhere . as in the two - level case ,
a disorder in the width of the different valleys would change this result .
despite their disarming simplicity , the three case studies above turn out to be extremely informative in qualifying the profound difference of qa from ca , and their surprising consequences .
we expect that these results will be very important in understanding more realistic qa problems .
of course , the cases studied , although instructive , do not possess the real ingredient which makes annealing difficult , both in ca and qa , i.e. , some form of disorder in the distribution of the minima .
we believe , for instance , that even an irregular landscape with many minima , as the double - cosine potential @xmath305 ( with @xmath306 an irrational number ) shown in fig .
[ double_cos : fig ] , would already change drastically the behavior of qa from a power - law to a logarithm . on quite general grounds ,
anderson s localization @xcite would predict that wavefunctions are localized for a genuinely disordered potential for large enough mass ( i.e. , small enough kinetic energy bandwidth ) in any @xmath307 ( this localization occurs for all value of the mass in @xmath308 ) . therefore , quantum annealing should always , via a cascade of landau - zener events , end up into some localized state which has , a priori , nothing to do with our search of the actual potential minimum .
, showing an irregular landscape with many minima.,title="fig:",width=302 ] [ double_cos : fig ] a very simply illustration of the crucial role of disorder is given by the @xmath309 disordered ising ferromagnet : @xmath310 where @xmath311 are positive random variables in the interval @xmath312 $ ] , and @xmath60 is the transverse field inducing quantum fluctuations . obviously , the ground state is the ferromagnetic state with all spins aligned up ( or down ) .
however , arbitrarily weak values of the @xmath313 can pin domain walls between up and down ferromagnetic regions , with a very small energy cost @xmath314 . for a finite system with periodic boundary conditions domain walls
appear in pairs , and separate sections of the system with alternating @xmath315 and @xmath316 ferromagnetic ground states .
given two domain walls pinned at weak-@xmath313 points a distance @xmath317 apart , healing the system via single spin flip moves requires flipping @xmath318 spins , which can be a formidable barrier to tunnel through .
the system will have a very slow annealing ( quantum , as well as classical ) while showing , at the same time , no complexity whatsoever : simple disorder is enough .
summarizing , we have compared schrdinger versus fokker - planck annealing in various simple cases of one - dimensional potential , in particular a double well potential and a parabolic washboard potential with many minima but no disorder . in all cases
the two annealings , quantum and classical , were seen to behave in a remarkably different way .
classical annealing is influenced only by the height of the barriers surrounding the relevant minima , via arrhenius - like thermal promotion over the barriers , with probability distributions which are quite localized on those minima .
quantum annealing is influenced by the structure of the eigenvalue spectrum of the problem very small landau - zener tunneling gaps associated to large barriers are highly detrimental to it . in some cases , for instance in the case of the parabolic washboard potential , quantum annealing can be much more effective then classical annealing , but , generally speaking , both strategies suffer whenever the potential landscape is disordered . as an outcome of the discussion
, it is quite clear that quantum annealing , although potentially useful and sometimes more convenient than classical annealing , is not capable , in general , of finding solutions of np - complete problems in polynomial time : indeed , quite interestingly , even trivial optimization problems ( trivial , obviously , only from the complexity point of view ) , like the one - dimensional ferromagnetic random ising model , can show a very slow annealing behavior .
this project was sponsored by miur through firb rbau017s8r004 , firb rbau01lx5h , cofin2003 and cofin2004 , and by infm ( `` iniziativa trasversale calcolo parallelo '' ) .
we acknowledge illuminating discussions with demian battaglia .
if is straightforward to find the solution of the fokker - planck eq .
( [ fp : eqn ] ) when the the potential is harmonic , @xmath65 , and the initial condition is the boltzmann distribution @xmath319 . indeed , it is simple to verify that the gaussian _ ansatz _ @xmath320 fulfills the initial condition ( if @xmath321 ) and solves the fp equation , as long as the two functions @xmath88 and @xmath322 satisfy the following ordinary differential equations : @xmath323 the normalization constant @xmath322 turns out to be irrelevant in calculating the average potential energy which we need @xmath324 and can be completely forgotten , since the equation for @xmath88 does not involve it . the equation for @xmath88 appears to be _ non - linear _ , but can immediately transformed into a linear equation by dividing up both sides by @xmath325 and recognizing that the correct variable to use is precisely @xmath326 , or better yet , @xmath68 . in terms of @xmath68
we can therefore write a linear equation of the form : @xmath327 the initial condition being simply given by the equipartition value @xmath328 . an alternative way @xcite of deriving eq .
( [ der_e_pot : eqn ] ) consists in taking the derivative with respect to time of both sides of eq .
( [ e_pot_fp : eqn ] ) , using then the fp equation for @xmath329 on the left hand - side of the ensuing equation , and finally integrating by parts the terms containing spatial derivatives of @xmath330 ( this procedure results in a closed differential equation for @xmath68 only if the potential @xmath16 is harmonic ) . as every one - dimensional linear first - order differential equation , eq .
( [ der_e_pot : eqn ] ) can be integrated by quadrature , the solution being : @xmath331 if we now anneal the temperature down to @xmath181 in a time @xmath3 in the usual way , @xmath332 , assuming that the diffusion constant behaves as @xmath333 , we readily get an analytic expression for the residual energy at the end of the annealing , @xmath75 , which can be shown to behave , for large @xmath3 , as a power law : @xmath334 quite evidently , annealing proceeds here extremely fast , with a power - law exponent @xmath77 that can increase without bounds ( for instance if @xmath78 ) upon increasing the exponent @xmath79 of the annealing schedule .
notice , however , that large values of @xmath267 are , on the contrary , detrimental .
consider now the problem of a particle moving in a parabolic potential @xmath65 , with a time - dependent mass , such that the hamiltonian is given by : @xmath335 where @xmath336 denotes the coefficient of the laplacian operator .
the schrdinger evolution of the wavefunction @xmath85 is then , @xmath337 where @xmath83 for a real time ( rt ) evolution , and @xmath84 for an imaginary time ( it ) evolution .
now , whereas general solutions of the time - dependent schrdinger equation for arbitrary initial condition @xmath338 are not easy , it turns out that if @xmath16 is quadratic , then any initial gaussian wavefunction propagates into a gaussian , which is enough for our goal . in detail , write the following _ ansatz _ for @xmath85 : @xmath339 substituting the _ ansatz _ for @xmath85 into the schrdinger evolution ( in rt or in it ) , one immediately verifies that @xmath85 satisfies eq .
( [ schr_evol : eqn ] ) for arbitrary @xmath52 as long as @xmath88 and @xmath322 satisfy the following ordinary differential equations : @xmath340 once again , the normalization constant @xmath322 turns out to be irrelevant in calculating the average potential energy @xmath341 and can be completely forgotten , since the equation for @xmath88 does not involve it .
the initial condition @xmath342 is , by assumption , such that at @xmath229 the system is in the _ ground state _ corresponding to @xmath343 .
such a ground state value @xmath344 is easily calculated by equating to zero the right - hand side of eq .
( [ gauss_ode : eqn ] ) with @xmath345 , i.e. , @xmath346 .
a few general considerations can be based on a purely qualitative analysis of eq .
( [ gauss_ode : eqn ] ) .
consider first the it case , where the equation for @xmath88 reads @xmath347 .
one can easily get convinced that @xmath88 is forced to be a real , positive and monotonically increasing function of @xmath34 , i.e. , @xmath348 and @xmath349 for @xmath350 .
therefore , we can easily write an inequality of the form : @xmath351 from which we immediately conclude , by integrating over @xmath34 the two sides of the inequality , that @xmath352 i.e. , the residual energy @xmath353 can not decrease faster than @xmath93 for @xmath194 .
we will now assume , without great loss of generality , that the laplacian coefficient @xmath52 is given by @xmath354 , where @xmath3 is an annealing time - scale ( for instance the annealing time , when a linear schedule is used ) and @xmath355 is a positive decreasing function for @xmath356 such that @xmath357 .
it is useful to switch to dimensionless variables by measuring times in unit of @xmath3 , @xmath358 , and @xmath88 in units of its initial ground state value @xmath359 .
the appropriate dimensionless quantity is therefore @xmath360 , with @xmath358 , where the parametric dependence on the annealing time - scale @xmath3 has been explicitly indicated .
the equation for @xmath361 is given by : @xmath362 \qquad \alpha = \tau \sqrt{2 k \gamma_0 } /(-\xi ) \nonumber \\
b(0;\tau ) & = & 1 \;,\end{aligned}\ ] ] where the dot , from now on , will denote a derivative with respect to @xmath363 . notice that the parametric dependence on @xmath3 is all buried into the coefficient @xmath168 , which reabsorbs also the @xmath364 appearing in the dynamics ( rt versus it ) .
this kind of non - linear differential equation is of the well known _ riccati _
it can be transformed into a _
linear second - order _ differential equation by operating the following substitution @xmath365 where , evidently , @xmath366 is defined up to an overall normalization constant .
indeed , simple algebra shows that we can re - express eq .
( [ non - linear : eqn ] ) as a second - order linear equation for @xmath366 , as follows : @xmath367 as long as @xmath368 , it is simple to verify that the second - order equation for @xmath369 can be also equivalently written as : @xmath370 finally , denoting by @xmath371 the indefinite integral of @xmath369 , such that @xmath372 , and integrating over @xmath363 both sides of eq .
( [ linear_2:eqn ] ) , we can also write : @xmath373 eq .
( [ gen_airy : eqn ] ) is easily solved when the annealing schedule for @xmath52 is linear , @xmath374 , i.e. , when @xmath375 with @xmath127 , a case in which eq .
( [ gen_airy : eqn ] ) is of the _ airy _ type . in the latter case , it is useful to perform a final change of independent variable to @xmath376 , so that , defining @xmath377 , we can write the equation for @xmath378 in the standard airy form : @xmath379 the general solution of eq . ( [ airy : eqn ] )
is given in terms of the two airy s functions @xmath380 and @xmath381 , @xmath382 where @xmath383 and @xmath145 are two constant coefficients .
going back to @xmath371 and @xmath369 , we then have the explicit expressions : @xmath384 where the prime indicates a derivative with respect to @xmath385 , and we have used the property of airy s functions that @xmath386 and @xmath387 . finally , substituting back the expressions in eq .
( [ solution_y : eqn ] ) into the original function @xmath88 , see eq .
[ b_versus_y : eqn ] , we get : @xmath388 this general solution correctly depends on one parameter only , i.e. , @xmath389 , so that we can put @xmath390 without loss of generality .
we impose the initial condition @xmath359 by requiring : @xmath391 solving for @xmath145 , we get : @xmath392 due to the asymptotic properties of the airy s functions ( @xmath393 , @xmath394 and @xmath395 , @xmath396 , when @xmath397 on the real axis ) , we conclude that for the it case : @xmath398 so that , finally , @xmath399 where we used that @xmath400 and @xmath401 . on the other hand , for the rt case we have to take the limit @xmath402 ( on the real axis ) instead , and in that region all the airy s functions oscillate
. nevertheless is possible to show that the value of @xmath145 is uniformly bounded for @xmath403 .
we conclude , therefore , that for large @xmath3 and with a linear annealing schedule , @xmath127 , @xmath404 and , consequently , the residual energy behaves asymptotically as @xmath405 the generalization of this result to an arbitrary annealing exponent @xmath406 in @xmath407 , is a bit more involved .
it is however possible to establish a generalization of eq .
( [ eps_res_airy : eqn ] ) , for arbitrary @xmath100 , that we checked by means of direct numerical integration .
it reads : @xmath408 an expression that holds true for both rt and it annealing .
a side issue , but nonetheless an interesting one we wish to discuss here relates to classical annealing , and concerns the well - known mapping of a fokker - planck equation onto an imaginary - time schrdinger problem @xcite , and its implications on the relationship between ca and qa .
the bottom - line will be that the mapping does _ not _ imply that a fp - based ca is actually equivalent to qa , and moreover that such a mapping is not particularly useful in our annealing context .
consider , once again , the fp problem with a time - dependent temperature @xmath31 @xmath409 where both the friction coefficient and the diffusion constant are now time - dependent quantities , which we indicate by @xmath410 and @xmath44 . in order to map the problem in eq .
[ fp_2:eqn ] onto an imaginary - time schrdinger problem , the standard procedure @xcite is to pose @xmath411 and to determine @xmath412 in such a way as to eliminate the non - schrdinger - looking drift term , turning it onto a standard potential term .
the algebra is trivial .
one can show that the drift term is eliminated if , and only if , the @xmath413 satisfies the equation : @xmath414
whose solution is readily found to be : @xmath415 with @xmath416 a function of time only , which can even be taken to be constant without loss of generality . by plugging @xmath417 in the fp equation [ fp_2:eqn ] , with @xmath413 as above
, one can show that the resulting equation for @xmath85 is indeed of the schrdinger form @xmath418 with an effective potential @xmath419 given by @xmath420 + \frac{\partial_t \phi_0(x , t)}{\phi_0(x , t ) } \;.\ ] ] the first term in @xmath419 is the standard effective potential of the riccati form obtained in the time - independent case @xcite .
the second piece in @xmath419 is absent in the time - independent case , and can be easily seen to be @xcite : @xmath421 the main point we want to stress is that , by annealing @xmath31 and hence @xmath263 , we not only reduce the coefficient of the laplacian in eq .
[ fp_to_it : eqn ] , but we also strongly modify the _ potential _
@xmath419 , at variance with a genuine qa where only the kinetic term is annealed down .
the modifications of the potential are so strong that , at low temperature , the instantaneous eigenvalue spectrum associated to the fp equation , as discussed in sec .
[ tls : sec ] , is vastly different from that of the quantum double well system .
it is easy to see , by direct substitution in eq .
( [ fp : eqn ] ) , that , at constant @xmath28 , the boltzmann distribution @xmath182 is a stationary solution of the fokker - planck equation only if @xmath422 and @xmath43 satisfy the einstein s relation , @xmath423 .
if , for instance , @xmath63 behaves in one scheme as a power - law , @xmath424 , and in another as a slowly decreasing logarithm , @xmath425 , then the first scheme will be sooner or later more convenient independently of most details .
typically , we restrict the variable @xmath17 in the interval @xmath426 $ ] , beyond which the potential is too large and @xmath33 is negligible , and use a grid of up to 1000 points , checking for convergence of the results .
this is generally speaking a difficult fit : the estimated value of @xmath427 is between 0.14 and 0.22 , depending on the inclusion of sub - leading corrections , to be compared with the theoretical value of 0.22 . neglecting the log - corrections leads to a wrong estimate of the power - law exponent . for the sake of precision , we should stress the fact that the logarithm is , strictly speaking , born out of the continuum limit , which is indeed valid only for @xmath428 .
if the lattice constant @xmath106 is kept finite , there is a minimum non - zero value @xmath155 for the splitting between the bottom of any two valleys , and , based on the two - level system results , we could anticipate a final power - law behavior @xmath429 .
nevertheless , the logarithmic behavior should apply in a whole temperature window with a final crossover to the power - law behavior . | this paper investigates the basic behavior and performance of simulated quantum annealing ( qa ) in comparison with classical annealing ( ca ) .
three simple one dimensional case study systems are considered , namely a parabolic well , a double well , and a curved washboard .
the time dependent schrdinger evolution in either real or imaginary time describing qa is contrasted with the fokker planck evolution of ca . the asymptotic decrease of excess energy with annealing time is studied in each case , and the reasons for differences are examined and discussed . the huse - fisher classical power law of double well ca is replaced with a different power law in qa . the multi - well washboard problem studied in ca by shinomoto and kabashima and leading classically to a logarithmic annealing even in the absence of disorder , turns to a power law behavior when annealed with qa .
the crucial role of disorder and localization is briefly discussed . |
ogle ( optical gravitational lensing experiment , the leader : andrzej udalski ) is a long term project , with a gradual expansion of its capability .
the dedicated 1.3 meter telescope is located at the las campanas observatory in chile , a site owned and operated by the carnegie institution of washington .
ogle is operated by the warsaw university observatory .
information about it may be found at : 0.2 cm ogle : 1.0 cm http://sirius.astrouw.edu.pl/~ ogle/ ogle : 1.0 cm http://bulge.princeton.edu/~ ogle/ 0.05 cm 0.2 cm ogle - iii is the current , third stage of the project .
the observations are done with a mosaic @xmath0 ccd camera built by a. udalski . ogle - iii begun in may 2001 . in the 2002 galactic bulge season
almost 400 candidate microlensing events were discovered : ogle - ews : 1.0 cm http://sirius.astrouw.edu.pl/~ ogle / ogle3/ews / ews.html a search for planetary transits was conducted on 32 nights during a 45 day interval in 2001 .
a total of 59 stars were identified , for which low depth flat bottom transits with orbital periods shorter than 10 days were found ( udalski et al . 2002a , b ) .
some of these are likely caused by ` hot jupiters ' , others by brown and red dwarfs .
the analysis of ogle - ii stage ( udalski , kubiak & szymaski 1997 ) , which covered years 1997 - 2000 , is advanced , and almost all data is in public domain .
this includes catalogs of i , v , b magnitudes and positions for a total of almost 40 million stars in the galactic bulge , and in the magellanic clouds ( udalski et al .
2002c , 2000b , 1998a ) , catalogs of over 500 microlensing events ( udalski et al .
2000a , woniak et al . 2001 ) , catalogs of over 270 thousand variable stars ( zebru et al .
2001 , woniak et al .
2002 ) , and in particular thousands of eclipsing binaries in the smc ( udalski et al .
1998a ) and cepheids in the lmc and smc ( udalski et al .
1999a , b ) .
in addition ogle astrometry provided proper motion measurements of thousands of stars ( soszyski et al .
2002 , sumi et al .
2002 ) , leading to the discovery of streaming motion ( rotation ) in the galactic bar . of particular interest
are special microlensing events .
ogle-2000-bul-43 was found to have a spectacular parallax effect ( soszyski et al .
even more dramatic was ogle-1999-bul-19 , the first event with multiple peaks in its apparent brightness caused by the earth s orbital motion and a very long time scale : @xmath1 days ( smith et al .
even longer time scales were found for ogle-1999-bul-32 ( mao et al .
2002 , @xmath2 days and ogle sc_5 2859 ( smith 2002 , @xmath3 days ) .
the last two events are likely caused by massive lenses , probably stellar mass black holes .
asas ( all sky automated survey , the leader : grzegorz pojmaski ) is a long term project , with gradual expansion of its capabilities .
currently it has 4 small instruments , with apertures of 2 cm , 7 cm , 7 cm , and 20 cm , all located at the las campanas observatory in chile , at a site owned and operated by the carnegie institution of washington .
asas is operated by the warsaw university observatory .
information about it may be found at : 0.2 cm asas : 1.0 cm http://www.astrouw.edu.pl/~ gp / asas / asas.html asas : 1.0 cm http://archive.princeton.edu/~ asas/ 1.6 cm 0.2 cm all four instruments use as detectors apogee @xmath4 ccd cameras . the two instruments with 7 cm aperture cover all sky every two nights in standard v and i filters down to about 14 magnitude .
a total of over 6 thousand variables was discovered so far by asas ( pojmaski 1998 , 2000 , 2002 ) , and all photometric data are on the www .
all instruments are fully robotic , with some support provided by the ogle observers .
both projects aim at real time data processing , and in particular at real time recognition and verification of any new events in the sky , among them microlensing events , supernovae , novae , dwarf novae , stellar flares , grb afterglows , etc .
two specific goals of ogle are : a firm detection of microlensing by planets , and by stellar mass black holes .
so far stellar mass black holes were suggested as lensing masses for several long microlensing events ( bennett et al .
2002 , mao et al .
2002 , smith 2002 ) , but evidence is not conclusive .
no definite planetary microlensing event has been discovered so far ( cf . bennett et al .
1999 , albrow et al . 2000 , gaudi et al .
2002 ) . inspecting almost 400 microlensing events reported by ogle - iii ews ( early warning system ) in 2002
jaroszyski & paczyski ( 2002 ) noticed that the event ogle-2002-blg-055 has a single data point deviating from otherwise smooth microlensing light curve by 0.6 magnitudes , while the photometric errors were about 0.01 magnitude .
udalski kindly examined the ccd image and found that there was nothing wrong with it , i.e. the bright point appears to be real .
as nearby data points show no obvious departure from a smooth light curve a plausible interpretation of the phenomenon is in terms of a binary lens with a very extreme mass ratio .
while a unique value of the mass ratio can not be determined from so sparse time coverage , a good fit to all data was obtained for the mass ratio of 0.01 and 0.001 , indicating a planetary mass companion to a stellar mass lens .
obviously , it is not possible to make a strong claim based on a single data point , but a modest modification of the future ogle observing procedure will provide a far better coverage of future planetary events .
there are typically only several hundred stellar microlensing events unfolding at any given time , compared to over 150 million stars monitored on a given night .
small ccd sub - frames covering known events can be processed within minutes of data acquisition . when an anomalous data point is noticed the observation of the field will be repeated . if the anomaly is confirmed the field
will be observed every 30 or 60 minutes , to provide a good coverage of the rapidly changing brightness , and allowing a unique determination of the mass ratio .
we expect that ogle - iii ews system will have this capability by the spring of 2003 ( a. udalski , private communication ) , and it is likely that the first definite planetary microlensing event will be detected in 2003 . at the opposite end of the lens mass spectrum are the very long duration events , which are plausible candidates for stellar mass black holes .
currently two independent lens parameters can be determined for long events with the ogle - iii photometry : the event time scale @xmath5 and the magnitude of the parallax effect .
one additional parameter : the angular separation between the two images , or the astrometric shift in the combined light centroid is needed to determine uniquely the lens mass .
it will be possible to make such measurements with the future vlt interferometer ( delplancke et al .
2001 , segransan et al .
2002 ) , or with the existing hst . a massive lens like the one associated with ogle-1999-bul-32 had the two images separated by several milli arc seconds ( mao et al . 2002 ,
9 ) , leading to the centroid motion of a comparable amount .
this astrometric effect is within easy reach of the hst ( e.g. benedict et al .
the long events last several years , and with the ogle - iii data rate there will be several of them unfolding at any given time within several degrees of the galactic center . very likely the vlti and/or hst observations may be scheduled in advanced , with no need for the too ( target of opportunity ) mode of operation .
little is known about the number and the distribution of stellar mass black holes in the galaxy , and the range of their masses .
ogle is likely to lead to the first definite mass determination in 2004 - 2005 .
we expect that ogle as well as asas will expand their capability in the future .
there are no definite and specific plans for the expansion , but there is a general idea how to proceed . a natural step to ogle - iv would be another 1.3 - 1.8 meter telescope with a ccd camera with a total of @xmath6 giga pixels , i.e. an instrument with the data rate 16 times higher than current ogle - iii .
the data rate is the single most important parameter for the search and/or monitoring projects like ogle and asas . for asas
we expect more frequent all sky coverage , with a more gradual increase in its depth .
a virtue of bright transients is that they may be followed up for a longer time , in greater detail and with more instruments than faint transients can .
obviously , bright transients are rare , hence the need for all sky coverage and a frequent time sampling .
smith , m. c. submitted to mnras ( 2002 ) smith , m. c. et al .
( ogle ) mnras 336 , 670 = astro - ph/0206503 ( 2002 ) soszyski , i. et al .
( ogle ) apj , 552 , 731 = astro - ph/0012144 ( 2001 ) soszyski , i. et al .
( ogle ) aca 52 , 143 = astro - ph/0205289 ( 2002 ) sumi , t. , eyer , l. , & woniak , p. r. : astro - ph/0210381 ( 2002 ) udalski , a. , kubiak , m. , & szymaski , m. : aca 47 , 319 = astro - ph/9710091 ( 1997 ) udalski , a. et al .
( ogle ) : aca 48 , 147 = astro - ph/9806313 ( 1998a ) udalski , a. et al .
( ogle ) : aca 48 , 563 = astro - ph/9812348 ( 1998b ) udalski , a. et al .
( ogle ) : aca 49 , 223 = astro - ph/9908317 ( 1999a ) udalski , a. et al .
( ogle ) : aca 49 , 437 = astro - ph/9912096 ( 1999b ) udalski , a. et al .
( ogle ) : aca 50 , 1 = astro - ph/0002418 ( 2000a ) udalski , a. et al .
( ogle ) : aca 50 , 307 = astro - ph/0010150 ( 2000b ) udalski , a. et al .
( ogle ) : aca 52 , 1 = astro - ph/0202320 ( 2002a ) udalski , a. et al .
( ogle ) : aca 52 , 115 = astro - ph/0207133 ( 2002b ) udalski , a. et al .
( ogle ) : aca 52 , 217 = astro - ph/0210278 ( 2002c ) woniak , p. r. et al .
( ogle ) : aca 51 , 175 = astro - ph/0106474 ( 2001 ) woniak , p. r. et al .
( ogle ) : aca 52 , 129 = astro - ph/0201377 ( 2002 ) zebru , k. et al .
( ogle ) : aca 51 , 317 = astro - ph/0110623 ( 2001 ) | ogle and asas are long term observing projects operated by the warsaw university observatory at the las campanas site in chile .
ogle is currently monitoring almost 200 million stars in the galactic bulge and the magellanic clouds , and has detected so far almost 1,000 events of gravitational microlensing with the dedicated 1.3-meter telescope .
asas uses several very small instruments to monitor all southern sky for variability down to approximately 14 magnitude .
a total of almost 300 thousand variable stars were discovered so far by the two projects , and all photometric data is available on the www .
both projects aim at real time recognition and verification of all new phenomena in the sky .
ogle is likely to discover planets by 2003 and stellar mass black holes by 2004 - 2005 .
all ogle and asas data is made public domain as soon as possible , and may be used by a virtual observatory . |
in recent studies of the most metal - poor stars known , an interesting trend has been discovered regarding the abundance of carbon .
several authors have noted that roughly 25% of the most metal - poor stars studied exhibit strong enhancements of carbon , as reflected in the unusual strength of their ch g - band features .
these carbon - enhanced metal - poor ( cemp ) stars occur with such a large frequency that one is forced to consider the implications on the nature of possible production mechanisms of carbon in the early universe .
over the metallicity range @xmath5 [ fe / h ] @xmath6 , there exists an upper limit to the level of carbon enhancement amongst cemp stars at [ c / h ] @xmath2 @xcite .
this immediately suggests that at some early time in the universe a significant amount of carbon was produced , in all likelihood by one of the following sources : ( 1 ) a _ primordial _ mechanism from massive stellar progenitors , ( 2 ) _ intrinsic _ internal production by low - mass stars of extremely low [ fe / h ] , or ( 3 ) _ extrinsic _ production of carbon by stars of intermediate mass , which can be prodigious manufacturers of carbon during their agb stages , followed by mass transfer to a surviving lower - mass companion .
indeed , it remains possible that all three sources might play a role .
the first alternative , in which the observed levels of carbon in at least some of the cemp stars is primordial , or close to primordial , and was produced in the first generations of ( presumably quite massive ) stars , receives some support from models of element production in zero metal abundance stars in the mass range 18 30 @xmath7 @xcite . in this scenario a large amount of carbon was produced early on , and then not as prodigiously after that .
the second possibility is that , early in the universe , when there were few heavy elements present , unusually effective mixing episodes triggered at the time of helium core flash dredges up internally produced carbon and deposits it on the surfaces of low - mass stars @xcite .
the final alternative is that intermediate - mass stars ( e.g. , @xmath8 ) in binary pairs with lower - mass companions , quickly evolved , producing large amounts of carbon during their agb evolution .
then , a significant amount of this carbon - rich material was transferred to the long - lived companion , via roche - lobe overflow or a wind .
the lower - mass companion is presently observed to be carbon - rich , while the higher - mass carbon - producing star is now a faint white dwarf ( see , e.g. , @xcite ) . in this scenario
, one would expect to be able to detect the presence of the binary system , either directly through high - angular - resolution observations , or indirectly by detecting the tell - tale wobble in the orbit of the visible companion .
in order to investigate these different scenarios , new medium - resolution spectroscopic data has been gathered , with a variety of 2m-4 m class telescopes , for over 350 stars in the large sample of carbon - rich stars reported by christlieb et al .
it should be kept in mind that this sample of stars was selected based , not on metallicity , but rather , on the apparent level of carbon enhancement revealed in low - resolution objective - prism specta . as a result ,
only those metal - poor stars which exhibited enhancement of carbon were selected , resulting in a determination of the upper limit of carbon enhancement , without restricting the metallicity range .
these data allow for the determination of reasonably accurate estimates of carbon abundances ( and [ fe / h ] ) for the christlieb et al .
carbon - rich stars , using techniques described by rossi et al . @xcite .
the calibration of rossi et al .
, based on the kp and gp indices described by beers et al .
@xcite , in combination with de - reddened @xmath9 colors obtained from the 2mass survey @xcite , achieves accuracies of @xmath10 dex for [ fe / h ] and @xmath11 dex for [ c / fe ] , respectively .
figure 1 shows the derived abundances that have been obtained to date for a subset of the christlieb et al .
carbon - rich stars .
roughly 50% of the carbon - enhanced candidates proved to be very metal - deficient stars with [ fe / h ] @xmath12 , in other words , they are cemp stars .
the remainder appear to be stars of higher metallicity , up to and including solar .
these latter stars may be examples of intermediate - mass intrinsic agb stars , whose presence in the halo of the galaxy might well be accounted for by stripping from dwarf galaxies such as sagittarius @xcite . with declining [ fe / h ] .
these stars were selected as carbon - enhanced , independent of metallicity .
the upper envelope of abundances correspond to [ c / h ] = @xmath13 .
note that abundances have not yet been determined for stars more metal - rich that [ fe / h ] = @xmath13.,width=377,height=226 ] in addition to elemental abundances , radial velocities have been obtained for most of these candidates .
around half of these stars have velocites consistent with membership of the general halo population .
the remainders are a mix of high- and low - velocity stars , some of which are likely members of the metal - weak thick disk @xcite , while others are possibly associated with the sagittarius stream . | very metal - deficient stars that exhibit enhancements of their carbon abundances are of crucial importance for understanding a number of issues the nature of stellar evolution among the first generations of stars , the shape of the initial mass function , and the relationship between carbon enhancement and neutron - capture processes , in particular the astrophysical s - process .
one recent discovery from objective - prism surveys dedicated to the discovery of metal - deficient stars is that the frequency of carbon - enhanced metal - poor ( cemp ) stars increases with declining metallicity , reaching roughly 25% for [ fe / h ] @xmath0 . in order to explore this phenomenon in greater detail
we have obtained medium - resolution ( 2 ) spectroscopy for about 350 of the 413 objects in the christlieb et al .
catalog of carbon - rich stars , selected from the hamburg / eso objective prism survey on the basis of their carbon - enhancement , rather than metal deficiency . based on these spectra , and near - ir @xmath1 photometry from the 2mass point source catalog ,
we obtain estimates of [ fe / h ] and [ c / fe ] for most of the stars in this sample , along with reasonably accurate determinations of their radial velocities .
of particular importance , we find that the upper envelope of carbon enhancement observed for these stars is nearly constant , at [ c / h ] @xmath2 , over the metallicity range @xmath3 [ fe / h ] @xmath4 ; this same level of [ c / h ] applies to the most iron - deficent star yet discovered , he 0107 - 5240 , at [ fe / h ] = -5.3 . |
the be / x - ray binaries represent the largest subclass of high - mass x - ray binaries .
about two - thirds of the identified systems fall into this category .
these systems consist of a be star ( i.e. , a b star with an equatorial disc ) and , generally , a neutron star .
the orbit is wide ( several tens of days @xmath2 several hundred days ) and eccentric ( @xmath3 ) .
most of the be / x - ray binaries show only transient x - ray activity due to transient accretion of the circumstellar matter of the be star , while some show persistent x - ray emission . each be / x - ray binary exhibits some or all of the following three types of x - ray activity : 1 .
periodic ( type i ) x - ray outbursts , coinciding with periastron passage ( @xmath4 ) , 2 .
giant ( type ii ) x - ray outbursts ( @xmath5 ) , which show no clear orbital modulation , 3 .
persistent low - luminosity x - ray emission ( @xmath6 ) ( @xcite ; see also @xcite ) .
these features imply a complicated interaction between the be - star envelope and the neutron star .
a be star has a two - component extended atmosphere , a polar region and a cool ( @xmath7 ) equatorial disc .
the polar region consists of a low - density , fast ( @xmath8 ) outflow emitting uv radiation .
the wind structure is well explained by the so - called line - driven wind model , in which the radiative acceleration results from the scattering of the stellar radiation in an ensemble of spectral lines ( @xcite ; @xcite ) . on the other hand , the equatorial disc , which is geometrically thin and nearly keplerian , consists of a high - density plasma from which the optical emission lines and the ir excess arise .
the radial velocity of the disc is smaller than a few @xmath9 , at least within @xmath10 stellar radii ( @xcite ; @xcite ; @xcite ) .
although there is no widely accepted model for discs around be stars , the viscous decretion disc model proposed by @xcite explains many of the observed features and thus seems promising ( @xcite ; see also @xcite ) . in this model ,
the matter supplied from the equatorial surface of the star drifts outwards because of the viscous effect and forms the disc .
the basic equations for viscous decretion discs are the same as those for viscous accretion discs , except that the sign of @xmath11 ( mass decretion / accretion rate ) is opposite .
the boundary conditions for decretion discs , however , are different from those for accretion discs
. therefore , the decretion disc has a structure different from that of the accretion disc @xcite . until quite recently
, models for type i x - ray outbursts in be / x - ray binaries had assumed a large disc around the be star so that the neutron star can accrete gas when it passes through the disc near periastron .
however , @xcite and @xcite recently performed a semi - analytical study based on the viscous decretion disc model for be stars and showed that the be disc in be / x - ray binaries is truncated at a radius smaller than the periastron distance , as long as @xmath0 , where @xmath12 is the shakura - sunyaev viscosity parameter ( see fig . [
fig : bex ] for a schematic view of a be / x - ray binary ) .
the truncation of the disc is due to the resonant torque exerted by the neutron star , which removes the angular momentum from the disc .
the disc radii they obtained for seven particular systems ( 4u0115 + 63 , v0332 + 53 , a0535 + 262 , exo2030 + 375 , 2s1417@xmath13624 , groj1008@xmath1357 , and 2s1417@xmath13624 ) are consistent with the x - ray behaviour of those systems . moreover , the result is in agreement with the result of @xcite that there is a positive correlation between the orbital size and the maximum equivalent width of h@xmath14 ever observed in a system , a measure of the maximum disc size around the be star in the system .
the truncation of the be disc in be / x - ray binaries is not surprising .
the resonant interaction is important in various contexts even in a fly - by encounter with a perturber . in fly - by encounters between a disc galaxy and a point mass perturber ,
the energy is always transported from the disc to the perturber through the resonant interaction , except for overhead encounters where the energy transfer is small @xcite . in distant encounters between a circumstellar accretion disc and a perturbing mass with @xmath15 , where @xmath16 and @xmath17 are the periastron distance and disc radius , respectively , the disc material loses energy and angular momentum to the perturber s orbit through a resonance feature @xcite . in the case of
be / x - ray binaries , the surface density of the be disc is expected to increase more rapidly than that for isolated be stars , as a consequence of truncation .
this qualitatively agrees with the result found by @xcite that the be discs in be / x - ray binaries are about twice as dense as those around isolated be stars .
the disc may finally become optically thick , and become unstable to radiation - driven warping ( @xcite ; see also @xcite ) .
multi - wavelength , long - term monitoring observations of v635 cas , the optical counterpart of 4u0115 + 63 , revealed that the be disc in 4u0115 + 63 undergoes a quasi - cyclic ( @xmath18 ) dynamical evolution @xcite : after each disc - loss episode , the disc starts reforming , grows until it becomes unstable to warping , and after that a type ii outburst occurs .
although a direct link between the warped disc and the type ii outburst is still missing , the dynamical evolution of the be disc is likely to be the agent that controls the x - ray behaviour of the system .
this way , the truncated disc model , at least qualitatively , explains many of the observed features of be / x - ray binaries . the semi - analytical model adopted by @xcite and @xcite , however , only compares the resonant torque integrated over the whole orbit with the viscous torque to determine at which radius the disc is truncated .
hence , it can not make a quantitative prediction about how perfect or imperfect the truncation is .
moreover , it predicts nothing about phase - dependent features , such as the disc deformation and the change in the mass capture rate .
therefore , in order to study the efficiency of the resonant truncation and the orbital - phase dependence of the interaction , we simulate the interaction between the be - star disc and the neutron star in be / x - ray binaries , using a 3d sph code . in a general context
, such simulations will also enable us to study the interaction between the viscous disc and the companion in an eccentric orbit . in this paper , which is the first of a series of papers dedicated to understanding the interaction between the
be disc and the neutron star , we study the effects of viscosity on disc truncation in a coplanar system .
simulations presented here were performed with a three - dimensional , smoothed particle hydrodynamics ( sph ) code .
the sph code is based on a version originally developed by benz ( @xcite ; @xcite ) .
the smoothing length is variable in time and space .
the code uses a tree structure to calculate the nearest neighbours of particles .
the sph equations with the standard cubic spline kernel are integrated using a second - order runge - kutta - fehlberg integrator with individual time steps for each particle @xcite , which results in an enormous computational saving when a large range of dynamical time - scales are involved . in our code , the be disc is modelled by an ensemble of gas particles , each of which has a negligible mass chosen to be @xmath19 with a variable smoothing length . for simplicity
, the gas particles are assumed to be isothermal at the temperature of @xmath20 , where @xmath21 is the effective temperature of the be star . on the other hand ,
the be star and the neutron star are modelled by two sink particles @xcite with corresponding masses .
gas particles which fall within a specified accretion radius are accreted by the sink particle .
we assume that the be star has the accretion radius of @xmath22 , @xmath23 being the radius of the be star . for the neutron star
, we adopt a variable accretion radius of @xmath24 , where @xmath25 is the roche - lobe radius for a circular binary .
this is because having a small accretion radius is time - consuming and it is unphysical to adopt an accretion radius smaller than the smoothing length of the particles near the disc outer radius .
an approximate formula for @xmath25 is given by @xmath26 ( e.g. , @xcite ) with the mass ratio @xmath27 , where @xmath28 and @xmath29 are the masses of neutron star and be star , respectively , and the distance between the stars , @xmath30 .
the sph artificial viscosity , @xmath31 , between particles @xmath32 and @xmath33 enters the momentum equation as @xmath34 where @xmath35 is the velocity , @xmath36 is the mass , @xmath37 is the pressure , @xmath38 is the density , @xmath39 is the standard cubic spline kernel , @xmath40 is the distance between particles @xmath32 and @xmath33 , @xmath41 is the mean of the smoothing lengths of particles @xmath32 and @xmath33 , and @xmath31 has the following standard form , @xmath42 @xcite , where @xmath43 and @xmath44 are the linear and nonlinear artificial viscosity parameters , respectively , @xmath45 , @xmath46 , @xmath47 is the isothermal sound speed , and @xmath48 with @xmath49 . in the continuum limit ,
the viscous force @xmath50 in the above sph formalism is written as @xmath51 \label{eq : fv1}\ ] ] @xcite , where @xmath52 in the 3d sph code with the cubic spline kernel and @xmath53 is the deformation tensor .
if we assume that the density varies on a length - scale much longer than the velocity , we have @xmath54 .
\label{eq : fv2}\ ] ] this implies that the shear viscosity @xmath55 and the bulk viscosity @xmath56 are given by @xmath57 and @xmath58 respectively , in the continuum limit of the 3d sph formalism . on the other hand , in the shakura - sunyaev viscosity prescription , the shear viscosity @xmath55 is written as @xmath59 where @xmath60 is the scale - height of the disc . from equations ( [ eq : nusph ] ) and ( [ eq : nuss ] )
, we have the relation between the sph artificial viscosity parameter @xmath43 and the shakura - sunyaev viscosity parameter @xmath12 as @xmath61 if we assume that @xmath62 . in general flows , however , @xmath63 .
moreover , the viscosity is artificially turned off for divergent flows in our model [ see equation ( [ eq : pi ] ) ] .
therefore , equation ( [ eq : alphass ] ) should be taken as a rough approximation to the relation between @xmath12 and @xmath43 .
in several simulations we report in this paper , we adopted @xmath64 and @xmath65 , in which case @xmath66 is variable in time and space . in the other simulations , however , we adopted constant values of @xmath12 in order to roughly model the @xmath14 disc . in these simulations , @xmath67 was variable in time and space and @xmath68 .
the mass ejection mechanism from the be star is not known . in our simulations , it is modelled by injecting gas particles at @xmath69 . in our normal resolution simulations , which finally have @xmath70 particles , we take @xmath71 ,
while in a high - resolution simulation we performed with @xmath72 , which finally has about @xmath73 particles , we take @xmath74 .
these values were adopted so that @xmath75 , where @xmath76 is a typical smoothing length near the stellar surface . in order to avoid an additional complexity
, we kept the injection rate constant in each simulation .
once injected , gas particles interact with each other . as a result ,
most of the injected particles fall onto the be star by losing the angular momentum and a small fraction of particles drift outwards , getting the angular momentum from the other particles . as the be star , we take a b0v star with @xmath77 , @xmath78 , and @xmath79 , which has the typical spectral type of be / x - ray binaries . with these parameters ,
the disc scale - height @xmath60 is @xmath80 at @xmath81 and increases as @xmath82 . for the neutron star , we take @xmath83 and @xmath84 .
we set the binary orbit on the @xmath85-@xmath86 plane with the major axis along the @xmath85-axis . at @xmath87 ,
the neutron star is at apastron .
it orbits about the be star primary with the orbital period @xmath88 and the eccentricity @xmath89 .
the angle of misalignment , @xmath90 , between the binary orbital plane and the be disc plane is set to 0 in this paper .
the unit of time is @xmath91 , unless noted otherwise .
each simulation is very time - consuming .
it takes three to four weeks to perform each of the normal - resolution simulations , which runs on a single processor of an hp exemplar v2500 , and about four months to perform the high - resolution simulation , which runs efficiently on eight processors of an sgi origin 3800 .
this is particularly so for long - period systems .
thus , in this paper , we present results for only a restricted range of parameter space as the first step in the study of the interaction between the be disc and the neutron star .
we fix the orbital period @xmath91 , the orbital eccentricity @xmath89 , and the misalignment angle @xmath92 , allowing only the viscosity parameter to vary .
we adopt @xmath93 and @xmath94 , targeting 4u0115 + 63 , one of the best studied be / x - ray binaries .
then , the semi - major axis @xmath95 and @xmath25 given by equation ( [ eq : roche ] ) ranges between @xmath96 at periastron and @xmath97 at apastron . before attempting the binary simulations , we applied the code to an isolated be star in order to test whether it gives reliable results compared to those obtained by using a one - dimensional diffusion - type equation for the surface density .
as mentioned above , we adopt the viscous decretion disc model , which yields a geometrically thin , nearly keplerian disc around the be star . for simplicity , we assume that the disc around an isolated be star is axisymmetric and keplerian and in hydrostatic equilibrium in the vertical ( @xmath86- ) direction .
the evolution of such a disc is described by @xmath98 \label{eq : sigma1d}\ ] ] with @xmath99 ( e.g. , @xcite ) , where @xmath100 is the surface density , @xmath101 is the radial velocity , and @xmath102 is the angular frequency of disc rotation . in order to create a situation similar to that in sph simulations , we inject mass at a constant rate at @xmath71 . at the inner boundary @xmath81
, we adopt the torque - free boundary condition , @xmath103 .
we also take @xmath103 as the outer boundary condition at @xmath104 , which affects the disc structure only in a region near @xmath104 .
the evolution of @xmath100 for @xmath105 and @xmath72 for the initial three years is shown by dashed lines ( @xmath106 1yr , 2yr and 3yr from left ) in the left panels of figs .
[ fig : isolated_a1 ] and [ fig : isolated_a01 ] , respectively .
the surface density is measured in units of @xmath107 , where @xmath108 is the highest local density at @xmath109yr normalised by @xmath110 , a typical value for be stars @xcite .
it should be noted that , in this framework , no steady solution is present for decretion discs unlike for accretion discs @xcite .
instead , the formal solution of equation ( [ eq : sigma1d ] ) with @xmath111 and @xmath112 gives the disc structure at @xmath113 , which is given by @xmath114 in our isothermal disc model .
using the model described in @xmath115[sec : sph_code ] , we performed two 3d sph simulations of the disc evolution around an isolated be star with @xmath105 and @xmath72 . in these simulations , @xmath43 is variable in time and space and @xmath68 to keep @xmath12 constant , as described in the previous section . figs .
[ fig : isolated_a1 ] and [ fig : isolated_a01 ] show the surface density evolution ( left ) and the disc structure at @xmath116yr ( right ) for @xmath105 and @xmath72 , respectively .
the time interval between adjacent contours in the left panel is 1yr . in the right panel ,
the solid , the dashed , and the dash - dotted lines denote the surface density , the azimuthal velocity normalised by the stellar critical velocity , and the radial mach number , respectively . in the figures ,
the density is integrated vertically and averaged azimuthally , while the velocity components are averaged vertically and azimuthally .
the dotted line shows the vertically and azimuthally averaged distribution of @xmath117 required to keep @xmath12 constant . from figs .
[ fig : isolated_a1 ] and [ fig : isolated_a01 ] , we observe that our 3d sph code is capable of producing the disc evolution similar to that in the 1d model , except that for small viscosity the disc evolves a bit faster than in the 1d model .
the disc structure is almost keplerian and the radial velocity , which decreases with time , is very subsonic for @xmath118 at @xmath119yr .
these features are in agreement with the observed characteristics of be stars . for the purpose of comparison , we present in fig . [
fig : isolated_sph2 ] the result from a simulation with @xmath64 and @xmath120 . in this simulation
, @xmath12 is variable in time and space and is roughly proportional to @xmath121 . in the initial phase of disc formation
, the density distribution has a very steep slope in the radial direction , so that @xmath12 stays high except for a region near the injection radius .
as time goes on , @xmath12 in the outer region decreases , because of the increase in the density there . as a result
, @xmath12 has a local maximum near the star , which , in turn , causes a dip in the density distribution , the feature not seen in simulations with constant @xmath12 .
[ fig : isolated_sph2 ] also shows that the disc structure outside the dip ( @xmath122 ) is , in a rough sense , similar to that in simulations with constant @xmath12 .
since our purpose is to study the interaction between the be disc and the neutron star in a system , of which the periastron distance is @xmath123 , the difference in the disc structure close to the star does not matter very much . in the following sections , we will see that both kinds of models give similar results . in our model , in which the mass injection rate is kept constant , the net mass - loss rate from the star through the disc decreases as the disc mass increases .
this is because a larger fraction of injected particles must lose the angular momentum and fall back to the star to support a larger disc .
[ fig : isosph_mdot ] gives the change in the mass - loss rate from the star , @xmath124 , ( thin line ) measured at @xmath125 and the disc mass @xmath126 ( thick line ) for the simulations shown in figs .
[ fig : isolated_a1 ] and [ fig : isolated_a01 ] . to reduce the fluctuation noise , @xmath124
is averaged over @xmath127 days .
as expected , a larger viscosity makes the decrease in @xmath124 faster .
from the figure , we note that , for the first several years , @xmath128 several @xmath129 for a wide range of the viscosity parameter .
it is interesting to compare the model mass - loss rate with the observed one . the observed equatorial mass - loss rate is , however , a poorly determined quantity .
it has been measured only during the disc - formation phase for several stars . among them , x per is the only one star for which both the equatorial mass - loss rate and the highest disc density are known .
@xcite studied the long - term behaviour of the be disc of x per ( 4u0352 + 30 ) , a be / x - ray binary system with @xmath130 and @xmath131 , and found that the equatorial mass - loss rate for a disc build - up phase of less than 230d was greater than @xmath132 and that for the following 380d was @xmath133 .
they also found that the base density of the be disc of x per was as high as @xmath134 .
it should be noted that the model mass - loss rate shown in fig .
[ fig : isosph_mdot ] is in good agreement with the observed equatorial mass - loss rate of x per .
the simulations shown above assume a continuous mass supply from the be star for studying the formation and evolution of a persistent disc .
some be stars exhibit a transient disc formation by an episodic mass loss .
such a situation was investigated by @xcite .
they studied the evolution of the gas explosively ejected from a be star to model the transient disc formation and disc decay , using a 3d sph code .
they found that the gaseous particles form a nearly keplerian disc in the viscous time - scale . because of the difference between the boundary conditions adopted in this paper and by @xcite , the structure of our persistent disc is different from that of the decaying disc studied by @xcite
in the previous section , we have seen that our model with a 3d sph code is capable of reproducing the results from 1d simulations of formation and evolution of the disc around isolated be stars and explaining the observed equatorial mass - loss rate from the be star . in this section , we apply our model to a be / x - ray binary with @xmath135 and @xmath94 , assuming coplanarity between the be disc and the binary .
the semi - major axis of the binary , @xmath136 , is @xmath137 ( @xmath138 ) .
we have run simulations with @xmath105 for @xmath139 ( @xmath140 ) , by which time the disc almost reaches an equilibrium state , while we have run the other simulations for @xmath141 ( @xmath142 ) . in all simulations but one , the number of sph particles at the end of the simulation was about @xmath143 . in order to confirm the reliability of the results obtained by those simulations
, we performed a simulation with @xmath72 with a higher resolution , in which the number of sph particles at the end of the simulation was about @xmath144 . [
cols="^,^,^,^,>,>,^,^,^,^,^ " , ] [ tab : summary ] @xcite investigated the tidal / resonant truncation of circumstellar and circumbinary discs in eccentric binaries and found that a gap is always formed between the disc and the binary orbit .
following their formulation , @xcite and @xcite showed that the be disc in be / x - ray binaries is truncated via the resonant interaction with the neutron star as long as @xmath0 . in the following ,
we investigate the resonant interaction between the be disc and the neutron star in more detail , by analysing the results from 3d sph simulations .
table [ tab : summary ] lists some characteristic quantities from these simulations , which will be discussed below . figs .
[ fig : sd_a1 ] and [ fig : sd_a01 ] show the disc evolution under the influence of the neutron star for @xmath105 and @xmath72 , respectively .
the truncation of the disc is obviously more evident for @xmath72 than for @xmath145 . from fig .
[ fig : sd_a01 ] , we clearly see how the resonant torque works on a viscous decretion disc . when the disc size is small ( @xmath146 for @xmath72 ) , the disc evolution is almost identical to that around an isolated be star . as the disc grows , however , the effect of the resonant torque from the neutron star becomes apparent and the radial density distribution begins to have a break at a radius around the 4:1 to 5:1 resonance radii ( for @xmath72 ) .
we call this radius the truncation radius .
since the resonant torque prevents disc material from drifting outwards , the disc density increases more rapidly than in discs around isolated be stars . outside the truncation radius ,
the density decreases rapidly .
the wavy patterns seen in the surface density distribution in the left panel and in the radial velocity distribution in the right panel of fig .
[ fig : sd_a01 ] are due to the tightly wound spiral density wave excited in the disc .
in contrast to the low - viscosity simulation shown in fig.[fig : sd_a01 ] , the resonant torque has little effect on the disc structure when the viscosity is very high . as seen in the left panel of fig.[fig : sd_a1 ] , there is an only very modest break in the surface density distribution for @xmath105 . in this simulation
, the disc almost reaches an equilibrium state at @xmath147 ( @xmath148 ) in the sense that the disc structure varies regularly , depending on the orbital phase .
we also performed a simulation with @xmath149 .
the disc structure obtained is something between those with @xmath1 and @xmath105 .
there is a clear break and a wavy feature in the radial surface - density distribution but they are not so strong as those for @xmath72 . for comparison , we present in fig . [
fig : sd_copl2 ] the result from a simulation with @xmath64 and @xmath65 , in which @xmath12 is variable in time and space . from fig .
[ fig : sd_copl2 ] , we observe that the disc with constant artificial viscosity parameters evolves in a similar way to that of the @xmath14-disc with a similar viscosity parameter , except for the presence of the dip in the density distribution close to the star , as was expected from the previous section . in order to study the interaction between the be disc and the neutron star in more detail , we performed a high - resolution simulation with @xmath72 . in this simulation ,
the number of sph particles is about an order of magnitude larger than , and so the smoothing length is on average less than a half of , that of other simulations .
[ fig : sd_ukaff ] shows the surface density evolution for @xmath150 and the disc structure at @xmath151 in this high - resolution simulation ( unfortunately , the allocated computing time ran out at @xmath152 ) . from fig .
[ fig : sd_ukaff ] , we easily see more detailed disc structure than that in the corresponding simulation with a lower resolution shown in fig .
[ fig : sd_a01 ] .
the wavy pattern in the surface density distribution caused by the spiral density wave is more remarkable in the high - resolution simulation .
this is because a larger number of particles give a higher resolution of the interacting region , which makes the interaction more localised and stronger .
although the surface density profile has breaks near the 5:1 resonance radius ( @xmath153 ) and the 4:1 radius ( @xmath154 ) , the steep density decrease already begins at a much smaller radius , which coincides with the outermost density peak of the spiral wave . in the rest of this paper , we mainly present the results from this high - resolution simulation as the representative ones with @xmath1 , because it gives a better understanding of the star - disc interaction .
most be / x - ray binaries with known orbital parameters have orbital eccentricities in the range from 0.3 to 0.9 .
in such systems , the star - disc interaction is likely to be strongly phase - dependent . in this subsection , we discuss phase - dependent features except for the mass capture rate by the neutron star , which will be discussed separately .
figs.[fig : ss_a1 ] and [ fig : ss_a01 ] give snapshots covering one orbital period for @xmath105 and @xmath72 , respectively .
each panel shows the surface density in a range of about two orders of magnitude in the logarithmic scale . from these figures , we note a remarkable difference in the disc structure between @xmath105 and @xmath72 . for @xmath145 ,
the disc has a significant density up to the periastron distance and experiences a strong interaction at and after the periastron passage of the neutron star . on the other hand , for @xmath72 ,
the resonant torque from the neutron star is much more effective at truncating the disc than for the high viscosity disc .
the sharp decline in the disc density outside the 5:1 resonance causes a gap between this radius and the periastron distance , apparently reducing the mass capture rate by the neutron star , as will be seen in section [ sec : capture ] . in both cases ,
the spiral density waves are clearly seen between the periastron passage and the apastron passage .
the opening angle of the spirals , which is related to the effective gravity in the disc , is smaller for a larger binary separation . through the resonant interaction ,
the angular momentum is transported from the disc to the binary .
we have to admit that the effect of the angular momentum transport on the binary turned out to be much stronger than we had expected .
despite the fact that the mass of each particle is only @xmath155 so that the disc mass is only about @xmath156 in our simulations , the increase in the binary orbital period is visible in the late stage of simulations . the computed phase lags behind the correct orbital phase by @xmath157 at @xmath158 in the simulation with @xmath105 and by @xmath159 and @xmath160 at @xmath161 in the high and normal resolution simulations with @xmath1 , respectively . in the following , figures should be read taking this phase shift into account . in order to have a measure of the disc radius ,
i.e. , the radius at which the disc density has a major break , we applied a non - linear least - square fitting method to the radial distribution of the azimuthally - averaged surface density @xmath100 , adopting the following simple fitting function , @xmath162 where @xmath163 and @xmath164 are constants and @xmath165 is the disc radius .
[ fig : rd_copl ] shows the phase dependence of the disc radius ( thick line ) in the simulations shown in figs.[fig : ss_a1 ] and [ fig : ss_a01 ] . to reduce the fluctuation noise , we folded the data on the orbital period over @xmath166 for @xmath145 and @xmath167 for @xmath72 . in the figure ,
the horizontal solid lines denote some of the lowest @xmath168 resonance radii ( the 2:1 , the 3:1 , @xmath169 , the 10:1 from top ) , while the thin sinusoidal line denotes the orbit of the neutron star .
the origin of the phase is at the periastron passage of the neutron star . from fig .
[ fig : rd_copl ] , we note that the disc radius coincides with the 4:1 resonance radius ( @xmath170 ) for @xmath105 , whereas the disc has a radius intermediate between the 4:1 radius and the 5:1 radius ( @xmath171 ) for @xmath72 ( the mean @xmath165 is @xmath172 for @xmath105 and @xmath173 for @xmath72 ) .
the latter is a typical feature expected for a disc in which the wave damps locally @xcite .
we also note that the disc radius modulates around the mean value .
the disc shrinks after the periastron passage of the neutron star , which gives a negative torque to the disc .
after that it restores its radius by viscous diffusion , so that the amplitude of the modulation is larger for @xmath105 than for @xmath72 . since the disc structure depends on the binary phase , the disc emission is expected to exhibit an orbital modulation as well . to see whether this is the case , we calculated the continuum flux from the be disc for @xmath174 , assuming that the disc is optically thin and the emissivity is proportional to @xmath175 , where @xmath38 is the local density . for simplicity
, we ignored the effect of the obscuration of the disc by the star , which will become important for systems with high inclination angles .
we then obtained a base flux curve by performing the cubic - spline fitting of the fluxes at apastron passages .
the base flux describes the long - term change in the continuum flux . finally , we computed the orbital modulation by subtracting the base flux from the instantaneous fluxes .
[ fig : lc_copl ] shows the orbital modulation in the optically thin continuum from the be disc for @xmath105 ( left ) and @xmath72 ( right ) . for @xmath72 ,
the modulation is more clearly seen in the high - resolution simulation ( thick line ) than in the normal one ( thin line ) . in order to reduce the fluctuation noise , we folded the data on the orbital period over the period annotated in the panel .
contrary to what is expected from fig.[fig : rd_copl ] , the continuum exhibits little modulation for @xmath105 , and about one per cent of modulation is seen for @xmath72 .
the negative result for @xmath105 suggests that the strongly - perturbed outer disc contributes little to the optically thin emission , because of its low density . on the other hand , the positive result for @xmath72 , in particular in the high - resolution simulation , though the amplitude is still very small , seems to come from the region where the density is enhanced by the spiral wave , because the disc radius does not change significantly and the rise and the subsequent decline of disc emission are in phase with the density enhancement and the subsequent damping caused by the spiral density wave .
the difference between the modulation patterns from the high and normal resolution simulations with @xmath1 suggests that resolving the spiral density wave is important to obtain a reliable modulation pattern of such low amplitude .
although fig .
[ fig : lc_copl ] gives little observability of the orbital modulation in optically thin disc emission , it is still likely that optically thick emission , such as balmer lines , will show significant orbital modulation .
calculating the optically thick emission from the disc is , however , beyond the scope of this paper . in a circumbinary disc around an eccentric binary , an eccentric mode is excited through direct driving due to a one - armed bar potential @xcite . in this subsection
, we study whether the same mechanism works in be / x - ray binaries .
we analyse the evolution of the eccentricity in the be disc by decomposing the surface density distribution into fourier components which vary as @xmath176 , where @xmath177 and @xmath178 are the azimuthal and time - harmonic numbers , respectively , and @xmath179^{1/2}$ ] is the frequency of mean binary rotation .
following @xcite , we define the mode strength by @xmath180 where @xmath181 and @xmath182 are either @xmath183 or @xmath184 functions .
the surface density here is computed by summing up @xmath185 functions at particle positions , not by taking the kernel into account as has been done in the previous sections .
then , the total strength of the mode @xmath186 is defined by @xmath187 fig .
[ fig : mode_a1 ] shows the excitation and precession of the ( 1,0 ) mode ( i.e. , the eccentric mode ) for @xmath105 .
the upper panel shows the strengths of several modes , while the lower panel shows the evolution of the angle @xmath188 between the eccentric vector of the disc and that of the binary orbit defined by @xmath189 the eccentric mode grows initially linearly in time ( @xmath190 ) , as is predicted by the theory and was also found by @xcite . at @xmath191
the strength of the eccentric mode is saturated at @xmath192 .
as it is saturated , the mode stops precessing and is locked at @xmath193 . in order to study the structure of the eccentric mode in more detail ,
we analyse the orbits of individual particles
. the position and velocity of each particle are instantaneously equal to those of an elliptical keplerian orbit of semi - latus rectum @xmath194 and the eccentric vector @xmath195 with the amplitude @xmath196 and the longitude of periastron , @xmath197 , with respect to the stellar periastron ( e.g. , @xcite ) .
the radial coordinate @xmath198 is then given by @xmath199 given @xmath200 , @xmath201 in our simulations .
the upper panels of fig .
[ fig : ecc_lowres ] present the distribution of the orbital parameters of disc particles for @xmath105 at @xmath202 . from the left panel
, we note that the disc eccentricity increases almost linearly in @xmath194 for @xmath203 and is roughly constant for @xmath204 .
the right panel for the radial dependence of the longitude of periastron shows that the eccentric mode is twisted in a trailing sense . and
the longitude of periastron , @xmath197 , of the particle orbits from the normal - resolution simulations with @xmath105 at @xmath202 ( upper ) and @xmath72 at @xmath205 ( lower ) .
the gray - scale plot in each panel shows the particle distribution in linear scale . @xmath206 and @xmath207 are mean values of @xmath208 and @xmath209.,scaledwidth=40.0% ] and the longitude of periastron , @xmath197 , of the particle orbits from the normal - resolution simulations with @xmath105 at @xmath202 ( upper ) and @xmath72 at @xmath205 ( lower ) . the gray - scale plot in each panel shows the particle distribution in linear scale . @xmath206 and @xmath207 are mean values of @xmath208 and @xmath209.,scaledwidth=40.0% ] the normal - resolution simulation with @xmath72 showed a similar trend .
the eccentric mode grew initially linearly in time ( @xmath210 ) .
the mode strength was almost saturated at @xmath211 at @xmath212 .
the mode also stopped precessing and was locked at @xmath213 .
the distribution of the orbital parameters of disc particles at @xmath205 from this simulation is presented in the lower panels of fig .
[ fig : ecc_lowres ] . from the left panel ,
we note that the eccentricity has a maximum at @xmath214 .
the right panel shows that the eccentric mode is twisted in a trailing sense , as in the case for @xmath105 .
recently , @xcite has developed a non - linear theory of the evolution of the shape and surface density of a three - dimensional eccentric accretion disc .
when the eccentricity of the disc is small so that terms of relative order @xmath215 may be neglected , this theory provides a linear evolutionary equation for the complex eccentricity @xmath216 of the disc .
we have reworked this theory for the case of a strictly isothermal decretion disc with no radial velocity , and also evaluated the tidal forcing terms associated with a companion object of small eccentricity .
the governing equation is then of the form @xmath217 , \label{eq : dedt}\end{aligned}\ ] ] where @xmath218 and @xmath219 are the ( constant ) eccentricity and semi - latus rectum of the binary orbit , @xmath220 is the laplace coefficient of celestial mechanics , and @xmath221 .
the dimensionless coefficients @xmath222 and @xmath223 are given by @xmath224 where @xmath225 are dimensionless complex constants that depend only on the shear and bulk viscosity parameters @xmath12 and @xmath226 . in the limit @xmath227 these coefficients become purely imaginary and give rise to precession induced by the pressure of the disc .
most importantly , the real part of @xmath228 , which has the role of a diffusion coefficient for short - wavelength eccentric perturbations , turns out to be positive when @xmath229 , as is true in the sph simulations [ see equation ( [ eq : nubsph ] ) ] .
if this condition were not satisfied , the disc would experience a local eccentric instability and equation ( [ eq : dedt ] ) could not be evolved forward in time .
equation ( [ eq : dedt ] ) has the character of a dispersive and diffusive linear wave equation for @xmath230 .
the eccentricity of the binary provides a tidal forcing that is independent of time . starting from an initially circular state @xmath231
, the eccentricity of the disc first grows linearly in time and then approaches a steady state .
the steady eccentric shape can be determined either by solving the time - dependent problem until the transient response decays , or by solving the second - order ordinary differential equation obtained by setting the time - derivative to zero .
we have performed both of these calculations , adopting the boundary condition @xmath231 at the stellar surface , a ` stress - free ' boundary condition @xmath232 at a notional outer edge ( @xmath233 ) , and a surface density distribution @xmath234 .
the steady configurations of @xmath235 and @xmath236 for the cases @xmath105 and @xmath72 are illustrated in fig .
[ fig : ecc_model ] .
one should not expect a perfect agreement with the sph simulations because the theory is based on the assumption of small eccentricities and only approximates the actual surface density distribution .
however , the steady eccentric shapes based on linear theory offer a fair explanation of the results of the sph simulations , especially for the high - viscosity case . and longitude of periastron @xmath197 of the particle orbits for @xmath105 ( upper ) and @xmath72 ( lower ) .
the notional outer edge is at @xmath237.,scaledwidth=40.0% ] and longitude of periastron @xmath197 of the particle orbits for @xmath105 ( upper ) and @xmath72 ( lower ) .
the notional outer edge is at @xmath237.,scaledwidth=40.0% ] fig .
[ fig : mode_a01 ] shows the evolution of an eccentric mode from the high - resolution simulation with @xmath72 .
the detailed structure of the mode is shown in fig .
[ fig : ecc_a01hi ] by snapshots covering one precessional period . from these figures
, we note that the evolution and the structure of the eccentric mode are very similar to those from the corresponding normal resolution simulation until @xmath238 . at @xmath238 ,
the eccentric mode suddenly begins to precess in a prograde sense , as seen in the lower panel of fig.[fig : mode_a01 ] . as @xmath188 increases from @xmath239 to @xmath240 , the mode strength increases .
the mode stagnates at @xmath241 for @xmath242 around @xmath243 , suggesting that the eccentric disc at @xmath244 is an unstable configuration .
the precession decelerates before reaching @xmath245 and accelerates after that . at the same time
, the mode changes its shape . as seen in the lower panel of fig .
[ fig : ecc_a01hi ] , the eccentric mode is twisted in a trailing sense for @xmath246 and in a leading sense after that . at @xmath247 ,
apsidal axes of all particle orbits align . as seen in fig .
[ fig : ecc_a01hi ] , the twist of the mode increases , as the apsidal axis of the disc moves toward the stellar periastron ( @xmath248 ) .
the strong radial dependence of the phase caused by the twist reduces the eccentricity of the disc , as shown in the upper panel of fig .
[ fig : mode_a01 ] .
the precession is fastest at @xmath193 , suggesting that there is a stable point at @xmath193 .
after that , the mode becomes trailing and its strength increases .
the precessional period is about @xmath249 .
it is important to note that a similar behaviour is found in circumbinary discs around eccentric binaries . according to @xcite ,
this behaviour occurs as follows : when the eccentricity of the disc edge is small , @xmath188 is locked at a stable value @xmath250 .
however , as the eccentricity grows , the locking action weakens , and the prograde precession due to the quadrupole moment of the binary potential dominates .
the disc edge begins to precess when its eccentricity becomes @xmath251 , and afterwards the eccentricity oscillates with a precessional period .
the disc typically attains the eccentricity of @xmath252 .
we note that the above behaviour of the eccentric mode in circumbinary discs around eccentric binaries is strikingly similar to that shown in fig .
[ fig : mode_a01 ] , except that , in our simulation for be / x - ray binaries , the growth and precessional time - scales of the eccentric mode are much shorter and the disc eccentricity attained is significantly smaller than in circumbinary discs . ,
but for the high - resolution simulation with @xmath72.,scaledwidth=35.0% ] , but for the high - resolution simulation with @xmath72.,scaledwidth=35.0% ] as mentioned in section [ sec : intro ] , most of the be / x - ray binaries show transient x - ray activities . among them , some exhibit periodical x - ray outbursts called type i , which coincide with the periastron passage , while the others show occasional giant x - ray outbursts called type ii and little or no detectable x - ray emission in quiescent phase . 4u0115 + 63 , the system we are modelling in this paper , belongs to the latter group . in this subsection ,
we first study how much mass is captured by the neutron star in a general context , and then discuss whether our model predicts the mass capture rate consistent with the observed x - ray behaviour of 4u0115 + 63 .
[ fig : mdot_a1 ] shows the change in the mass capture rate by the neutron star , @xmath253 , and the disc mass @xmath126 for @xmath105 .
the upper panel shows the evolution of @xmath253 and @xmath126 , while the lower panel shows the orbital - phase dependence of @xmath253 . in the lower panel ,
we folded @xmath253 on the orbital period over @xmath254 to reduce the fluctuation noise .
the horizontal dashed line and dash - dotted line denote the mean mass capture rate by the neutron star and the mean mass - loss rate from the be star , respectively .
we have already seen that there is little truncation of the be disc for @xmath105 .
[ fig : mdot_a1 ] confirms this . for @xmath255
the be disc is almost in equilibrium in the sense that the disc mass and the mass capture rate only shows regular orbital modulations with constant amplitude . for @xmath166 , the mean mass - loss rate from the be star is @xmath256 , while the mean mass capture rate for the same period is @xmath257 .
thus , in this high viscosity disc , the neutron star captures the disc mass at about the same rate as the mass - loss rate from the be star .
and the thick line denotes the disc mass @xmath126 . in the lower panel ,
the data are folded on the orbital period over @xmath166 .
the horizontal dashed line and the dash - dotted line in the lower panel denote the mass capture rate by the neutron star and the mass - loss rate from the be star averaged over the period annotated at the bottom of the panel.,scaledwidth=35.0% ] the simulation with @xmath149 also gives a high fraction ( about 60 per cent for @xmath258 ) of mass capture rate with respect to the mass - loss rate , as listed in table [ tab : summary ] .
this indicates that the resonant truncation effect is not effective for @xmath149 , either .
in contrast to the simulations with @xmath105 and @xmath149 , the high - resolution simulation with @xmath1 revealed a subtle feature in the mass capture rate , as shown in fig . [
fig : mdot_a01 ] .
before the precession of the eccentric mode began , the mass capture rate @xmath253 increased monotonically : @xmath259 before @xmath260 with the resolution of this simulation .
then , the peak mass capture rate showed a gradual increase to a level at @xmath261 at @xmath262 .
after the eccentric mode began to precess , @xmath253 exhibited a long - term modulation , depending on the longitude of the eccentric mode , @xmath188 .
it gradually increased as @xmath188 increased from 0 to @xmath263 and decreased as @xmath188 increased from @xmath263 to @xmath240 . at the periastron passage at @xmath264
, @xmath253 was a maximum of @xmath265 , which was about a factor of three higher than the level before precession .
the eccentricity of the disc was @xmath266 at @xmath241 and @xmath267 at @xmath193 .
it should be noted that even this small eccentricity caused a factor of three enhancement in the mass capture rate .
if the disc had a much stronger disturbance , the enhancement in @xmath253 is likely to be much larger .
in addition to the long - term modulation due to the precession of the eccentric mode , we note that the mass capture rate by the neutron star is much smaller and more strongly phase - dependent for @xmath1 than for @xmath105 . for @xmath72 ,
the phase at which the neutron star captures the disc mass most slightly lags behind the periastron passage , because of the presence of a gap between the disc outer radius and the periastron . in the high resolution simulation ,
the peak mass capture rate for @xmath268 is @xmath269 , which is one order of magnitude smaller than that for @xmath105 .
the mass capture rate then decreases by two orders of magnitude by the apastron passage . even in the normal resolution simulation ,
in which the disc density around the truncation radius is significantly higher than that in the high - resolution simulation , the peak mass capture rate is about a factor of three smaller than that for @xmath145 , and decreases by a factor of fifty by apastron . note that similar , strongly phase - dependent accretion was found in simulations by @xcite for circumbinary discs around eccentric binaries . in the high - resolution simulation with @xmath72 , the neutron star , on average ,
captures the disc mass at the rate of @xmath270 for @xmath268 , while the mean mass - loss rate from the be star for the same period is @xmath271 .
this indicates that , even after three years of disc growth , about 6/7 of the gas lost from the star still accumulates in the disc , making the disc continually denser . , but for the high - resolution simulation with @xmath72 . in the lower panel ,
the data are folded on the orbital period over @xmath272 . for comparison ,
the mass capture rate from the normal - resolution simulation with @xmath72 ( thin solid line ) is overlapped in the lower panel.,scaledwidth=35.0% ] the transient nature in the x - ray activity of be / x - ray binaries is considered to result from the interactions between the accreted material and the rotating magnetised neutron star @xcite . if the accreted material is dense enough to make the magnetospheric radius smaller than the corotation radius , the accretion onto the neutron star causes a bright x - ray emission ( the direct accretion regime ) . otherwise , the magnetospheric radius is larger than the corotation radius , and the accretion onto the neutron star is prevented by the propeller mechanism .
the system is then in quiescence ( the propeller regime ) .
recently , @xcite found that 4u0115 + 63 showed a dramatic x - ray luminosity variation from @xmath273 to @xmath274 in less than 15hr close to periastron , whereas it showed a nearly constant x - ray luminosity at a level of @xmath275 near apastron .
they concluded that the system was in the transition regime between the direct accretion regime and the propeller regime close to periastron and in the propeller regime near apastron , because the direct accretion regime and the propeller regime , respectively , apply for @xmath276 and @xmath277 , where @xmath278 is a parameter which depends on the geometry and physics of accretion .
below we try to compare our simulation results with the above criteria by @xcite .
although it is likely that the captured material will form an accretion disc around the neutron star , we do not know how the accretion rate is related to the mass capture rate shown in figs .
[ fig : mdot_a1 ] and [ fig : mdot_a01 ] , which is based on the moments at which particles enter the variable accretion radius of the neutron star .
therefore , we consider two extreme situations , in which @xmath279 or @xmath280 , where @xmath281 is the accretion time - scale . in the former situation ,
the accretion rate profile is approximately the same as the profile of mass capture rate , whereas in the latter situation , the variation in the accretion rate will be much smaller than that in the mass capture rate .
we assume that the x - ray luminosity is given by @xmath282 with @xmath283 . for @xmath105 , @xmath284 at periastron if @xmath285 and all the captured mass accretes onto the neutron star .
note that this level of luminosity enters the direct accretion regime .
since 4u0115 + 63 is considered to be in the direct accretion regime only in occasional giant x - ray outbursts , we conclude that the @xmath105 model is inconsistent with the observation if @xmath279 .
if @xmath280 , the neutron star will emit the x - ray luminosity corresponding to the mean mass capture rate of @xmath286 ( see the bottom panel of fig . [
fig : mdot_a1 ] ) .
the x - ray luminosity is then @xmath287 .
this level of luminosity enters the transition regime .
hence , the @xmath105 model is not ruled out by the observed constraint if @xmath280 .
we have the same conclusion for the @xmath149 model .
since it gives the mass capture rate about a half of that for @xmath105 , the model is inconsistent with the observation if @xmath279 , but is not ruled out if @xmath288 . on the other hand , for @xmath72 , @xmath289 at periastron even if @xmath290 and all the captured material accretes onto the neutron star .
this suggests that the system is in the transition regime even at periastron .
it is likely that the low mass capture rate in this simulation put the system into the propeller regime in most of the orbital phases .
thus , the @xmath72 model for 4u0115 + 63 is consistent with the observed x - ray behaviour , irrespective of the accretion time - scale .
in this paper , we have presented results from 3d sph simulations of the disc formation around isolated be stars and of the interaction between the be star disc and the neutron star in be / x - ray binaries , based on the viscous decretion disc model for be stars @xcite . in several simulations we adopted constant values of artificial viscosity parameters @xmath43 and
@xmath44 , for which the shakura - sunyaev viscosity parameter @xmath12 is variable in time and space . in the other simulations , we adopted constant values of @xmath12 , for which @xmath43 is variable in time and space and @xmath68 .
we preferred constant @xmath12 simulations because the analysis of the results becomes easier . in all simulations ,
the be disc was nearly keplerian and the radial velocity was highly subsonic .
these features are consistent with those observed for be discs . the simulations of isolated be stars showed that our code is capable of producing results similar to those from the 1d simulations .
the simulated mass - loss rate from the be star in the first several years of disc formation was several @xmath291 for a wide range of viscosity parameter , which is consistent with the observed equatorial mass - loss rate .
here , @xmath108 is the highest local density at @xmath292yr normalised by @xmath110 , a typical value for be stars . in binary simulations ,
we have studied the effect of viscosity on the star - disc interaction in the case of a coplanar system with @xmath135 and @xmath94 , the parameters for 4u0115 + 63 , one of the best studied be / x - ray binaries .
we have chosen these orbital parameters because they enable us to compare the simulation results with the observed ones and the short orbital period makes the computing time bearable .
some of the results from these simulations are summarised in table [ tab : summary ] .
our simulations showed that there is a radius outside which the azimuthally averaged surface density decreases steeply .
for a smaller @xmath12 , the slope outside this radius is steeper , giving a stronger truncation effect on the disc . among the simulations with @xmath105 , 0.3 , and 0.1
, we found that the resonant truncation of the be disc is effective only for @xmath72 .
for @xmath105 and 0.3 , the neutron star captures the disc mass at a rate comparable to the mass - loss rate from the be star .
these results confirm the previous semi - analytical result by @xcite and @xcite on disc truncation that the resonant truncation is effective for a disc with @xmath0 .
the truncation radius for @xmath72 roughly agrees with that derived semi - analytically . our simulations , in particular the high - resolution simulation with @xmath72 , showed how the disc grows under the influence of the neutron star .
in the initial build - up phase , the disc evolution is similar to that for isolated be stars .
but , later on , the effect of the resonant torque becomes apparent , preventing the disc gas from drifting outwards at several resonance radii .
the effect is most remarkable at the 4:1 and 5:1 radii ( @xmath293 and 0.33 , respectively ) for @xmath72 . as a result ,
the disc density increases more rapidly than that for isolated be stars .
this feature is consistent with @xcite , who found that the discs in be / x - ray binaries are about twice as dense as those of isolated be stars . since the neutron star orbits in an eccentric orbit about the be star , the interaction is phase dependent .
the disc shrinks a little at periastron and then recovers gradually .
consequently , the disc emission will vary with phase . for our system with @xmath94
, this effect turned out not to be large enough to cause an observable variation in the disc continuum as long as it is optically thin .
it is possible , however , that the orbital modulation in the disc continuum in systems with much higher orbital eccentricity is observable .
moreover , it is likely that the profiles of optically thick emission lines from the disc , such as balmer lines , will show an orbital modulation even in systems with moderate eccentricity , because the disc structure is made non - axisymmetric by the periodic excitation and damping of the spiral density wave . in the be disc in be / x - ray binaries
, an eccentric mode is excited through direct driving due to the ( 1,0 ) harmonic of the binary potential . in high - viscosity simulations
, the mode grows initially linearly in time and then approaches a steady state , of which steady eccentric shapes based on linear theory of the evolution of a 3d eccentric decretion disc described in section [ sec : ecc_mode ] offer a fair explanation .
the strength of the mode in the steady state is larger for a smaller value of @xmath12 . on the other hand , in the high - resolution simulation with @xmath72 ,
the eccentric mode undergoes prograde precession at some point .
the precession period is about @xmath294 .
the precession rate is not constant .
it accelerates for @xmath295 and decelerates for @xmath296 , where @xmath188 is the angle between the eccentric vector of the disc and that of the binary orbit .
the precession rate is radius - dependent . it is larger at a larger radius .
since the eccentric mode is leading for @xmath297 and trailing for @xmath298 , the twist of the mode increases with time for @xmath295 and decreases for @xmath298 .
the strength of the eccentric mode , which anti - correlates with the twist of the mode , decreases for @xmath295 and increases for @xmath299 . as the mode precesses , the mass capture rate , @xmath253 , by the neutron star modulates .
it is a maximum at @xmath300 .
even for an eccentricity of 0.1 , @xmath253 at maximum is about a factor of three higher than the level before precession .
if the disc had a much stronger disturbance , the enhancement in @xmath253 is likely to be much larger .
such a system may temporarily show periodic x - ray outbursts . in our model for be / x - ray binaries , in which the disc material overflows toward the neutron star via the @xmath301 point , the mass capture rate by the neutron star becomes strongly phase dependent . the dependence is stronger for a smaller value of viscosity . in the disc with @xmath72 ,
the mass capture rate decreases by two orders of magnitude between periastron and apastron passages , and after apastron passage , no disc mass is captured by the neutron star . note
that our model gives a much stronger contrast in the mass capture rate than that expected for the stellar wind accretion .
we have also compared the simulated mass capture rate with the observed x - ray behaviour of 4u0115 + 63 , considering two extreme situations , in which @xmath279 or @xmath288 , where @xmath281 is the accretion time - scale .
we found that the disc model for @xmath72 gives a result consistent with the observation for both situations , whereas the higher viscosity models for @xmath105 and 0.3 are ruled out unless @xmath280 .
analysing multi - wavelength long - term monitoring observations of 4u0115 + 63 , @xcite found that the be star undergoes quasi - cyclic ( @xmath302 ) activity , losing and reforming its circumstellar disc .
they also found that , at some point , the growing disc becomes unstable to warping and then tilts and starts precessing .
type ii x - ray outbursts take place after the warping episode . as shown in this paper
, our @xmath72 model explains many of the observed features of 4u0115 + 63 in the phase before the warping occurs .
our simulation , however , showed no dynamical instability , and therefore is incapable of explaining the warping episode . including the effect of radiation from the be star may turn out to be essential to have a model that explains the whole cycle of the disc evolution , which is beyond the scope of this paper . in this paper , we have concentrated our study on the viscous effects on star - disc interaction in a coplanar system with fixed orbital period and eccentricity .
this was done not only as a first step to have a comprehensive understanding of the interaction between the be disc and the neutron star in be / x - ray binaries , but also to have an archetypal model with which we can compare the results from our future simulations . in the next paper , we will study the effects of the misalignment angles between the be disc and the binary . we will discuss the effects of the orbital eccentricity in the third paper , and the fourth paper will conclude the series by studying the effects of the orbital period .
we thank the anonymous referee for constructive comments .
ato acklowledges the institute of astronomy , cambridge , uk for the warm hospitality .
the high - resolution simulation reported here was performed using the uk astrophysical fluids facility ( ukaff ) .
the other simulations were done using the facility at the hokkaido university computing center , japan .
this work was supported in part by grant - in - aid for scientific research ( 13640244 ) of japan society for the promotion of science .
abbott d.c . , 1982 ,
apj 259 , 282 artymowicz p. , lubow s.h . , 1994 , apj , 421 , 651 artymowicz p. , lubow s.h . , 1996a , in beckwith s. , staude j. , quetz a. , natta a. , ed .
disks and outflows aound young stars , springer , berlin , p.115 artymowicz p. , lubow s.h . , 1996b , apj , 467 , l77 bate m. r. , bonnell i.a . , price n.m . , 1995 , mnras , 285 , 33 benz w. , 1990 , in buchler j.r . , ed , the numerical modelling of nonlinear stellar pulsations
kluwer , dordrecht , p.269 benz w. , bowers r.l . ,
cameron a.g.w .
, press w.h . , 1990 ,
apj , 348 , 647 campana s. , gastaldello f. , stella l. , israel , g.l . , colpi m. , pizzolato f. , orlandini m. , dal fiume d. , 2001 , apj , 561 , 924 castor j.i . ,
abbott d.c . ,
klein r.i . , 1975 , apj 195 , 157 hall s.m . ,
clarke c.j .
, pringle j.e . , 1996 , mnras , 278 , 303 hanuschik r.w . , 1994 , in balona l.a .
, henrichs h. , le contel j.m . , ed .
pulsation , rotation and mass loss of early - type stars , iau symp .
162 , kluwer academic publishers , dordrecht , p.265 hanuschik r.w . , 2000 , in smith m.a .
, henrichs h.f . , fabregat j. , ed .
be phenomenon in early - type stars , iau colloq .
185 , asp , san francisco , p.518 kroll p. , hanuschik r.w . , 1997 , in wickramasinghe d.t . , bicknell g.v .
, ferrario l. , ed . accretion phenomena and related outflows , iau colloq .
163 , asp , san francisco , p.494 lee u. , saio h. , osaki y. , 1991 , mnras , 250 , 432 lubow s.h . , 1991 , apj , 381 , 268 lubow s.h . , artymowicz p. , 2000 , in mannings v. , boss a.p . , russell s.s . ,
protostars and planets iv , univ . of arizona press , p.731 meglicki z. , wickramasinghe d. , bicknell g.v . , 1993 , mnras , 264 , 691 monaghan j.j . ,
gingold r.a . , 1983 , j. comput .
phys . , 52 , 374 negueruela i. , okazaki a.t . , 2001 , a&a , 369 , 108 negueruela i. , reig p. , coe m.j . , fabregat j. , 1998 , a&a , 336 , 251 negueruela i. , okazaki a.t .
, fabregat j. , coe m.j . , munari u. , tomov t. , 2001 , a&a , 369 , 117 ogilvie g.i . , 2001 , mnras , 325 , 231 okazaki a. t. , 2001 , pasj , 53 , 119 okazaki a.t . , negueruela i. , 2001a , a&a , 377 , 161 okazaki a.t . , negueruela i. , 2001b , in giacconi r. , serio s. , stella l. , ed .
x - ray astronomy 2000 , asp conf . series , vol .
234 , 2001 , asp , san francisco , p.281 palmer p.l . , 1983 , mnras , 204 , 507 porter j.m . , 1998 , a&a , 336 , 966 porter j.m . , 1999 , a&a , 348 , 512 pringle j.e . , 1981 ,
ara&a , 19 , 137 pringle j.e . , 1991 , mnras , 248 , 754 pringle j.e . , 1996 , mnras , 281 , 357 reig p. , fabregat j. , coe m.j
, 1997 , a&a , 322 , 193 stella l. , white n.e . , rosner r. , 1986 , apj , 208 , 669 telting j.h . , waters l.b.f.m . ,
roche p. , boogert a.c.a .
, clark j.s . , de martino d. , persi p. , 1998 ,
a&a , 296 , 785 warner b. , 1995 , cataclysmic variable stars .
cambridge university press , cambridge waters l.b.f.m .
, marlborough j.m . , 1994 , in balona l.a .
, henrichs h. , le contel j.m . , ed .
pulsation , rotation and mass loss of early - type stars , iau symp .
162 , kluwer academic publishers , dordrecht , p.399 waters l.b.f.m . , taylor a.r .
, van den heuvel e.p.j , habets g.m.h.j , persi p. , 1988 ,
a&a , 198 , 200 zamanov r.k .
, reig p. , mart j. , coe m.j .
, fabregat j. , tomov , n.a . , valchev t. , 2001 , a&a , 367 , 884 | we study the viscous effects on the interaction between the coplanar be - star disc and the neutron star in be / x - ray binaries , using a three - dimensional , smoothed particle hydrodynamics code . for simplicity , we assume the be disc to be isothermal at the temperature of half the stellar effective temperature . in order to mimic the gas ejection process from the be star , we inject particles with the keplerian rotation velocity at a radius just outside the star . both be star and neutron star are treated as point masses .
we find that the be - star disc is effectively truncated if the shakura - sunyaev viscosity parameter @xmath0 , which confirms the previous semi - analytical result . in the truncated disc , the material decreted from the be star accumulates , so that the disc becomes denser more rapidly than if around an isolated be star .
the resonant truncation of the be disc results in a significant reduction of the amount of gas captured by the neutron star and a strong dependence of the mass capture rate on the orbital phase .
we also find that an eccentric mode is excited in the be disc through direct driving due to a one - armed bar potential of the binary .
the strength of the mode becomes greater in the case of a smaller viscosity . in a high - resolution simulation with @xmath1
, the eccentric mode is found to precess in a prograde sense .
the mass capture rate by the neutron star modulates as the mode precesses . [ firstpage ] accretion , accretion discs binaries : close hydrodynamics instabilities stars : emission - line , be x - rays : stars . |
the drinfeld doubles of bosonizations of braided hopf algebras over abelian groups , and their quotients by central group - likes , are known in the folklore as _
quantum groups_. such is the case of the _ quantum enveloping algebra @xmath5 _ or the _ small quantum group @xmath6 _ @xcite .
these quantum groups have been intensely studied , both their intrinsic structures and their representation theories . however , to the best of our knowledge , there is no research which contemplates non - abelian groups .
the purpose of our work is to give a first step in this direction .
more precisely , let @xmath2 be a finite non - abelian group and @xmath7 a yetter - drinfeld module over @xmath8 with finite - dimensional nichols algebra @xmath1 ( we will work over an algebraically closed field @xmath9 of characteristic zero ) .
we denote by @xmath0 the drinfeld double of the bosonization @xmath10 and call it _ `` quantum group at a non - abelian group '' _ recall that @xmath5 is called _ quantum group at a root of @xmath11 _ if the indeterminate @xmath12 is specialized to a root of @xmath11 @xcite . in this work
, we deal with the category of representations of @xmath0 .
we use the methods coming from the theory of lie algebras which were also used in the study of quantum groups over abelian groups .
we find similarities as well as differences between our results and their analogues in the context of lie algebras or the mentioned quantum groups .
to explain these similarities and differences , we first recall briefly the situation in those frameworks .
assume that @xmath13 is either an enveloping algebra of a lie algebra as in ( * ? ? ?
* chapter 7 ) or a quantum group as in ( * ? ? ?
* chapter 5 ) , ( * ? ? ?
* chapter 3 ) ; the reader can find all the details of the following exposition in these chapters . roughly speaking
, @xmath13 has a distinguished commutative and cocommutative hopf subalgebra @xmath14 .
hence 1 .
[ intro : weights ] the maximal spectrum of @xmath14 is an abelian group @xmath15 .
the elements of @xmath15 , the algebra maps @xmath16 , are called _
weights_. the module corresponding to the weight @xmath17 is denoted @xmath18 .
[ intro : prod of weights ] the product of @xmath15 is implemented by tensoring .
that is , the tensor product of @xmath18 and @xmath19 is the module @xmath20 . also , @xmath13 admits a _ triangular decomposition _ , that means that there are subalgebras @xmath21 and @xmath22 such that the multiplication @xmath23 gives a linear isomorphism .
indeed , @xmath13 is a @xmath24-graded algebra such that the degrees of @xmath14 , @xmath21 and @xmath22 are zero , negative and positive , respectively .
let @xmath25 be the subalgebra generated by @xmath14 and @xmath22 .
given a weight @xmath26 , this can be seen as an @xmath25-module by letting @xmath27 act trivially on it .
we denote it again by @xmath28 .
let @xmath29 be an @xmath13-module and @xmath30 its _ weight space _ of weight @xmath26 .
we restrict our attention to the @xmath13-modules which decompose as the direct sum of their weight spaces .
for instance , @xmath13 regarded as a module with respect to the _ adjoint action_. then 1 .
[ intro : decomp weights ] @xmath31 for all weights @xmath17 and @xmath26 .
an @xmath13-module @xmath29 is called a _ highest - weight module _
( of weight @xmath26 ) if it is generated by an element @xmath32 such that @xmath33 .
notice that @xmath34 by the triangular decomposition of @xmath13 .
the basic examples of highest - weight modules are the _ verma modules _
these are essential in the study of the representation theory of @xmath13 because 1 .
[ intro : verma ] every verma module has a unique simple quotient and every simple @xmath13-module is a quotient of a unique verma module .
[ intro : verma free ] every verma module is a free @xmath36-module of rank @xmath11 .
other important features of the verma modules can be found in the references above .
let us consider now a quantum group @xmath0 at a finite non - abelian group @xmath2 .
the role of @xmath14 shall be played by the drinfeld double @xmath3 of @xmath8 .
this is a semisimple but not commutative hopf subalgebra of @xmath0 .
we will see that @xmath0 admits a triangular decomposition @xmath37 where @xmath38 denotes the dual object of @xmath7 in the category of @xmath3-modules and @xmath39 is its nichols algebra . in this
setting the bosonization @xmath40 shall play the role of @xmath25 .
we will calculate the commutation rules between the generators of @xmath3 , @xmath7 and @xmath38 , and deduce that @xmath0 is a @xmath24-graded algebra with homogeneous spaces @xmath41 the classification of the simple @xmath3-modules is well - known , see for instance ( * ? ? ?
* subsection 3.1 ) ; unlike , there are simple modules of dimension greater than one .
the simple @xmath3-modules are parametrized by pairs @xmath42 , where @xmath43 is a conjugacy class in @xmath2 and @xmath44 is an irreducible representation of the centralizer of a fixed @xmath45 . if @xmath46 denotes the corresponding simple @xmath3-module , cf .
, then it becomes a @xmath47-module by letting @xmath39 act trivially on it .
therefore we can define the verma modules for a quantum group at a non - abelian group as the induced modules @xmath48 thus @xmath49 is a free @xmath1-module of rank @xmath50 , compare with .
our main result asserts that holds true in our context , _
i. e. _ every verma module @xmath49 has a unique simple quotient and every simple @xmath0-module is a quotient of a unique verma module , theorem [ teo : bi con l ] .
therefore we obtain a bijective correspondence @xmath51 where @xmath52 denotes the head of @xmath49 .
moreover , we prove that the socle @xmath53 of @xmath49 is simple what provides another bijective correspondence between the set of simple @xmath3-modules and the set of simple @xmath0-modules , theorem [ teo : bi con socle ] .
we also give a criterion to decide whether or not a verma module is simple , corollary [ cor : verma simple xtop ] , and show that the socle and the head are related by @xmath54 where @xmath55 and @xmath56 is the homogeneous component of maximum degree of @xmath1 , theorem [ teo : l dual s ] ; recall that @xmath56 is one - dimensional . in order to compute explicitly the simple @xmath0-modules
, we have to study the submodules of the verma modules .
this is done in the abelian case using the properties and among others , which allow to obtain remarkable results under certain general assumptions . although the @xmath0-modules decompose as the direct sum of simple @xmath3-modules , our situation is more complex because and do not hold true . here
the tensor product between simple @xmath3-modules is not necessarily simple and hence we have to know their fusion rules .
we give a general strategy to compute the highest - weight submodules of any @xmath0-module @xmath29 .
we use that @xmath0 is a @xmath24-graded @xmath3-module with respect to the adjoint action , which respects the triangular decomposition , and the fact that the action @xmath57 is a morphism of @xmath3-modules , [ subsec : algorithm ] .
we carry out this strategy to compute the simple modules in a concrete example in section [ sec : s3 ] as we summarize below .
the first genuine example of a finite - dimensional nichols algebra over a non - abelian group is the fomin - kirillov algebra @xmath58 @xcite .
it is isomorphic to the nichols algebra @xmath1 of the yetter - drinfeld module @xmath59 over @xmath60 .
the action and coaction on @xmath7 are @xmath61 for any transposition @xmath62 and @xmath63 @xcite where @xmath64 denotes the sign map .
let now @xmath0 be the drinfeld double of @xmath65 .
as an algebra , @xmath0 is generated by @xmath66 we shall see that @xmath67 as @xmath68-modules .
these elements are subjected to the next relations : @xmath69 @xmath70 @xmath71 for all transpositions @xmath62 , @xmath72 and @xmath73 . on the other hand
, the simple @xmath0-modules are parametrized by the simple @xmath68-modules according to our main result .
let @xmath74 and @xmath75 be permutations in @xmath4 . then @xmath76 , @xmath77 and @xmath78 are the conjugacy classes of @xmath4 and @xmath68 has eight non - isomorphic simple modules .
namely , @xmath79 we recall the structures of them and their fusion rules in [ subsub : s3 sigma simples][subsub : fusion rules ] .
we compute the lattice of submodules of the corresponding verma modules @xmath49 and classify the simple @xmath0-modules .
in particular , we prove that * @xmath80 as @xmath68-modules and @xmath81 , corollary [ cor : l e mas cociente de t ] . *
@xmath82 as @xmath68-modules and @xmath83 , corollary [ cor : l e rho cociente de vqn ] .
* @xmath84 as @xmath68-modules and @xmath85 , corollary [ cor : l tau cero cociente de upo ] .
* @xmath86 as @xmath68-modules and @xmath87 , theorem [ teo : el latice de sigma menos ] . * the verma modules @xmath88 , @xmath89 , @xmath90 and @xmath91 are simple , theorem [ teo : los simples ] .
their dimensions are @xmath92 , @xmath93 , @xmath93 and @xmath94 , respectively . as @xmath68-modules they are the tensor product of @xmath1 with the associated simple @xmath68-module .
we finish by pointing out other facts about these modules . *
the head @xmath52 and the socle @xmath53 are isomorphic , theorem [ teo : los simples ] , lemma [ le : socle sigma menos ] and corollaries [ cor : l e mas cociente de t ] , except to @xmath95 * the simple @xmath0-modules are self - duals except to @xmath96 * @xmath97 has submodules which are not homogeneous , lemma [ le : t ] , and its maximal submodule is not generated by highest - weight submodules , theorem [ teo : el latice de sigma menos ] .
the first author was financially supported by capes - brazil .
the second author was partially supported by anpcyt - foncyt , conicet and secyt ( unc ) .
this work was carried out in part during the visit of the first author to the university of crdoba ( argentina ) .
she would like to thank the faculty of mathematics , astronomy and physics for its warm hospitality and support . both authors are grateful to nicols andruskiewitsch for drawing their attention to this problem and also for so many suggestions .
the second author thanks vyacheslav futorny for very useful conversations during his visit to the university of so paulo under the framework of the math
amsud program .
we also thank the referee for the careful reading of our article and for providing constructive comments to improve the exposition of this paper .
through this work @xmath9 denotes an algebraically closed field of characteristic zero .
the dual of a vector space @xmath7 will be denoted by @xmath98 .
if @xmath99 and @xmath100 , then @xmath101 denotes the evaluation of @xmath102 in @xmath103 .
let @xmath104 be a set .
we write @xmath105 for the free vector space on @xmath104 .
let @xmath106 be an algebra . by an @xmath106-module ,
we mean a left @xmath106-module .
if @xmath104 is a subset of an @xmath106-module @xmath29 and @xmath107 , then @xmath108 denotes the set of all @xmath109 with @xmath110 and @xmath111 .
let @xmath112 be a finite - dimensional hopf algebra .
we denote by @xmath113 , @xmath114 and @xmath115 the comultiplication , the antipode and the counit of @xmath112 .
we will use the sweedler notation @xmath116 for the comultiplication of any @xmath117 , and for the coaction @xmath118 of an element @xmath119 belonging to an @xmath112-comodule .
recall that @xmath120 denotes the category of yetter - drinfeld modules over @xmath112 , whose objects are the @xmath112-modules and @xmath112-comodules @xmath29 such that for every @xmath117 and @xmath121 it holds that @xmath122 [ [ subsub : dh ] ] we consider the drinfeld double @xmath123 of @xmath112 according to ( * ? ? ?
* theorem 7.1.1 ) .
namely , @xmath123 is @xmath124 as coalgebra .
meanwhile , the multiplication and the antipode are given by @xmath125 in consequence , we have that @xmath112 and @xmath126 are hopf subalgebras of @xmath123 .
recall that the category @xmath120 is braided equivalent to the category @xmath127 of @xmath123-modules .
namely , if @xmath128 , then @xmath29 is a @xmath123-module by setting @xmath129 for every @xmath117 , @xmath130 and @xmath121 .
[ [ subsub : bv ] ] the nichols algebra of @xmath131 is constructed as follows , see for instance ( * ? ? ?
* @xmath1322.1 ) .
first , we consider the tensor algebra @xmath133 as a graded braided hopf algebra in @xmath120 by defining @xmath134 for all @xmath99 .
let @xmath135 be the maximal ideal and coideal of @xmath133 generated by homogeneous elements of degree @xmath136 .
then the nichols algebra of @xmath7 is the quotient @xmath137 which is a graded braided hopf algebra in @xmath120 .
its homogeneous component of degree @xmath138 will be denoted @xmath139 .
note that @xmath140 and @xmath141 .
moreover , if @xmath1 is finite - dimensional , then its homogeneous component of maximum degree is one - dimensional and it is the space of left and right integrals , see for instance ( * ? ? ?
* @xmath1322.3 and @xmath1323.2 ) .
[ [ subsub : boson ] ] the bosonization @xmath142 @xcite is the hopf algebra structure defined on @xmath143 in such a way that @xmath112 is a hopf subalgebra , @xmath1 is a subalgebra , @xmath144 it is a graded hopf algebra where its homogeneous component of degree @xmath138 is @xmath145 . [ [ subsub : g ] ] let @xmath2 be a finite group .
the unity element of @xmath2 is denoted by @xmath146 .
we set @xmath147 , the dual hopf algebra of the group algebra @xmath8 .
let @xmath148 be the dual basis of the canonical basis @xmath149 of @xmath8 .
the comultiplication of an element @xmath150 is @xmath151 let @xmath29 be a @xmath152-module and @xmath153 .
then @xmath29 is @xmath2-graded with homogeneous component of degree @xmath154 : @xmath155=\delta_g m=\left\{m\in m\mid f\cdot m = f(g)m\quad\forall f\in{\bbbk}^g\right\}.\ ] ] if @xmath156 , we set @xmath157=s\cap m[g]$ ] .
we denote by @xmath158 the subset of @xmath2 formed by those elements whose homogeneous component in @xmath29 is non - zero .
the one - dimensional @xmath152-module of degree @xmath154 will be denoted @xmath159 .
if @xmath152 is a subalgebra of @xmath106 , then we will consider @xmath106 as a @xmath152-algebra with the adjoint action , that is @xmath160 for any @xmath161 and @xmath162 .
we denote by @xmath3 the drinfeld double of @xmath8 . since @xmath152 is a commutative algebra ,
@xmath152 and @xmath8 are hopf subalgebras of @xmath3 .
then the algebra structure of @xmath3 is completely determined by @xmath163 we will define verma modules in @xmath132[subsec : verma ] by inducing from the simple @xmath3-modules .
these are well - known because they are equivalent to the simple objects in @xmath164 and a description of these last can be found for instance in ( * ? ? ?
* subsection 3.1 ) .
we recall this description but in the context of modules over @xmath3 .
let @xmath165 be the conjugacy class of @xmath153 , @xmath166 the centralizer of @xmath154 and @xmath167 an irreducible representation of @xmath166 .
the @xmath8-module induced by @xmath167 , @xmath168 is also a @xmath152-module if we define the action by @xmath169 then @xmath170 is of @xmath2-degree @xmath171 and @xmath172 .
note that @xmath173 . therefore @xmath46 is a @xmath3-module .
moreover , @xmath46 is simple and every simple @xmath3-module is of this form by ( * ? ? ?
* proposition 3.1.2 ) .
a @xmath3-module is _ of weight @xmath174 _ if it is isomorphic to @xmath46 .
let @xmath4 be the group of bijections on @xmath175 .
we set @xmath74 and @xmath75 .
these two cycles generate @xmath4 and satisfy the relations @xmath176 and @xmath177 .
the conjugacy classes of @xmath4 are @xmath178 next , we describe the simple @xmath68-modules which we will consider in @xmath132[sec : s3 ] .
the centralizer @xmath179 is just the cyclic subgroup generated by @xmath180 .
then @xmath179 has only two irreducible representations : the trivial one and the induced by the sign map @xmath64 .
therefore the simple @xmath68-modules attached to @xmath180 are @xmath181 let us consider the set of symbols @xmath182 as a basis of @xmath183 .
sometimes we write @xmath184 instead of @xmath185 , if @xmath186 , and omit the subscript if there is no place for confusion .
hence the action of @xmath68 on @xmath183 is defined in such a way that @xmath184 has @xmath4-degree @xmath187 and @xmath188 from now on , we fix a root of the unit @xmath189 of order @xmath190 .
the centralizer @xmath191 is the cyclic subgroup generated by @xmath192 .
then @xmath191 has ( up to isomorphisms ) three irreducible representations .
these are given by the group maps @xmath193 , @xmath194 for @xmath195 .
therefore the simple @xmath68-modules attached to @xmath192 are @xmath196 let us consider the set of symbols @xmath197 as a basis of @xmath198 .
sometimes we write @xmath199 instead of @xmath200 , if @xmath201 , and omit the subscript if there is no place for confusion .
hence the action of @xmath68 on @xmath198 is defined in such a way that @xmath202 is of @xmath4-degree @xmath203 and @xmath204 for @xmath205 .
it is not difficult to check that @xmath206 is an isomorphism of @xmath60-modules .
let @xmath207 be the map defining the two - dimensional specht @xmath4-module .
then @xmath208 , @xmath209 and @xmath210 is a complete list of non - isomorphic irreducible @xmath4-modules .
therefore the simple @xmath68-modules attached to @xmath146 are @xmath211 these are concentrated in @xmath4-degree @xmath146 .
the modules @xmath212 are one - dimensional , we denote by @xmath213 its generators and omit the subscript if there is no place for confusion
. we can describe the @xmath60-action on @xmath214 using the canonical representation of @xmath60 on the vector space spanned by @xmath215 .
in fact , @xmath216 decomposes into the direct sum @xmath217 where the submodules of weight @xmath208 and @xmath210 are spanned by @xmath218 and @xmath219 , respectively .
another special basis of @xmath214 is the set of symbol @xmath220 where @xmath221 this basis is special because it gives the following isomorphisms of @xmath60-modules @xmath222 we omit the subscript in @xmath223 if there is no place for confusion .
let @xmath224 and @xmath225 be simple @xmath68-modules .
we want to decompose the tensor products @xmath226 into a direct sum of simple @xmath68-modules .
first , we have a decomposition into the direct sum of two submodules which are not necessarily simple : @xmath227{{\otimes}}n[g]\right)\oplus\left(\bigoplus_{g , h\in{{\mathbb s}}_3,\,g\neq h}w[g]{{\otimes}}n[h]\right).\end{aligned}\ ] ] note that the first submodule is zero if @xmath228 .
this decomposition is useful for us because each submodule has a basis which is a transitive @xmath4-set in the sense of the next lemma .
let @xmath229 and @xmath230 be the bases of @xmath224 and @xmath225 given in @xmath132[subsub : s3 sigma simples ] , [ subsub : s3 tau simples ] and [ subsub : s3 e simples ] . then the sets @xmath231{{\otimes}}{\mathbf{b}}_n[g]\quad\mbox{and}\quad{\mathbf{b}}_2=\bigcup_{g , h\in{{\mathbb s}}_3,\ , g\neq h}{\mathbf{b}}_w[g]{{\otimes}}{\mathbf{b}}_n[h]\end{aligned}\ ] ] are bases of the first submodule and the second one in , respectively .
[ le : transitive action ] if @xmath232 , @xmath233 , then there is @xmath234 such that @xmath235 for some non - zero scalar @xmath26 . the sets @xmath236 $ ] and @xmath237 $ ] , @xmath63 ,
are either empty or have only one element @xmath238 except to @xmath214 , but in this case the basis is @xmath239 . in these bases , we see from the definition that @xmath240 for some non - zero scalar @xmath26 .
we conclude by remarking that @xmath4 acts transitively by conjugation on the sets @xmath241 and @xmath242 . as a consequence of the above lemma we have the next remark wich
will be useful in @xmath132[sec : s3 ] where the action of @xmath7 , or @xmath38 , on @xmath225 will play the role of @xmath243 .
[ obs : description submodule gen by a submodule 1 ]
let @xmath244 be a map of @xmath68-modules .
assume there is @xmath245 , respectively @xmath246 , such that @xmath247 .
hence @xmath243 restricted to @xmath248{{\otimes}}n[g]$ ] , respectively @xmath249{{\otimes}}n[h]$ ] , is zero since @xmath4 acts transitively on the basis @xmath250 , respectively @xmath251 .
we next list the precise fusion rules only for those tensor products which will appear in section [ sec : s3 ] .
we give the assignments ( or describe the submodules ) which realize the listed isomorphisms but we leave to the reader the verification that these really are maps of @xmath68-modules ( or @xmath68-submodules ) .
* @xmath252 and @xmath253 .
the isomorphisms are given by @xmath254 .
* @xmath255 and * @xmath256 for all @xmath195 .
the assignments @xmath257 give these isomorphisms .
in the sequel , by abuse of notation , @xmath258 and @xmath259 denote the multiplication and sum module @xmath190 .
* @xmath260 for all @xmath195 .
we obtain this isomorphism keeping in mind that @xmath261 * @xmath262 for all @xmath195 .
the isomorphism follows by considering the submodules @xmath263 * @xmath264 with @xmath265 .
here we use that @xmath266 * @xmath267 for all @xmath195 . in the first isomorphism
@xmath268 identifies with the element @xmath269 meanwhile in the second isomorphism , @xmath270 , for @xmath271 , identifies with @xmath272 * @xmath273 for all @xmath195 . here
we take the assignments @xmath274 for @xmath271 .
* @xmath275 .
for this isomorphism we use that @xmath276 is a one - dimensional submodule ; the maps @xmath277 with @xmath278 , define inclusions of @xmath68-modules and @xmath198 is included as @xmath68-module by identifying the element @xmath279 of @xmath198 with the element @xmath280 for @xmath281 and @xmath195 .
* @xmath282 . here
we have to identify @xmath283 , for @xmath284 , with the element @xmath285
through this section , we fix a finite non - abelian group @xmath2 and a yetter - drinfeld module @xmath286 such that its nichols algebra @xmath1 is finite - dimensional .
we denote by @xmath0 the drinfeld double of the bosonization @xmath10 . for shortness
we say that @xmath0 is a _ quantum group at a non - abelian group_. in the first part of the section we describe the algebra structure of @xmath0
. then we introduce and study the verma modules for @xmath0 .
[ def : ov ] we set @xmath38 to be @xmath98 endowed with the yetter - drinfeld module structure over @xmath152 defined by the following properties : @xmath287 for every @xmath288 , @xmath289 , @xmath153 and @xmath162 . it is a straightforward computation to check that these structures satisfy the compatibility for yetter - drinfeld modules .
also , this is a consequence of the next lemma .
recall the hopf algebra structure of a bosonization in @xmath132[subsub : boson ] .
[ le : iso dual op ] the algebra map @xmath290 defined by @xmath291 for all @xmath153 , @xmath162 , @xmath289 , @xmath288 , @xmath292 and @xmath293 , is an isomorphism of graded hopf algebras . in particular , the hilbert series of the nichols algebras @xmath1 and @xmath39 are equals
. we can deduce that @xmath294 where @xmath295 is the nichols algebra of its homogeneous space of degree @xmath11 following for instance ( * ? ? ?
* section 2 ) . also , we see from the definition in the statement that @xmath296 , with @xmath297 for all @xmath288 and @xmath162 , is a linear map which is bijective in degree @xmath298 and @xmath11
. therefore the lemma follows if we show that @xmath299 for all @xmath300 and @xmath288 .
in fact , ensures that @xmath301 is a hopf algebra map . by , we determine that @xmath38 is the space of coinvariants in degree @xmath11 of @xmath302 with respect to the projection over @xmath152 and the corresponding yetter - drinfeld structure is given by definition [ def : ov ] . hence @xmath295 is the nichols algebra of @xmath38 .
we check the first equality of , the remainder ones can be checked in a similar way . as @xmath10 is a graded hopf algebra
it is enough to see that @xmath303 for all @xmath304 and @xmath289 .
we identify the hopf subalgebra @xmath302 of @xmath0 with @xmath305 by invoking the above lemma .
the quantum group @xmath0 at a non - abelian group can be presented as an algebra generated by the elements belonging to @xmath7 , @xmath38 , @xmath8 and @xmath152 subject to their relations in @xmath1 , @xmath39 , @xmath8 and @xmath152 , plus the commutation rules @xmath306 @xmath307 for all @xmath300 , @xmath289 and @xmath288 .
the equations correspond to the bosonization , see .
meanwhile , and follow from .
the subalgebra of @xmath0 generated by @xmath8 and @xmath152 is a hopf subalgebra isomorphic to @xmath3 and it is the coradical of @xmath0 . in particular , @xmath0 is non - pointed .
it follows from lemma [ le : iso dual op ] and . due to the above lemmata
, a quantum group at a non - abelian group has a triangular decomposition , that is @xmath308 and it is a @xmath24-graded algebra by setting @xmath309 we consider @xmath7 and @xmath38 as yetter - drinfeld modules over @xmath3 with the adjoint action and the same coaction as @xmath8-comodule and @xmath152-comodule , respectively .
that is possible because the rules and guarantee that @xmath7 and @xmath38 are stable by the adjoint action of @xmath3 , _ i. e. _ @xmath310 and @xmath311 for all @xmath312 , @xmath289 and @xmath288 .
also , @xmath313 as coalgebra .
we extend these structures to @xmath1 and @xmath39 .
hence the bosonization @xmath314 identifies with the subalgebra of @xmath0 generated by @xmath3 and @xmath7 , respectively @xmath38 .
[ obs : v as dgmod ] the adjoint action of @xmath3 on @xmath7 coincides with the action defined by the equivalence of categories between @xmath164 and @xmath315 , see .
we would like to remark other facts about @xmath7 and @xmath38 .
we refer to @xcite for details about the item ( iv ) below .
[ le : v ov son duales sobre dg ] 1 .
@xmath38 is the dual object of @xmath7 in the tensor category @xmath315 .
2 . @xmath1 and @xmath39 are the nichols algebras of @xmath7 and @xmath38 in @xmath316 , respectively .
3 . @xmath1 and @xmath39 are the nichols algebras of @xmath7 and @xmath38 in @xmath315 , respectively .
@xmath39 is isomorphic to the opposite and copposite hopf algebra @xmath317 in @xmath315 .
\(i ) let @xmath288 , @xmath289 and @xmath153 . by , we have that @xmath318 . on the other hand , and imply that @xmath319 .
then @xmath320 . in a similar way
, we see that @xmath321 .
\(ii ) @xmath1 and @xmath39 are braided hopf algebras in @xmath316 , because @xmath322 and @xmath47 are hopf algebras , which satisfy the defining properties of a nichols algebra \(iii ) follows from ( ii ) because the braiding of @xmath316 on @xmath7 coincides with that of @xmath315 and the same holds for @xmath38 .
\(iv ) let @xmath323 be the dual object of @xmath7 in @xmath164 .
by ( * ? ? ?
* theorem 3.2.30 ) , @xmath324 in @xmath164 .
we said before that the adjoint action of @xmath3 on @xmath7 coincides with the action defined by the functor given the equivalence of categories between @xmath164 and @xmath315 .
then , by ( i ) , @xmath38 is the image of @xmath323 by this functor and ( iv ) follows because the braidings of these categories are equal via this functor .
a classical technique in representation theory is to study modules induced by simple modules of a subalgebra .
such is the case of the verma modules for quantum groups , see for instance ( * ? ? ?
* chapter 7 ) , ( * ? ? ? * chapter 5 ) and ( * ? ? ?
* chapter 3 ) .
following this idea , we shall induce from the subalgebra @xmath47 .
every simple @xmath47-module is isomorphic to a simple @xmath3-module where @xmath325 acts via the counit .
this holds because @xmath39 is local and hence @xmath326 is the jacobson radical of @xmath47 .
[ def : verma ] let @xmath46 be a simple @xmath3-module . the _ verma module _
@xmath49 is the @xmath0-module induced by @xmath46 seen as a module over @xmath47 .
explicitly , @xmath327 we fix a simple @xmath3-module @xmath46 and set @xmath328 . immediately from the definition , we get that @xmath329 is free as @xmath1-module of rank @xmath330 .
moreover , @xmath331 since @xmath332 , for @xmath312 , @xmath333 and @xmath334 , and the last term is the definition of the action in the tensor product of two @xmath3-modules .
also @xmath329 inherits the @xmath24-grading of @xmath0 and its homogeneous spaces are @xmath3-submodules .
namely , its homogeneous space of degree @xmath335 is @xmath336 thus @xmath329 turns out to be an @xmath24-graded @xmath0-module since @xmath337 in fact , the first equality holds as the action by an element of @xmath7 is just the multiplication in the nichols algebra and this is generated by @xmath7 .
we proceed by induction to prove the second inclusion . if @xmath338 , then @xmath339 on which @xmath38 acts by zero . if @xmath340 , then @xmath341 where the first inclusion follows from and the second one by inductive hypothesis . as @xmath322-module ,
@xmath329 is generated by any element of degree 0 , that is @xmath342 in fact , @xmath343 since @xmath46 is @xmath3-simple .
however we have more generators for a verma module as @xmath322-module .
[ le : si esta 1 mas algo entonces es todo el verma ] if @xmath344 is non - zero , then @xmath345 for any @xmath346 . by the lemma follows if @xmath347 .
otherwise , we write @xmath348 with @xmath349 and @xmath350 . by there is @xmath351 such that @xmath352 , moreover @xmath353 .
then @xmath354 and the maximum degree of @xmath355 is smaller than @xmath356 .
hence the lemma follows by induction in the maximum degree of @xmath357 since @xmath1 is finite - dimensional and @xmath358 if @xmath359 is the maximum degree of @xmath1 . using the above lemma
we prove one of the main properties of a verma module .
[ teo : unico submod max ] a verma module has a unique maximal @xmath0-submodule and it is homogeneous .
if @xmath360 is a strict @xmath0-submodule of @xmath329 , then there exists a non - zero negative integer @xmath361 such that @xmath362 by lemma [ le : si esta 1 mas algo entonces es todo el verma ] .
hence the sum @xmath363 of all strict @xmath0-submodules is the unique maximal @xmath0-submodule of @xmath329 .
let @xmath364 with @xmath365 . if we see that @xmath366 for all @xmath367 , then @xmath363 is homogeneous .
otherwise , without loss of generality , we can assume @xmath368 with @xmath367 maximal , then @xmath369 .
thus there is @xmath370 such that @xmath371 and hence @xmath372 with @xmath373
. then @xmath374 by lemma [ le : si esta 1 mas algo entonces es todo el verma ] but this is not possible because @xmath375 . as it is common ,
we introduce the highest - weight modules in such a way that a verma module is a highest - weight module .
the weights in our case are the simple @xmath3-modules which can have dimension greater than one .
[ def : highest modules ]
let @xmath225 be a @xmath0-module and @xmath376 a simple @xmath3-submodule of weight @xmath174 .
assume that @xmath225 is generated as @xmath0-module by @xmath29 .
we say that @xmath225 is a _
highest - weight module of weight @xmath174 _ if @xmath377 .
we say that @xmath225 is a _
lowest - weight module of weight @xmath174 _ if @xmath378 .
hence we have that @xmath379 if @xmath225 is a highest - weight module , and @xmath380 in case that @xmath225 is a lowest - weight module .
these follow from the decomposition of @xmath0 and since @xmath29 is @xmath3-simple .
we set @xmath381 where @xmath359 is the maximum degree of the nichols algebra .
note that @xmath382 is simple as a @xmath3-module since @xmath383 is one - dimensional .
[ teo : socle simple ] the socle of the verma module @xmath329 is simple as a @xmath0-module and equals @xmath384 .
the socle is simple if we show that @xmath385 for any homogeneous element @xmath386 of degree @xmath387 in @xmath329 with @xmath388 . to show that , we write @xmath389 with @xmath390 and @xmath391 linearly independent .
we pick @xmath392 . since @xmath383 is the space of integrals of the nichols algebra
, there is @xmath393 such that @xmath394 .
then @xmath395 is an homogeneous element of degree @xmath396 .
hence @xmath397 for appropriated @xmath398 and therefore @xmath385 because @xmath382 is @xmath3-simple . finally , the socle is a lowest - weight module because it is generated by @xmath382 .
therefore the socle is equal to @xmath384 by . as a direct consequence
we obtain a criterion for the simplicity of a verma module . recall that the hilbert series of @xmath1 and @xmath39 are equal by lemma [ le : iso dual op ] .
[ cor : verma simple xtop ] let @xmath399 , @xmath400 and @xmath121 .
the verma module @xmath329 is simple as @xmath0-module if and only if @xmath401 . by , @xmath402 and hence it generates @xmath329 by .
then the socle , which is simple by the above theorem , is exactly @xmath329 .
assume now that @xmath329 is a simple @xmath0-module .
in particular , @xmath329 is generated by @xmath403 and thus there is an element @xmath404 such that @xmath405 . by , @xmath406 .
since @xmath359 is the maximum degree of @xmath1 and @xmath39 , we have that @xmath407 .
finally , we can take @xmath408 because @xmath409 is one - dimensional . due to the above theorems we can introduce the following @xmath0-modules . [ def : max simple socle ] let @xmath46 be a simple @xmath3-module
. then * @xmath410 denotes the maximal @xmath0-submodule of @xmath49 .
* @xmath52 denotes the head of @xmath49 .
* @xmath53 denotes the socle of @xmath49 as @xmath0-module .
the following theorem states that the correspondence @xmath411 , between the sets of simple @xmath3-modules and simple @xmath0-modules , is bijective .
[ teo : bi con l ] 1 .
let @xmath46 be a simple @xmath3-module .
then @xmath52 is the unique simple highest - weight module of weight @xmath174 .
every simple @xmath0-module is isomorphic to @xmath52 for a unique simple @xmath3-module @xmath46 .
\(i ) @xmath52 is a simple @xmath0-module by theorem [ teo : unico submod max ] .
as @xmath49 is generated by @xmath46 , @xmath52 is so .
moreover , @xmath412 then @xmath52 is a highest - weight module .
the uniqueness follows from the fact that a highest - module @xmath413 of weight @xmath174 is a quotient of @xmath49 since @xmath46 turns out to be a @xmath47-submodule of @xmath413 . if also @xmath413 is @xmath0-simple , then @xmath414 since @xmath410 is the unique maximal submodule of @xmath49 .
\(ii ) every @xmath0-module @xmath413 has a simple @xmath47-module , say @xmath46 .
then we have a morphism @xmath415 of @xmath0-modules and this map is surjective if @xmath413 is @xmath0-simple . hence @xmath416 . on the other hand ,
if there exists another @xmath417 isomorphic to @xmath413 , then @xmath418 because they are highest weights of @xmath413 .
the correspondence @xmath419 also is bijective by the next theorem .
we set @xmath420 to be the simple @xmath3-module isomorphic to @xmath421 .
[ teo : bi con socle ] 1 .
let @xmath46 be a simple @xmath3-module .
then @xmath53 is the unique simple lowest - weight module of weight @xmath420 .
every simple @xmath0-module is isomorphic to @xmath53 for a unique simple @xmath3-module @xmath46 .
the socles of the verma modules are lowest - weight modules by theorem [ teo : socle simple ] . also , the socles of non - isomorphic verma modules are non - isomorphic because their lowest - weight components are not .
hence the uniqueness in ( i ) and ( ii ) follow from the fact that the sets of simple @xmath0-modules and simple @xmath3-modules are in bijective correspondence by theorem [ teo : bi con l ] .
we denote by @xmath422 the dual @xmath3-module of @xmath46 .
[ teo : l dual s ] the dual @xmath0-module @xmath423 is a highest - weight module of weight @xmath424 . therefore @xmath425 as @xmath0-modules .
let @xmath426 , @xmath427 , be the homogeneous component of degree @xmath428 of @xmath53 . then @xmath429 and @xmath430 as @xmath3-modules . since @xmath423 is simple , it is generated by @xmath431 .
moreover , @xmath432 .
in fact , @xmath433 because @xmath434 and @xmath435 for all @xmath427 .
hence the theorem follow from theorem [ teo : bi con l ] . 1 . even though the maximal submodule of @xmath329 is homogeneous , a submodule is not necessarily homogeneous , cf .
lemma [ le : t ] .
the maximal submodule is not necessarily generated by highest - weight submodules , cf .
theorem [ teo : el latice de sigma menos ] .
the head and the socle of a verma module are not necessarily isomorphic , cf .
corollary [ cor : l tau cero s e rho ] .
4 . there are examples with @xmath436 .
for instance , let @xmath295 be the nichols algebra considered in @xcite . then @xmath437 is not necessarily trivial as yetter - drinfeld module .
we fix a @xmath0-module @xmath225 and a @xmath3-submodule @xmath376 .
we will explain how we can compute the @xmath0-submodule generated by @xmath29 under the hypothesis that it is either a lowest - weight or highest - weight module .
we denote by @xmath243 , and call it _ action map _ , the restriction to @xmath438 of the action of @xmath0 over @xmath225 . by abuse of notation , we also denote by @xmath243 the restriction to @xmath439 .
the key of our idea is the simple observation that @xmath440 indeed , we want to see that @xmath441 for any @xmath442 , @xmath312 and @xmath121 .
the action on the tensor product is @xmath443 .
then we apply the action map and obtain @xmath444 . by and
, the @xmath0-submodule generated by @xmath29 is either @xmath445 or @xmath446 .
hence we can compute @xmath447 following the algorithm described in the next remark .
[ obs : description submodule gen by a submodule 2 ] keep the above hypothesis and notation . 1 .
decompose the tensor product @xmath448 , or @xmath438 depending on the case , into the direct sum @xmath449 of simple @xmath3-modules .
this is possible because @xmath3 is semisimple .
moreover , its simple modules are well - know , recall @xmath132[subsec : dg ] . 1 .
apply the action map to each simple @xmath3-module @xmath450 .
the restriction of the action map to @xmath450 is either zero or an isomorphism by schur s lemma .
therefore the image of the action map is isomorphic as @xmath3-module to the direct sum of the simple modules @xmath450 that are not annulled .
note that @xmath451 if and only if @xmath452 for some @xmath453 . 1 .
repeat the process with the @xmath3-submodule @xmath454 , or @xmath455 depending on the case , instead of @xmath29 .
we have to repeat the process as many times as the maximum degree of @xmath39 or @xmath1 . 1 .
the @xmath0-submodule generated by @xmath29 is the sum of all @xmath3-submodules obtained in the step ( ii ) .
throughout this section @xmath59 is the yetter - drinfeld module over @xmath60 given by @xmath61 for any transposition @xmath62 and @xmath63 .
let @xmath456 be the yetter - drinfeld module attached to @xmath7 from definition [ def : ov ] .
from @xcite we know that @xmath1 is isomorphic to the @xmath92-dimensional fomin - kirillov algebra introduced in @xcite .
we denote by @xmath0 the drinfeld double of the bosonization @xmath65 .
the aim of this section is to apply the results of the previous section in the specific example of this quantum group .
we have to consider @xmath7 and @xmath38 as @xmath68-modules with the adjoint action in @xmath0 . using remark [ obs : v as dgmod ] , we see that @xmath457 via the assignment @xmath458 for every transposition @xmath62 . by lemma [ le :
v ov son duales sobre dg ] , @xmath38 is the dual object of @xmath7 in the category @xmath459 .
we denote by @xmath460 the basis of @xmath38 dual to @xmath461 , that is @xmath462 .
then it is not difficult to check that @xmath463 via the assignment @xmath464 for every transposition @xmath62 .
the defining relations of the nichols algebra @xmath1 are @xmath465 we denote by @xmath466 the basis of @xmath1 which is obtained by choosing one element per row of the next list and multiply them from top to bottom , see _
e. g. _ @xcite : @xmath467 we set @xmath468 , @xmath469 .
the element of maximum degree in @xmath466 is @xmath470 [ descomposicion de bv as ds3mod ] we have the following isomorphisms of @xmath68-modules . 1 .
[ item : 0 4 ] @xmath471 , 2 .
[ item : 1 3 ] @xmath472 ; the last isomorphism is given by the assignment @xmath473 3 .
[ item : descomposion de bv tau]@xmath474 , the isomorphism is given by @xmath475 \(i ) for @xmath476 the isomorphism is clear .
for @xmath477 it is enough to see that @xmath478 to prove ( ii ) we note that @xmath479 and @xmath480 \(iii ) follows from and using .
we fix a simple @xmath0-module @xmath29 and take the basis @xmath481 of @xmath29 which consists of elements of the form @xmath238 , @xmath63 , recall @xmath132[subsub : s3 sigma simples ] , @xmath132[subsub : s3 tau simples ] and @xmath132[subsub : s3 e simples ] .
let @xmath329 be the verma module of @xmath29 .
since @xmath482 , a basis of @xmath329 is the set of elements @xmath483 then the action of @xmath1 on @xmath329 is given just by the multiplication : @xmath484 the action of @xmath68 is the diagonal action by : @xmath485 computing the action of @xmath39 is more laborious .
we have to use the commutation rules between the generators of @xmath39 and @xmath1 given by . in our case
is rewritten as follow @xmath486 for all distinct transpositions @xmath62 and @xmath72 . however , if we know the action of @xmath487 on @xmath329 , then we can deduce the action of the remainder generators of @xmath39 .
in fact , let @xmath488 and @xmath489 such that @xmath490 .
hence @xmath491 in the appendix we give explicitly the action of @xmath487 on each element @xmath492 in the basis of @xmath329 .
we leave this for the appendix because it is a very long list and we do nt want to bore the reader now .
we shall use these computations in the next subsections without previous mention .
[ teo : los simples ] the verma modules @xmath493 , @xmath494 , @xmath89 and @xmath90 are @xmath0-simple . therefore @xmath495 holds for these weights . by corollary [ cor :
verma simple xtop ] , it is enough to check that @xmath496 for some @xmath497 . in the case
@xmath493 , using the calculations done in the appendix we have that @xmath498 . for @xmath494 , in the same way we obtain @xmath499 . finally , in both cases @xmath500 and @xmath501 we have that @xmath502 .
therefore the isomorphism @xmath495 holds because @xmath503 .
the aim of this subsection is to prove the next theorem which describes the submodules of @xmath504 . as a byproduct , we find @xmath505 and the remaining simple @xmath0-modules as subquotients of @xmath504 . ]
[ teo : el latice de sigma menos ] every @xmath0-submodule of @xmath504 is obtained by adding the @xmath0-submodules @xmath506 in particular , @xmath507 and @xmath508 as @xmath68-modules .
moreover , @xmath509 are the composition factors of @xmath97 . the submodules mentioned in the statement will be given in the successive lemmas but the figure [ fig : sigma menos ] helps to visualise them .
namely , each dot represents a weight of @xmath97 .
this follows from the fusion rules given in [ subsub : fusion rules ] since @xmath510 : @xmath511 is of weight @xmath512 , @xmath13 and @xmath7 are of weight @xmath513 , @xmath514 and @xmath515 are of weight @xmath516 , @xmath295 and @xmath382 are of weight @xmath517 , @xmath518 is of weight @xmath519 for any @xmath520 , @xmath521 is of weight @xmath522 for @xmath195 . of course , the weights in the same row are in the same homogeneous component and the degree decreases from the top to the bottom . then , the dots connected by a thick line represent the socle @xmath523 , lemma [ le : socle sigma menos ] .
the dots over the northwest lines form @xmath524 , lemma [ le : vqn ] , and those over the northeast lines form @xmath525 , lemma [ le : upo ] .
the dots enclosed by circles represent a @xmath0-submodule @xmath526 , lemma [ le : t ] .
the isolated dots on the right hand side represent the weights of @xmath527 .
we start by calculating the socle of the verma module and then we describe the highest - weight submodules . for that , we shall use the algorithm described in remark [ obs : description submodule gen by a submodule 2 ] . in order to avoid extra computations , we approximate the kernel of the action map before applying this algorithm using remark [ obs : description submodule gen by a submodule 1 ] .
let @xmath528 and @xmath529 be the @xmath68-submodules of @xmath504 generated by * @xmath530 , for @xmath195 , and * @xmath531 , respectively . by , @xmath532 identifies with the element @xmath533 and belongs to the submodule of weight @xmath522 .
hence @xmath534 for @xmath195 .
[ le : socle sigma menos ] the socle of @xmath97 is @xmath535 where @xmath536 and @xmath534 as @xmath68-modules , @xmath233 .
moreover , @xmath523 is a highest - weight module of weight @xmath517 and therefore @xmath537 we use the algorithm proposed in remark [ obs : description submodule gen by a submodule 2 ] to compute the socle . recall that @xmath538 by theorem [ teo : socle simple ] . as @xmath68-module ,
we have that @xmath539 by . since @xmath540 , the action map is zero in the first submodule by remark [ obs : description submodule gen by a submodule 1 ] .
the second submodule decomposes into the direct sum @xmath541 where @xmath542 belongs to the submodule of weight @xmath522 by .
the action map applied to these elements gives @xmath543 which is zero iff @xmath544 .
otherwise , we obtain @xmath545 and hence @xmath546 recall remark [ obs : description submodule gen by a submodule 2 ] . now , we calculate @xmath547 . by ,
@xmath548 as @xmath68-modules and the element @xmath549 belongs to the submodule of weight @xmath550 .
we apply the action map to these elements and obtain @xmath551 hence @xmath552 and @xmath553 $ ] .
therefore @xmath554 finally , we consider @xmath555{{\otimes}}{\mathsf{r}}[(12)]$ ] and @xmath556{{\otimes}}{\mathsf{r}}[(23)]$ ] . since @xmath557
, we see that @xmath558 by remark [ obs : description submodule gen by a submodule 1 ] .
therefore @xmath559 moreover , @xmath523 is simple and generated by @xmath560 with @xmath558 .
therefore @xmath561 .
let @xmath562 and @xmath563 be the @xmath68-submodules of @xmath97 generated by * @xmath564 and * @xmath565 , respectively .
note that @xmath562 is of weight @xmath513 by .
[ le : vqn ] let @xmath566
. then a. [ le : vqn i ] @xmath567 is a highest - weight submodule of weight @xmath513 .
b. [ le : vqn ii ] @xmath568 . c. [ le :
vqn iii ] @xmath563 is of weight @xmath516 .
since @xmath569 , @xmath570 by remark [ obs : description submodule gen by a submodule 1 ] and thus @xmath571 is a highest - weight module . hence we will use remark [ obs : description submodule gen by a submodule 2 ] to compute @xmath572 . by , @xmath573 where @xmath574 belongs to the submodule of weight @xmath550 .
the action map applied to these elements gives @xmath575 hence @xmath576 .
in particular , @xmath577 which proves . now , we just have to compute @xmath578 since @xmath579 . as @xmath580-modules , @xmath581 , cf .
[ subsub : fusion rules ] .
the action map on the components of weight @xmath582 and @xmath513 is zero since @xmath583 , recall remark [ obs : description submodule gen by a submodule 1 ] .
meanwhile , the action map on the components of weight @xmath522 is not zero .
in fact , @xmath584 belongs to the submodule of weight @xmath522 by and @xmath585 hence @xmath586 . since @xmath587
, we conclude that @xmath588 and follows . finally
, we proof by noting that @xmath589 and @xmath590 since @xmath591 using the characterization of the simple modules given in theorem [ teo : bi con l ] , the next result follows directly from the above lemma .
[ cor : l e rho cociente de vqn ] the quotient @xmath592 is a simple highest - weight module of weight @xmath513 . therefore @xmath593 as @xmath68-modules .
let @xmath594 , @xmath595 and @xmath596 be the @xmath68-submodules generated by * @xmath597 , * @xmath598 and * @xmath599 , for @xmath195 , respectively .
we see that @xmath594 is of weight @xmath513 by and lemma [ descomposicion de bv as ds3mod ] , and @xmath596 is of weight @xmath522 by . [ le : upo ] let @xmath600 .
then a. [ le : upo i ] @xmath601 is a highest - weight submodule of weight @xmath512 .
b. [ le : upo ii ] @xmath602 . c. [ le : upo iii ]
@xmath595 is of weight @xmath516 .
we have that @xmath603 is a highest - weight module , since @xmath604 and remark [ obs : description submodule gen by a submodule 1 ] provides @xmath605 .
then we compute @xmath606 . by ,
@xmath607 where @xmath608 belongs to the submodule of weight @xmath550 .
the action map gives @xmath609 and we obtain @xmath610 .
in particular , @xmath611 which proves .
now , we calculate @xmath612 . as @xmath580-modules , @xmath613 , cf .
[ subsub : fusion rules ] . using we see that the action map on the components of weight @xmath582 and @xmath512 is zero since @xmath614 and @xmath615 . meanwhile , from and , the action map on the components of weight @xmath616 , @xmath617 and @xmath513 is not zero .
in fact , for @xmath233 : @xmath618 for the component @xmath513 we have that @xmath619 .
hence @xmath620 .
since @xmath587 , @xmath621 and @xmath622 , we conclude that @xmath623 and follows . for the proof of we
note that @xmath624 and @xmath625 since @xmath626 the next result is a direct consequence of the above lemma .
[ cor : l tau cero cociente de upo ] the quotient @xmath627 is a simple highest - weight module of weight @xmath512 . therefore @xmath628 as @xmath68-modules .
[ cor : l tau cero s e rho ] as @xmath0-modules , @xmath629 and @xmath630 . since @xmath631 , @xmath632 .
hence @xmath594 is a lowest - weight in the quotient @xmath627 .
therefore this quotient is isomorphic to @xmath633 by theorem [ teo : bi con socle ] .
the proof of the second isomorphism is similar .
the @xmath580-submodules of @xmath97 of weight @xmath519 are * @xmath634 where @xmath520 , @xmath635 [ le : t ] let @xmath636 for any @xmath520 .
then @xmath637 .
in particular , @xmath638 is a highest - weight submodule of weight @xmath519 .
we see that @xmath639 and @xmath640 .
meanwhile , @xmath641 and clearly , @xmath642 .
then @xmath643 and the lemma follows .
[ cor : l e mas cociente de t ] the quotient @xmath644 is a simple highest - weight module of weight @xmath519 .
therefore @xmath645 as @xmath68-modules .
moreover , @xmath646 as @xmath0-modules .
the first part follows as the above corollaries .
in particular , @xmath647 is one - dimensional and then it is also a lowest - weight module . hence @xmath646 holds .
let @xmath225 be a @xmath0-submodule of @xmath97 .
then @xmath648 where @xmath649 is @xmath68-simple .
since @xmath650 , it is enough to compute the @xmath0-submodule generated by @xmath649 case - by - case according to the weight of @xmath649 . recall the weights of @xmath97 from figure [ fig : sigma menos ] . 1 . if @xmath649 is of weight @xmath519 , then @xmath651 for some @xmath520 by lemma [ le : t ] .
if @xmath649 is of weight @xmath513 , then there is an element @xmath652 with @xmath653 and @xmath654 .
assume @xmath655 , otherwise @xmath656 is either @xmath657 or @xmath524 by lemmas [ le : vqn ] and [ le : upo ] .
then @xmath658 because @xmath570 , and hence @xmath659 . thus @xmath660 and therefore @xmath661 .
3 . if @xmath649 is of weight @xmath512 or @xmath516 . proceeding as above , we can see that @xmath662 .
if @xmath649 is of weight @xmath522 with @xmath663 , then there is an element @xmath664 with @xmath665 and @xmath666 .
moreover , @xmath667 as @xmath668 .
then either @xmath669 , if @xmath670 , or @xmath671 because @xmath672 . 5 .
if @xmath649 is of weight @xmath517 , then either @xmath669 or there is @xmath673 with @xmath674 and @xmath675 .
if @xmath676 , then @xmath671 by lemma [ le : si esta 1 mas algo entonces es todo el verma ] when @xmath677 or by noting that @xmath678 when @xmath679 .
[ teo : el latice de e mas ] the proper @xmath0-submodules of @xmath680 are @xmath681 where a. [ teo : e mas max ] @xmath682 is a highest - weight submodule of weight @xmath517 . b. [ teo : e mas soc ] @xmath683 is a highest - weight submodule of weight @xmath519 .
therefore @xmath684 are the composition factors of @xmath685 . ] in the figure [ fig : e mas ] , we have schemed the submodules of @xmath680 .
the weights are represented by dots , notice that @xmath686 as @xmath580-module .
the big dot in the bottom represents the socle @xmath687 .
the dotted area represents the maximal @xmath0-submodule @xmath688 and hence the dot in the top corresponds to @xmath647 . by corollary [ cor : l e mas cociente de t ] , @xmath80 as @xmath580-module and hence @xmath682 .
since @xmath688 is homogeneous by theorem [ teo : unico submod max ] , @xmath689 .
moreover , @xmath690 generates @xmath688 because @xmath686 as @xmath1-modules .
therefore @xmath688 is a highest - weight submodule of weight @xmath517 and follows . by and theorem [ teo : bi con l ]
, @xmath688 has a quotient isomorphic to @xmath527 . by theorem [ teo : el latice de sigma menos ] , @xmath86 as @xmath580-modules .
then , we deduce that the unique @xmath0-submodule of @xmath688 is @xmath382 by inspection in the weights of the verma module , see figure [ fig : e mas ] . therefore @xmath683 and follows .
the last sentence of the statement is immediate .
let @xmath691 and @xmath692 be the @xmath68-submodules of weight @xmath517 and @xmath513 with basis * @xmath693 , @xmath271 , see ; * @xmath694 and @xmath695 , recall and lemma [ descomposicion de bv as ds3mod ] .
[ teo : el latice de tau cero ] the proper @xmath0-submodules of @xmath696 are @xmath697 where a. [ teo : tau cero mas ] @xmath698 is a highest - weight submodule of weight @xmath517 . b. [ teo : tau cero soc ] @xmath699 is a highest - weight submodule of weight @xmath513 .
therefore @xmath700 are the composition factors of @xmath701 .
the weights of @xmath696 are represented by dots in the figure [ fig : tau cero ] which can be computed using the fusion rules in [ subsub : fusion rules ] .
the weights conected by a line form the socle @xmath702 and those over the dotted area form the maximal @xmath0-submodule @xmath703 . the weights on the left hand side correspond to @xmath704 .
] by corollary [ cor : l tau cero s e rho ] , @xmath705 is a highest - weight module of weight @xmath513 .
we see that @xmath706 using remark [ obs : description submodule gen by a submodule 1 ] .
therefore @xmath699 .
by corollary [ cor : l tau cero cociente de upo ] , @xmath84 as @xmath580-module .
hence the sum of the simple @xmath580-submodules over the dotted area in figure [ fig : tau cero ] have to form the maximal @xmath0-submodule @xmath703 .
since @xmath703 is homogeneous , @xmath707 and then @xmath708 is a highest - weight submodule of weight @xmath517 . on the other hand
, @xmath708 has a quotient isomorphic to @xmath527 and contains the socle @xmath705 .
therefore @xmath698 and @xmath705 is the unique @xmath0-submodule of @xmath703 .
the proof of the next theorem and the description of figure [ fig : e rho ] are similar to the above subsection .
let @xmath709 and @xmath710 be the @xmath68-submodules of weight @xmath517 and @xmath512 with basis * @xmath711 , @xmath271 , see ; * @xmath712 and @xmath713 , recall and lemma [ descomposicion de bv as ds3mod ] .
[ teo : el latice de erho ] the proper @xmath0-submodules of @xmath714 are @xmath715 where a. [ teo : e rho max ] @xmath716 is a highest - weight submodule of weight @xmath517 . b. [ teo : e rho soc ] @xmath717 is a highest - weight submodule of weight @xmath512 .
therefore @xmath718 are the composition factors of @xmath714 . ]
here we compute the action of @xmath719 on the verma modules .
we noticed in that it suffices to calculate the action of @xmath487 to know the action of the generators @xmath720 and @xmath721 .
for the modules @xmath722 and @xmath714 we have only one list since all elements have weight @xmath146 . for the module
@xmath723 we have three lists ( as the elements may have weight ( 12 ) , ( 13 ) or ( 23 ) ) and for the module @xmath724 we have two lists ( for the possible weights ( 123 ) and ( 132 ) ) .
list 1 : action on @xmath722 @xmath725 @xmath726 @xmath727 @xmath728 @xmath729 @xmath730 @xmath731 @xmath732 @xmath733 @xmath734 @xmath735 list 2 : action on @xmath714 @xmath736 @xmath737 @xmath738 @xmath739 @xmath740 @xmath741 @xmath742 @xmath743 @xmath744 @xmath745 @xmath746 list 3 : action on @xmath723 @xmath747 @xmath748 @xmath749 @xmath750 @xmath751 @xmath752 @xmath753 @xmath754 @xmath755 @xmath756 @xmath757 list 4 : action on @xmath723 @xmath758 @xmath759 @xmath760 @xmath761 @xmath762 @xmath763 @xmath764 @xmath765 @xmath766 @xmath767 @xmath768 list 5 : action on @xmath723 @xmath769 @xmath770 @xmath771 @xmath772 @xmath773 @xmath774 @xmath775 @xmath776 @xmath777 @xmath778 @xmath779 list 6 : action on @xmath724 @xmath780 @xmath781 @xmath782 @xmath783 @xmath784 @xmath785 @xmath786 @xmath787 @xmath788 @xmath789 @xmath790 list 7 : action on @xmath724 @xmath791 @xmath792 @xmath793 @xmath794 @xmath795 @xmath796 @xmath797 @xmath798 @xmath799 @xmath800 @xmath801
00 n. andruskiewitsch , m. graa , braided hopf algebras over non abelian finite groups , bol .
acad . ciencias ( crdoba ) 63 ( 1999 ) 4578 . | the drinfeld double @xmath0 of the bosonization of a finite - dimensional nichols algebra @xmath1 over a finite non - abelian group @xmath2 is called a _ quantum group at a non - abelian group_. we introduce verma modules over such a quantum group @xmath0 and prove that a verma module has simple head and simple socle .
this provides two bijective correspondences between the set of simple modules over @xmath0 and the set of simple modules over the drinfeld double @xmath3 . as an example
, we describe the lattice of submodules of the verma modules over the quantum group at the symmetric group @xmath4 attached to the 12-dimensional fomin - kirillov algebra , computing all the simple modules and calculating their dimensions .
hopf algebras , nichols algebras , fomin - kirillov algebras , quantum groups , verma modules , representation theory . |
modern big - bang cosmology firmly predicts the existence of a relic neutrino background , and relates its temperature to the temperature of the background microwave radiation @xmath2 see , e.g. the basic texts @xcite .
verifying the existence of the relic neutrino sea represents one of the main challenges of modern cosmology .
clearly , in contrast to the study of the background microwave radiation that has a long history and has reached an unprecendented accuracy ( see , e.g. the latest results in @xcite ) , detection of relic neutrinos remains an unfulfilled dream .
various strategies have been proposed so far , based on laboratories searches @xcite and astrophysical observations @xcite , such as absorption dips in the flux of ultra high energy neutrinos ( for a review see e.g. @xcite ) . as far as their detection in laboratory experiments
is concerned , one needs to overcome two main obstacles : the low cross section characteristic of weak interactions and the low energy of relic neutrinos .
the second obstacle can be overcome if the corresponding detection reactions have vanishing thresholds .
therefore we discuss here the possibility of detecting the relic neutrinos by the charged current reactions using radioactive unstable nuclei as targets .
the paper is organized as follows : in the next section we derive expressions for the charged current neutrino induced reaction cross sections involving nonrelativistic neutrinos . we show that such cross sections , when the corresponding reaction has a vanishing threshold , scale with @xmath3 so that the number of events converges to a constant for @xmath4 . in the following section we discuss the possibility to use a tritium target to detect the cosmological @xmath5 .
we show that the main challenge is the separation of the produced electrons with energies just above the endpoint of the @xmath6 spectrum from the overwhelming flux of the electrons from the tritium @xmath6 decay that extends just below the 18.6 kev endpoint .
we also discuss the possibility of a resonance enhancement of reactions involving cosmological neutrinos .
we show that the charged current reactions , included the radiative ones , do not have a resonance character . in the conclusion
we summarize our findings and stress the need for an extreme energy resolution if sensitivity to detect sub - ev mass relic neutrinos should be reached .
let us first recapitulate briefly the cross section for massless neutrinos .
we use the reaction @xmath7 as an example .
this can be easily modified for reactions without threshold such as @xmath8 since we are interested in very low energy neutrinos , we can treat nucleons nonrelativistically , and keep only the lowest order terms in @xmath9 and @xmath10 .
the standard expression is then ( @xmath11 , see e.g. @xcite ) @xmath12 where @xmath13 is the momentum transferred squared and @xmath14 is the square of the center - of - mass ( cm ) energy .
starting with the usual current @xmath15 current weak interaction @xmath16 [ \bar{v}_{\nu } \gamma^{\mu } ( 1 - \gamma_5 ) v_e]\ ] ] where @xmath17 are the vector and axial - vector form factors respectively , we arrive at the squared matrix element ( see , e.g. @xcite ) @xmath18 evaluating it in the laboratory frame where the proton is at rest , and keeping only the leading terms one gets @xmath19 ~.\ ] ] with @xmath20 the neutron and proton masses .
furthermore , using @xmath21 in the laboratory frame , and using the jacobian @xmath22 we obtain the usual lowest order expression @xcite @xmath23 ~.\ ] ] with @xmath24 .
let us now consider the case of massive neutrinos . in eq.([e:4 ] ) one should then substitute @xcite @xmath25[s - ( m_p - m_{\nu})^2 ] = 4 m_p^2 p_{\nu}^2 ~,\ ] ] where the last expression is again in the laboratory frame .
the jacobian in eq.([e:8 ] ) becomes instead @xmath26 the cross section is then given by ( using @xmath27 as in the eq .
( [ e:8 ] ) ) @xmath28 ~.\ ] ] presenting a @xmath29 dependence ) , was evaluated , these modifications were not made and are not reflected in fig . 1 of that paper . ] .
such a behaviour is in agreement with the general form for the cross section associated to exothermic reactions of nonrelativistic particles @xcite .
we see that this form is a natural one " that in most cases of practical importance , when @xmath30 , acquires the standard form eq.([e:9 ] ) . the interaction cross section of very low energy neutrinos , eq.([e:12 ] )
was implicitly used long time ago by weinberg @xcite . very recently
this process has attracted particular interest thanks to the work by cocco _
@xcite where the authors have considered the possibility of detecting cosmological neutrinos through their capture on radioactive beta decaying ( and hence with no threshold ) nuclei . in the case of nuclear ( stable or unstable ) targets , i.e. reactions
@xmath31 when possible , one would use the usually known @xmath32 value of the inverse radioactive decay to eliminate the fundamental constants and the nuclear matrix element ( see e.g. @xcite ) @xmath33 with @xmath34 in units of @xmath35 . here
the averaging is done over the incoming flux , @xmath36 is in seconds , the statistical function @xmath37 and the electron energy @xmath38 and momentum @xmath39 are evaluated with @xmath40 as a unit of energy , and the nuclear matrix element is excluded using the relation @xmath41 . for the neutrino capture on stable targets ( i.e. with a threshold )
the electron energy is simply ( neglecting recoil ) @xmath42 .
so , for the @xmath43 induced reactions @xmath44 and for the @xmath5 induced reactions @xmath45 .
for the capture on a radioactive target @xmath46 and @xmath47 .
let us now consider the possibility to use the neutrino capture by radioactive beta - decaying nuclei to detect cosmological neutrinos .
for this aim the relevant quantity is the number of events , i.e. the cross section times the flux . for the latter , the dependence on @xmath48 in eqs.([e:12 ] , [ e:14 ] ) is canceled out and one obtains the number of events that converges to a constant when @xmath4 , in agreement with @xcite . as an example of the possible application of the above
finding let us consider the @xmath5 capture on tritium , as done in @xcite . while our conclusions are qualitatively similar to the conclusions reached by cocco _
@xcite , they differ in several significant details .
tritium decays into @xmath49he with the half - life of 12.3 years .
the decay @xmath50 value is 18.6 kev , and @xmath51 = 1143 . from eq.([e:14]-[e:15 ] ) we deduce the cross section for @xmath52 k nonrelativistic neutrinos @xmath53 here , in the first equation we used that @xmath54 . in making that estimate
we neglected the @xmath55 virial motion of massive neutrinos in the galactic halo , and the motion of earth and solar system with respect to the random motion of the neutrinos . in order to evaluate possible count rate
, we have to know the number density of the background @xmath5 sea .
its average value , for neutrinos evenly distributed throughout the whole universe , corresponds to @xmath56 k. for neutrinos ( or antineutrinos ) of one flavor only @xmath57 is @xmath58 55 @xmath59 .
massive neutrinos will be gravitationally clustered on the scale of @xmath58 mpc for neutrinos with @xmath60 1 ev , that is on the scale of galaxy clusters ( probably the clustering scale is even larger ) . assuming that in that case the ratio of the dimensionless neutrino and baryon densities @xmath61 remains the same as in the universe as a whole , we obtain @xmath62 for @xmath63 ev and @xmath64 @xmath59 for a cluster of galaxies . in the following we do not use the last estimate and treat this ratio as an unknown @xmath65 dependent parameter .
a more elaborated study of neutrino clustering is made for example in @xcite , giving smaller but nonnegligible clustering for @xmath661 ev .
note , however , that much larger neutrino clustering was considered in refs .
the authors of these papers speculate that features in the cosmic - ray spectra , in particular the ` knee ' at @xmath583 pev , are associated with the threshold of the @xmath67 reaction on @xmath580.5 ev mass neutrinos .
the physics basis for the required clustering of @xmath68 is not provided in those papers .
the capture rate per tritium atom is @xmath69 let us assume , probably much too optimistically , that one can use a megacurie source of tritium ( 1 mcu = 3.7@xmath70 decays / s , i.e. 2.1@xmath71 tritium atoms @xmath58 100 g of tritium ) .
the number of events is then @xmath72 thus , if our assumption about the gravitational clustering is at least nearly correct , the capture rate would be reasonably large . however , the main issue would be whether the primordial background neutrino capture signal would be detectable given the overwhelming rate of the radioactive decay .
electrons from the ordinary @xmath6 decay are distributed over the kinetic energy interval @xmath73 , smeared by the resolution of the detection apparatus . on the other hand , electrons from the background neutrino captures
are monoenergetic with the kinetic energy @xmath74 again smeared by resolution .
thus , the signal to noise ratio will critically depend on the neutrino mass @xmath65 and on the energy resolution @xmath75 .
note that the fraction of electrons in an energy interval of width @xmath75 just below the endpoint is @xmath76 . to appreciate the problem we show in fig.[fig : schematic ] the tail of the spectrum of tritium @xmath6 decay folded with a gaussian resolution function and the signal of the cosmological @xmath5 capture electrons evaluated for @xmath77 and clearly separated in this idealized situation from the background .
remarkably , the ratio of the background neutrino capture rate and the competing @xmath6 decay with final electron within the resolution interval @xmath75 just below the endpoint , does not depend on the corresponding @xmath78 value ( see @xcite , their eq.(23 ) ) and , naturally , on the nuclear matrix element .
for @xmath79 the corresponding ratio is @xmath80 this appears to be a hopelessly small number . before discussing the issues further ,
let us point out that other sources of background , for example the capture of solar @xmath81 neutrinos , are not dangerous .
the total solar @xmath81 neutrino flux is @xmath82
@xmath83 s@xmath84 distributed over 420 kev @xcite .
thus , the flux in the lowest 10 ev is only about @xmath85 @xmath83 s@xmath84 , which is less than 1% of the effective flux of the primordial @xmath5 even for @xmath86 1 ev .
one should note that the discussed method is intresting only as long as neutrinos are non - relativistic with v@xmath87c .
for higher energy neutrinos ( like thermal solar neutrinos that have an estimated flux of @xmath88/@xmath35/sec / mev and energies @xmath58 1 kev @xcite ) signal becomes well separated from radioactive ion decay background , however the number of expected events will be very small , since one is obliged to work with very small amount of active material .
the @xmath58kev mass sterile neutrinos , considered in the literature ( see , e.g. @xcite ) are unobservable due to their extremely small mixing with the active neutrinos . if one could achieve a resolution @xmath75 that is less than the neutrino mass @xmath65 , the signal to background ratio would increase since the @xmath6 spectrum ends at @xmath89 while the electrons from neutrino capture have energy @xmath90 .
the corresponding gain , i.e. the suppression of the tail of the @xmath6 decay spectrum , is estimated in @xcite .
for @xmath91 it is @xmath92 i.e. , it is an extremely steep function of @xmath0 . according to such estimate
a signal to background ratio of the order of unity could be reached if the ratio @xmath93 .
since @xmath65 remains unknown , an experiment with a fixed resolution @xmath75 would be able to observe the background neutrino sea only if @xmath65 is large enough .
a numerical calculation suggests that in fact @xmath94 is enough to achieve a signal to background ratio of order of unity .
this is illustrated in fig .
[ fig : ratio ] where @xmath75 is the full width at half maximum of the assumed gaussian resolution function and the signal as well as the background are centered at @xmath95 and integrated over an interval of width @xmath75 .
the figure also shows that this ratio is such a steep function of @xmath0 that it is essentially independent of the enhancement of @xmath96 due to the gravitational clustering .
( note , however , that the signal itself , i.e. the number of events , is proportional to the clustering ratio @xmath97 . )
figure [ fig : ratio ] shows that in order to achieve sensitivity to sub ev neutrino masses a resolution width below @xmath58 0.5 ev would be necessary .
obviously , that is a very challenging requirement .
we wish now to discuss the possible resonance character of the neutrino capture on nuclei at threshold .
the interest in the possible resonance effects was whetted by the attempts of raghavan @xcite , that even so they are unsuccessful so far , stimulated lively discussion .
for the primordial background neutrinos the de broglie wavelength is extremely large , @xmath98 for @xmath99 = 1.9 k. that estimate is no longer valid for the gravitationally clustered massive neutrinos , that acquire the corresponding virial velocity .
nevertheless , their de broglie wavelength remains macroscopically large . if a resonance reaction could occur , and if it can be somehow described by a breit - wigner type formula where the integrated reaction cross section is @xmath100 a very large cross section could be obtained . here
@xmath101 is the partial width for the elastic scattering , @xmath102 for the capture , and @xmath103 is the total width . in the case of neutrino
induced reactions , obviously , there is only one channel , and all @xmath103 values should be identical .
mikaleyan _ et al . _
@xcite considered a resonance scenario for the endothermic reaction ( not observed as yet ) @xmath104 where @xmath105 is stable , @xmath106 is radioactive with an endpoint @xmath107 , and the captured electron is an orbital one in @xmath105 .
the reaction occurs when @xmath108 , where @xmath109 is the binding energy of the captured electron , so its threshold is @xmath110 below the threshold for the inverse @xmath6 decay .
it is shown in ref .
@xcite that for the process ( [ e:22 ] ) the factor @xmath111 in the cross section formula eq.([e:23 ] ) is compensated by the @xmath112 dependence of the total width @xmath103 ( @xmath113 in this case ) . moreover
, the resonance electron capture can not occur for a radioactive ( hence exothermic ) target @xmath105 . while the reaction ( [ e:24 ] ) is indeed of resonance character and
occurs only when the incoming @xmath43 has a fixed energy , the zero threshold reactions ( with radioactive @xmath105 ) @xmath114 are not of resonance character .
they can proceed for any energy of the incoming @xmath43 , including nonrelativistic energies , but they can not be described by the resonance formula and the cross section is never close to the value of @xmath115 as in eq.([e:23 ] ) .
we have shown that the charged current cross section of nonrelativistic neutrinos scales like @xmath29 in agreement with the generic behavior of the cross sections for slow particles .
that means , in particular , that the charged current reactions with vanishing threshold of nonrelativistic neutrinos have a rate that converges to a finite value as @xmath4 . with that in mind
we evaluate the cross section and reaction rate of the relic @xmath43 on tritium nuclei .
if a modest gravitational clustering enhancement of the relic neutrino number density is present , the number of events is large enough that it might be potentially observable .
the more important issue is the elimination of the overwhelming background of the electrons from tritium @xmath6 decay .
we show that signal / background ratio of the order of unity can be achieved only if the neutrino mass @xmath65 exceeds the characteristic experimental resolution width @xmath75 by a factor of two or more .
thus an energy resolution @xmath1 ev would be required in order to be possible , even in an ideal experiment , to detect relic neutrinos with sub - ev neutrino masses , independently of the gravitational neutrino clustering .
r. lazauskas and c. volpe acknowledge support from the project anr-05-jcjc-0023 `` non standard neutrino properties and their impact in astrophysics and cosmology '' ( neupac ) .
p. vogel appreciates the hospitality of the institut de physique nuclaire , orsay and of the aspen center for physics . | we discuss the cross section formula both for massless and massive neutrinos on stable and radioactive nuclei .
the latter could be of interest for the detection of cosmological neutrinos whose observation is one of the main challenges of modern cosmology .
we analyze the signal to background ratio as a function of the ratio @xmath0 , i.e. the neutrino mass over the detector resolution and show that an energy resolution @xmath1 ev would be required for sub - ev neutrino masses , independently of the gravitational neutrino clustering .
finally we mention the non - resonant character of neutrino capture on radioactive nuclei .
-0.50 in |
we begin by stating the main algorithm .
then , in the following subsections , we discuss each step in more detail , including precise definitions and references for terms and facts that are not commonly known .
[ main algorithm ] + inputs : 1 . a finite group @xmath11 ; 2 . an integer @xmath5 ; 3 .
a set of surface kernel generators @xmath12 determining a family of nonhyperelliptic riemann surfaces @xmath0 of genus @xmath6 with @xmath13 + output : a locally closed set @xmath14 and a family of smooth curves @xmath15 such that for each closed point @xmath16 , the fiber @xmath17 is a smooth genus @xmath6 canonically embedded curve with @xmath18 .
+ 1 . compute the conjugacy classes and character table of @xmath11 .
2 . use the eichler trace formula to compute the character of the action on differentials and on cubics in the canonical ideal .
3 . obtain matrix generators for the action on holomorphic differentials .
4 . use the projection formula to obtain candidate cubics .
5 . compute a flattening stratification and select the locus yielding smooth algebraic curves with degree @xmath19 and genus @xmath6 .
this step is purely for bookkeeping .
it is customary to list the conjugacy classes of @xmath11 in increasing order , and to list the rows in a character table by increasing degree .
however , there is no canonical order to either the conjugacy classes or the irreducible characters . given two different descriptions of a finite group @xmath11 , modern software such as ` magma ` may order the classes or the irreducible characters of @xmath11 differently . hence , we compute and fix these at the beginning of the calculation . here
we define surface kernel generators for the automorphism group of a riemann surface .
these generators determine the riemann surface as a branched cover of @xmath7 and are used in a key formula ( see theorem [ count fixed points ] below ) for counting the number of fixed points of an automorphism .
[ surface kernel definition ] a _ signature _ is a list of integers @xmath20 with @xmath21 , @xmath22 , and @xmath23 . +
a set of _ surface kernel generators _ for a finite group @xmath11 and signature @xmath20 is a sequence of elements @xmath24 such that 1 .
@xmath25 ; 2 .
@xmath26 ; and 3 .
@xmath27 \prod_{i=1}^{r } g_i = { \operatorname{id}}_g$ ] .
surface kernel generators have many other names in other papers ; they are called _ ramification types _ in @xcite and _ generating vectors _ in @xcite . as explained in @xcite*section 3.11 ,
surface kernel generators describe the quotient morphism @xmath28 as a branched cover . here
@xmath0 is a riemann surface of genus @xmath6 , @xmath11 is a subgroup of @xmath3 , the quotient @xmath29 has genus @xmath30 , the quotient morphism branches over @xmath31 points , and the integers @xmath32 describe the ramification over the branch points . in the sequel
we will be primarily interested in large automorphism groups , that is , @xmath1 . in this case
, the riemann - hurwitz formula implies that @xmath33 and @xmath34 .
surface kernel generators are used in the following formula for the number of fixed points of an automorphism : [ count fixed points ] let @xmath35 be an automorphism of order @xmath36 of a riemann surface @xmath0 of genus @xmath37 .
let @xmath38 be part of a set of surface kernel generators for @xmath0 , and let @xmath39 be the orders of these elements .
let @xmath40 be the set of fixed points of @xmath0 where @xmath35 acts on a neighborhood of the fixed point by @xmath41 .
then @xmath42 here @xmath43 is the centralizer of @xmath35 in @xmath11 , and @xmath44 denotes conjugacy .
next we recall the eichler trace formula . for a riemann surface @xmath0 ,
let @xmath45 be the holomorphic cotangent bundle , and let @xmath46 be the sheaf of holomorphic differentials .
the eichler trace formula gives the character of the action of @xmath3 on @xmath47 .
[ eichler trace formula ] suppose @xmath48 , and let @xmath35 be a nontrivial automorphism of @xmath0 of order @xmath49 .
write @xmath50 for the character of the representation of @xmath3 on @xmath47 .
then @xmath51 together , the previous two results give a group - theoretic method for computing the character of the @xmath3 action on @xmath52 starting from a set of surface kernel generators .
we can use the character of @xmath3 on @xmath52 to obtain the character of @xmath3 on quadric and cubics in the canonical ideal as follows .
let @xmath53 be the coordinate ring of @xmath54 , let @xmath55 be the canonical ideal , and let @xmath56 and @xmath57 denote the degree @xmath58 subspaces of @xmath53 and @xmath59 . by noether s theorem , the sequence @xmath60 is exact for each @xmath61 , and by petri s theorem , the canonical ideal is generated either by quadrics or by quadrics and cubics .
thus , beginning with the character of the action on @xmath62 , we may compute the characters of the actions on @xmath63 and @xmath64 and @xmath65 and @xmath66 , and then obtain the characters of the actions on @xmath67 and @xmath68 . from step 2
we have the character of the action on @xmath69 .
we seek matrix generators for this action .
it suffices to find matrix generators for each irreducible @xmath11-module appearing in @xmath69 .
given a finite group @xmath11 and an irreducible character @xmath70 of @xmath11 , software such as ` gap ` @xcite and ` magma ` contain commands for producing matrix generators of a representation @xmath71 of @xmath11 with character @xmath70 .
finding efficient algorithms to produce matrix generators with good properties ( for instance , sparse matrices , or matrices whose entries have small height , or matrices whose entries belong to a low degree extension of @xmath72 ) is a subject of ongoing research @xcite .
it seems that computer algebra systems implement several different algorithms that cover many special cases .
i do not know a reference for a general algorithm .
hence , i briefly present an algorithm that was suggested to me by valery alexeev and james mckernan .
this algorithm is not expected to perform efficiently ; it is included merely to establish that step 3 in algorithm [ main algorithm ] can be performed algorithmically .
+ inputs : 1 . a finite group @xmath11 with generators @xmath73 ; 2 .
an irreducible character @xmath74 of degree @xmath75 .
output : matrices @xmath76 such that the homomorphism @xmath77 is a representation with character @xmath70 1 .
compute matrix generators for the regular representation @xmath71 of @xmath11 .
these matrices are permutation matrices , and hence their entries are in @xmath78 .
2 . use the projection formula ( see theorem [ projection formula theorem ] below ) to compute matrix generators @xmath79 for a representation @xmath80 with character @xmath81 .
let @xmath82 be the smallest field containing @xmath83 .
note that @xmath84 $ ] .
then the matrix generators @xmath79 lie in @xmath85 .
3 . let @xmath86 be indeterminates .
let @xmath87 be the @xmath88 matrix over @xmath89 whose rows are given by the vectors @xmath90 .
let @xmath91 be the determinantal variety @xmath92 .
since representations of finite groups are completely reducible in characteristic zero , the representation @xmath80 is isomorphic over @xmath89 to the direct sum @xmath93 , and therefore @xmath94 is non - empty .
4 . intersect
@xmath0 with generic hyperplanes with coefficients in @xmath89 to obtain a zero - dimensional variety @xmath95 .
5 . if necessary , pass to a finite field extension @xmath96 of @xmath89 to obtain a reduced closed point @xmath97 .
the point @xmath98 ( regarded as a vector in @xmath99 ) generates the desired representation .
an example where this algorithm is used to produce matrix generators for the degree two irreducible representation of the symmetric group @xmath100 is available at my webpage @xcite .
finally , we note that in @xcite , streit describes a method for producing matrix generators for the action of @xmath3 on @xmath69 for some bely curves .
recall the projection formula for representations of finite groups .
( see for instance @xcite formula ( 2.31 ) ) .
[ projection formula theorem ] let @xmath71 be a finite - dimensional representation of a finite group @xmath11 over @xmath101 .
let @xmath102 be the irreducible representations of @xmath11 , let @xmath103 be their characters , and let @xmath104 .
let @xmath105 be the projection onto the @xmath106 isotypical component of @xmath71 .
then @xmath107 from step 3 , we have matrix generators for the @xmath11 action on @xmath108 .
thus , we can compute matrix generators for the actions on @xmath109 and @xmath100 , and use the projection formula to compute the isotypical subspace @xmath110 of degree @xmath58 polynomials on which @xmath11 acts with character @xmath111 . in some a few examples , we have @xmath112 , but more commonly , we have strict containment @xmath113 . in this case
we write elements of @xmath114 as generic linear combinations of the basis elements of @xmath110 and then seek coefficients that yield a smooth algebraic curve with the correct degree and genus .
the coefficients used to form these generic linear combinations form the base space @xmath115 of the family @xmath116 produced by the main algorithm .
let @xmath117 be a projective morphism with @xmath53 a reduced noetherian scheme .
then there exist locally closed subsets @xmath118 such that @xmath119 and @xmath120 is flat .
the stratification @xmath121 is called a flattening stratification for the map @xmath122 .
since @xmath53 is reduced , flatness implies that over each stratum , the hilbert polynomial of the fibers is constant .
we find the stratum with hilbert polynomial @xmath123 , then intersect this stratum with the locus where the fibers are smooth .
this completes the algorithm .
flattening stratifications have been an important tool in theoretical algebraic geometry for over 50 years .
there exist grbner basis techniques for computing flattening stratifications ; in the computational literature , these are typically called _ comprehensive _ or _ parametric grbner bases _ , or _ grbner systems_. the foundational work on this problem was begun by weispfenning , and many authors , including manubens and montes , suzuki and sato , nabeshima , and kapur , sun , and wang , have made important improvements on the original algorithm @xcite .
the size of a grbner basis can grow very quickly with the number of variables and generators of an ideal , and unfortunately , even the most recent software can not compute flattening stratifications for the examples we consider . thus , in section [ partial flattening stratification section ] below , we discuss a strategy for circumventing this obstacle .
many steps of algorithm [ main algorithm ] can be run using a computer algebra system , but even for modest examples , the flattening stratification required in the final step is intractable .
therefore we discuss various heuristics that can be employed to speed the computation . given a set of surface kernel generators , it is useful to discover as early as possible whether the corresponding riemann surface is hyperelliptic , trigonal , a plane quintic , or none of these .
we discuss these properties in turn .
_ hyperelliptic riemann surfaces_. algorithm [ main algorithm ] supposes that one begins with surface kernel generators corresponding to a nonhyperelliptic curve .
however , we can easily test for hyperellipticity if this property is not known in advance . a riemann surface
@xmath0 is hyperelliptic if and only if @xmath3 contains a central involution with @xmath124 fixed points .
thus , given a set of surface kernel generators , we can search for a central involution and count its fixed points using theorem [ count fixed points ] ( or even better , using @xcite*lemma 10.4 ) . in @xcite
, shaska gives equations of the form @xmath125 for hyperelliptic curves with automorphisms .
additionally , we can use the algorithm described in @xcite to get the equations of @xmath126 under a linear series such as the transcanonical embedding or bicanonical embedding .
so suppose the riemann surface is not hyperelliptic . by petri s theorem ,
the canonical ideal is generated by quadrics if @xmath0 is not hyperelliptic , not trigonal , and not a plane quintic .
thus , ruling out these possibilities allows us to work with quadrics instead of cubics , which significantly speeds up the algorithm .
this leads us to consider trigonal riemann surfaces and plane quintics . _ trigonal riemann surfaces . _
trigonal riemann surfaces may be divided into two types : cyclic trigonal and general trigonal @xcite .
cyclic trigonal curves can be detected by searching for degree three elements fixing @xmath127 points .
their automorphism groups have been classified @xcite*theorem 2.1 , and one may hope for a paper treating equations of cyclic trigonal riemann surfaces as the paper @xcite treats equations of hyperelliptic riemann surfaces .
less is known about general trigonal curves .
we have arakawa s bounds @xcite*remark 5 and a few additional necessary conditions @xcite*prop . 4 and lemma 5 .
we will not say more about general trigonal riemann surfaces here because after studying the riemann surfaces with large automorphism groups with genus @xmath128 , we learn _ a posteriori _ that very few of them are general trigonal . _ plane quintics .
_ plane quintics only occur in genus 6 , and the canonical model of a plane quintic lies on the veronese surface in @xmath129 .
thus , we have a necessary condition : @xmath0 is a plane quintic only if @xmath130 for some ( possibly reducible ) three - dimensional representation @xmath71 of @xmath11 . in practice
, it is generally quite fast to discover whether a nonhyperelliptic non - cyclic trigonal genus 6 riemann surface is a plane quintic . in this section
we use several notions from the theory of grbner bases .
we will not recall all the definitions here , and instead refer to @xcite*chapter 15 for the details .
the algorithms for comprehensive grbner bases described in section [ flattening stratification section ] all begin with the same observation .
let @xmath131 $ ] be a polynomial ring over a field .
let @xmath132 be a multiplicative term order on @xmath53 .
then a theorem of macaulay states that the hilbert function of @xmath59 is the same as the hilbert function of its initial ideal with respect to this term order ( see @xcite*theorem 15.26 ) .
therefore , whenever two ideals in @xmath53 have grbner bases with the same leading monomials with respect to some term order , they will have the same initial ideal for that term order , hence they must have the same hilbert function and hilbert polynomial , and therefore they will lie in the same stratum of a flattening stratification . to reach a different stratum of the flattening stratification ,
it is necessary to alter the leading terms of the grbner basis for instance , by restricting to the locus where that coefficient vanishes .
here is a brief example to illustrate this idea . let @xmath133 have coordinates @xmath134 , and let @xmath135 have coordinates @xmath136,@xmath137,@xmath138,@xmath139 .
the ideal @xmath140 defines a 2-parameter family of subschemes of @xmath135 . a grbner basis for @xmath59 in @xmath141[x_0,x_1,x_2,x_3]$ ] with respect to the lexicographic term order is @xmath142 over the locus where @xmath143 , @xmath144 , and @xmath145 are invertible , the initial ideal is + @xmath146 with hilbert polynomial @xmath147 . on the other hand , when @xmath148 , or @xmath149 , or @xmath150 , we get a different initial ideal and hilbert polynomial .
for example , the locus @xmath151 yields the twisted cubic with @xmath152 .
note that to discover this locus , it is not necessary to compute the entire grbner basis ; it would suffice for instance to compute the s - pair reduction for the first two generators , which yields @xmath153 .
modern software packages by nabeshima , montes , and kapur , sun , and wang can completely analyze this example .
however , these packages did not yield answers on the problems that arose in this work .
therefore , i used the strategy outlined above .
i partially computed a grbner basis in ` macaulay2 ` @xcite , and set some coefficients to zero . remarkably , this was sufficient to obtain the equations of the genus @xmath154 riemann surfaces with large automorphism groups .
some of the families analyzed in this manner had as many as six coefficients @xmath155 .
magaard , shaska , shpectorov , and vlklein s tables show that there exists a smooth , compact genus 7 riemann surface with automorphism group @xmath11 given by the group labeled @xmath156 in the gap library of small finite groups .
it has @xmath157 .
the quotient morphism is branched over 3 points of @xmath158 , and the ramification indices over these points are 2 , 4 , and 16 .
a naive search for a set of surface kernel generators in this group yields elements @xmath159 and @xmath160 with orders 2 and 4 such that @xmath161 has order 16 .
there are four relations among these generators : @xmath162 _ step 1 .
_ we use ` magma ` to compute the conjugacy classes and character table of @xmath11 .
there are 16 conjugacy classes . for convenience ,
write @xmath163 .
then a list of representatives of the conjugacy classes is @xmath164 next we compute the character table .
the irreducible characters are given below by their values on the sixteen conjugacy classes .
@xmath165 @xmath166 _ step 2 .
_ let @xmath167 be the irreducible @xmath11-module with character @xmath103 given by the table above .
for any @xmath11-module @xmath71 , let @xmath168 be its decomposition into irreducible @xmath11-modules .
we use the eichler trace formula in ` magma ` to compute these multiplicities @xmath169 for several relevant @xmath11-modules .
let @xmath170 $ ] , and let @xmath56 denote polynomials of degree @xmath58 .
let @xmath57 be the kernel defined by @xmath171 then we have @xmath172 we use ` gap ` to obtain matrix representatives of a @xmath11 action with character equal to the character of the @xmath11 action on @xmath173 .
such a representation is obtained by mapping the generators @xmath159 and @xmath160 to the matrices below .
@xmath174 , \qquad \left [ \begin{array}{rrrrrrr } \zeta_{8}^2 & 0 & 0 & 0 & 0 & 0 & 0\\ 0 & 0 & -\zeta_{8 } & 0 & 0 & 0 & 0\\ 0 & -\zeta_{8}^3 & 0 & 0 & 0 & 0 & 0\\ 0 & 0 & 0 & 0 & 0 & 0 & 1\\ 0 & 0 & 0 & 0 & 0 & -\zeta_{8 } & 0\\ 0 & 0 & 0 & 0 & -\zeta_{8}^3 & 0 & 0\\ 0 & 0 & 0 & 1 & 0 & 0 & 0 \end{array } \right]\ ] ] the decomposition of @xmath173 as a sum of three irreducible @xmath11-modules gives rise to the block diagonal form of these matrices .
_ we use the projection formula in ` magma ` to decompose the @xmath11-module of quadrics @xmath109 into its isotypical components .
when an isotypical component has multiplicity greater than 1 , we ( noncanonically ) choose ordered bases so that the @xmath11 action is given by the same matrices on each ordered basis .
@xmath175 the first isotypical subspace yields a polynomial of the form @xmath176 .
we may assume that @xmath143 and @xmath144 are nonzero , scale @xmath136 to make @xmath177 , and then divide by @xmath143 to obtain the polynomial @xmath178 .
the second isotypical subspace yields the polynomial @xmath179 . the third isotypical subspace yields polynomials of the form @xmath180 and @xmath181 .
we assume that @xmath182 and @xmath183 are nonzero , scale @xmath184 to make @xmath185 , then divide by @xmath182 and @xmath183 . in the remaining isotypical subspaces no further scaling is possible , and
hence we are left with two undetermined coefficients @xmath186 and @xmath187 .
thus , a riemann surface in this family has an ideal of the form @xmath188 _ step 5 .
_ to find values of the coefficients @xmath189 that yield a smooth curve , we partially compute a flattening stratification
. begin buchberger s algorithm .
we compute the s - pair reductions between the generators and find that @xmath190 therefore , in buchberger s algorithm , these polynomials will be added to the grbner basis .
this suggests that we study the locus given by the equations @xmath191 and @xmath192 as an interesting stratum in the flattening stratification .
we check in ` magma ` that the values @xmath193 and @xmath194 yield a smooth genus 7 curve in @xmath195 with the desired automorphism group . from these equations
, we can compute the betti table of this ideal : [ cols="^,^,^,^,^,^",options="header " , ] + genus 7 , locus 13 : group @xmath196 , signature ( 2,15,30 ) , hyperelliptic , @xmath197 | we present an algorithm for computing equations of canonically embedded riemann surfaces with automorphisms .
a variant of this algorithm with many heuristic improvements is used to produce equations of riemann surfaces @xmath0 with large automorphism groups ( that is , @xmath1 ) for genus @xmath2 .
the main tools are the eichler trace formula for the character of the action of @xmath3 on holomorphic differentials , algorithms for producing matrix generators of a representation of a finite group with a specified irreducible character , and grbner basis techniques for computing flattening stratifications .
riemann surfaces ( or algebraic curves ) with automorphisms have been important objects of study in complex analysis , algebraic geometry , number theory , and theoretical physics for over a century , as their symmetries often permit us to do calculations that would otherwise be intractable .
such riemann surfaces are special in the sense that a general riemann surface of genus @xmath4 has no nontrivial automorphisms .
moreover , the group of automorphisms of a riemann surface of genus @xmath5 is finite . breuer and conder performed computer searches that for each genus @xmath6 list the riemann surfaces of genus @xmath6 with large automorphism groups ( that is , @xmath1 ) . specifically , they list sets of surface kernel generators ( see definition [ surface kernel definition ] below ) , which describe these riemann surfaces as branched covers of @xmath7 .
breuer s list extends to genus @xmath8 , and conder s list extends to genus @xmath9 @xcite . even for small values of @xmath6 , these lists are extremely large , as a surface @xmath0 may appear several times for various subgroups of its full automorphism group
. in @xcite , magaard , shaska , shpectorov , and vlklein refined breuer s list by determining which surface kernel generators correspond to the full automorphism group of the riemann surface . to my knowledge , at this time there is no general algorithm published in the literature for producing equations of these riemann surfaces under any embedding from this data .
here , i present an algorithm to compute canonical equations of nonhyperelliptic riemann surfaces with automorphisms .
the main tools are the eichler trace formula for the character of the action of @xmath3 on holomorphic differentials , algorithms for producing matrix generators of a representation of a finite group with a specified irreducible character , and grbner basis techniques for computing flattening stratifications .
a variant of this algorithm with many heuristic improvements is used to produce equations of the nonhyperelliptic riemann surfaces with genus @xmath2 satisfying @xmath1 . here is an outline of the paper . in section [ algorithm section ] , i describe the main algorithm . in section [ heuristics section ] , i describe several heuristics that simplify or speed up the main algorithm . in section [ example section ] , i describe one example in detail , a genus 7 riemann surface with 64 automorphisms . in section [ results section ] , i give equations of selected canonically embedded riemann surfaces with @xmath10 along with matrix surface kernel generators .
a large number of people have supplied advice and encouragement on this project over a period of many years .
the undergraduates from my introductory vigre research group held in fall 2010 at the university of georgia and my student darcy chanin performed many calculations for genus 4 , 5 , and 6 surfaces .
i am grateful to the computational algebra group at the university of sydney , especially john cannon and mark watkins , for hosting me for a visit in june 2011 where i began programming the main algorithm in ` magma ` @xcite .
i have had many helpful conversations with my classmates and colleagues at columbia university , the university of georgia , and fordham university .
valery alexeev and james mckernan suggested the algorithm for matrix generators of representations outlined in section [ matrix generators section ] .
finally , i am grateful to jennifer paulhus , tony shaska , and john voight , whose encouragement was essential in completing this project .
this work was partially supported by the university of georgia s nsf vigre grant dms-03040000 , a simons foundation travel grant , and a fordham faculty research grant .
my webpage for this project is @xcite .
this page contains links to the latest version of my ` magma ` code , files detailing the calculations for specific examples , and many equations that are omitted in the tables in section [ results section ] . in future work , jennifer paulhus and
i plan to include much of the data described in this paper and on the website @xcite in the l - functions and modular forms database at ` lmfdb.org ` . |
the bjorken sum rule @xcite relates an integral over the spin distributions of quarks inside the nucleon to its axial charge .
this relation has been essential for understanding the nucleon spin structure and establishing , _ via _ its @xmath6-dependence , that quantum chromodynamics ( qcd ) describes the strong force when spin is included .
the bjorken integral has been measured in polarized deep inelastic lepton scattering ( dis ) at slac , cern and desy @xcite-@xcite and at moderate four - momentum transfer squared @xmath6 at jefferson lab ( jlab ) @xcite , see e.g. ref .
@xcite for a review .
the variable @xmath6 is inversely related to the space - time scale at which the nucleon is probed . in the perturbative qcd ( pqcd )
domain ( high @xmath6 ) the sum rule reads @xcite : @xmath7 + \sum_{i=2}^{\infty}{\frac{\mu_{2i}^{p - n}(q^{2})}{q^{2i-2 } } } \nonumber\end{aligned}\ ] ] where @xmath8 and @xmath9 are the spin - dependent proton and neutron structure functions , @xmath10 is the nucleon axial charge that controls the strength of neutron @xmath11-decay , @xmath12 is the strong coupling strength and @xmath13 , with @xmath14 the energy transfer and @xmath15 the nucleon mass .
the bracket term ( @xmath16 , known as the leading twist term ) is mildly dependent on @xmath6 due to pqcd soft gluon radiation .
the other term contains non - perturbative power corrections ( higher twists ) .
these are quark and gluon correlations that need to be understood to describe the nucleon structure away from the large @xmath6 limit .
the @xmath3-dependence of @xmath17 is calculable in principle from pqcd . in practice
, this has been done for @xmath16 and @xmath18 only @xcite . we stress that , as is almost always the case with pqcd , although the @xmath3-dependences are known , the absolute values of @xmath16 and @xmath18 are unknown and need to be measured or computed by non - perturbative means . besides its contribution to establishing pqcd ( at high @xmath6 )
, the bjorken sum rule can be used to extract higher twists , to check lattice qcd calculations ( at moderate @xmath6 ) , and to test effective theories of the strong force ( at low @xmath6 ) . in addition
, bjorken sum data and phenomenological models at lower @xmath6 can be described with a nearly constant `` effective strong coupling '' @xmath19 @xcite .
the lack of @xmath6-dependence of @xmath19 opens new avenues for non - perturbative qcd calculations using the ads / cft correspondence @xcite .
the elastic contribution to the bjorken sum is usually not included because the generalized bjorken sum rule is derived at large @xmath6 where such contribution is negligible .
furthermore , the bjorken sum rule naturally connects to the gerasimov - drell - hearn sum rule @xcite in which the elastic is inexistent .
consequently , when presenting the experimental measurement of the bjorken sum , the elastic contribution will not be included .
we refer to ref .
@xcite for a discussion on whether to include or not the elastic contribution to the gdh sum rule .
however , for higher twist analysis , all reactions should be included for a meaningful higher twist extraction @xcite-@xcite .
therefore , in the part of the paper discussing higher twist extraction , the elastic contribution to the bjorken sum will be added . in this paper , new data from the jlab clas
eg1b experiment @xcite-@xcite taken on polarized proton and deuteron targets are used to extract the bjorken integral over an extended @xmath3 range : @xmath20 gev@xmath1 compared to the previous jlab range @xmath21 gev@xmath1 @xcite .
the extension down to @xmath22 gev@xmath1 allows us to compare to chiral perturbation theory ( @xmath23 ) calculations in a domain where the chiral approximation should be valid .
the moderate @xmath3 range had been precisely measured @xcite .
the new data set , of equivalent precision , provides a useful check .
in particular , it verifies the neutron results , which come mostly from @xmath24he in ref .
@xcite and from the deuteron in this paper . at larger @xmath3 ( @xmath25 1 gev@xmath1 , where eq .
[ eq : bj(q2 ) ] holds ) , higher twists can now be studied with a statistical precision typically improved by a factor of 2 .
previous work @xcite has shown the necessity of precise @xmath3 mapping at moderate @xmath3 ( @xmath25 1 gev@xmath1 ) because of the surprisingly small size of the overall higher twist effect .
one might be tempted to lower the @xmath3 values at which the analysis is done ( see eq .
[ eq : bj(q2 ) ] ) but this is not reliable due to the fast @xmath26 rise of twist @xmath27 contributions at low @xmath3 and to the increasing uncertainty of the evolution of the twist-2 parts . the main contributor at low @xmath6 to this uncertainty is the strong coupling constant @xmath12 .
the bjorken integral is advantageous compared to the individual moments @xmath28 and @xmath29 because of simplifications arising from its non - singlet ( p - n ) nature : at moderate @xmath6 lattice qcd calculations are easier and more reliable because disconnected diagrams , which can not be easily computed on the lattice , cancel out . at higher
@xmath6 , the ( p - n ) simplification provides a sum rule ( the bjorken sum rule ) based on more solid grounds than the sum rules for individual nucleons ( the ellis jaffe sum rules @xcite that necessitate additional assumptions ) . at low @xmath6 ,
the ( p - n ) subtraction cancels the @xmath30 resonance contribution which makes the @xmath23 calculations significantly more reliable @xcite . by a similar argument
, the transverse - longitudinal polarizability @xmath31 @xcite , a higher moment of spin structure functions , also provides a reliable test of @xmath23 computations .
( in that case , the @xmath30 contribution is suppressed at low @xmath6 because the n-@xmath32 transition is mostly transverse , making the longitudinal - transverse ( lt ) interference term very small . ) nevertheless , calculations based on @xmath23 and data for @xmath31 on the neutron @xcite strongly disagree .
this calls for more low @xmath6 studies , especially the yet unmeasured @xmath33 @xcite .
the data discussed in this paper were taken with a longitudinally polarized target and hence can not be used to extract @xmath34 .
however , the generalized forward spin polarizability @xmath5 can be obtained and , just like the bjorken integral , its isovector part @xmath35 offers the same advantages as @xmath31 for checking the calculation techniques of @xmath23 .
we will also report on these results .
the measurements of structure functions @xmath8 and @xmath36 are described in refs .
the data cover an invariant mass range up to @xmath37 3 gev for 0.054 @xmath38 2.92 gev@xmath4 .
since experimental moments are integrated over a finite @xmath39 range , the data have to be supplemented by models for large @xmath39 .
we used the model described in ref . @xcite down to @xmath40 .
this part is known from dis experiments .
the rest is determined using a regge parametrization @xcite which was compared to that of bass and brisudova @xcite and found consistent with it .
a parameterization was also used to estimate the contributions between pion threshold ( 1.08 gev ) and 1.15 gev @xcite .
the bjorken integral is obtained from @xmath41 and @xmath42 assuming : @xmath43 with the deuteron d - state probability @xmath44 @xcite .
the data are given in table i ( a more detailed table is given in @xcite ) and shown in fig .
[ fig : bjsr ] . the elastic contribution ( @xmath45 )
is excluded .
data from slac e143 @xcite , hermes @xcite , jlab clas eg1a ( proton and deuteron ) , and jlab hall a e94010 ( neutron from @xmath24he ) combined with clas eg1a ( proton ) @xcite are also shown for comparison . .
the solid blue circles give the results from this work with the horizontal band giving the systematic uncertainties .
other symbols show the data from experiments e143 @xcite ( open diamonds ) , e155 @xcite ( open star ) , hermes @xcite ( open triangles ) and jlab @xcite ( open circles and open squares ) . for those ,
the error bars represent the quadratic sum of the statistic and systematic uncertainties .
the gray band represents the leading - twist nnlo pqcd calculation .
the curves correspond to @xmath23 calculations ( @xcite ) and phenomenological models ( @xcite ) . ]
.the measured ( @xmath46 ) and total ( @xmath47 ) bjorken integrals for different @xmath6 points ( in gev@xmath1 ) . the experimental systematic uncertainty @xmath48 is given in the 3@xmath49 column .
total systematics uncertainty , including the low and large-@xmath50 extrapolations , ( @xmath51 ) and statistical uncertainty ( @xmath52 ) on @xmath47 are given in the @xmath53 and @xmath54 columns . [ cols="^,^,^,^,^,^",options="header " , ]
the bjorken integral was extracted from polarized proton and deuteron data for @xmath55 gev@xmath4 . the results for intermediate @xmath3
( the parton to hadron transition domain ) are consistent with previous jlab data in which the neutron information was extracted from polarized @xmath56he .
this region exhibits a strong @xmath3-behavior , both from pqcd evolution and from some higher - twist effects . on the other hand , in the high-@xmath6 domain the bjorken integral
is rather flat .
the data together with kinematic constraints at @xmath57 also suggest a small @xmath6-dependence , in qualitative agreement with the generalized gdh sum predictions . at the lowest @xmath6 accessed by our data ,
@xmath23 calculations agree better with the bjorken integral ( an isovector quantity in which the @xmath30 resonance does not contribute ) than with moments on individual nucleons .
this is not trivial since the @xmath23 calculations fail to describe the generalized spin polarizability @xmath31 in which the @xmath30 is also suppressed .
it is clear from previously published data on @xmath31 and our analysis of @xmath5 that the @xmath30 resonance contribution is not responsible for the discrepancy between data and calculations .
the discrepancy between the @xmath23 calculations and the data occurs in all isospin channels , which makes it less likely that it is due to the contribution from heavier mesons in the chiral expansion .
the low @xmath6 @xmath23 regime has been recently mapped by two additional dedicated experiments in clas using polarized proton @xcite and deuteron targets @xcite and one in hall a using polarized @xmath56he @xcite .
these experiments will provide further precision tests of @xmath23 calculation techniques .
the moderate @xmath3 data ( 1 to 3 gev@xmath1 ) allow us to extract higher twist contributions and color polarizabilities . the twist-4 coefficient was found to be large : @xmath59 at @xmath60 gev@xmath4 ( compare to @xmath61 , @xmath62 and @xmath63 ) .
the uncertainty on @xmath64 remains relatively large ( @xmath65 ) ; however , we have completed several systematic studies both with the existing data as well as simulated data ( with no statistic fluctuations ) that indicate our result is stable .
the sign and magnitude of @xmath2 agree with a recent analysis performed on @xmath66 directly @xcite .
the observation that higher twist effects on @xmath67 are small overall does not imply that the net higher twist effect on the structure function @xmath68 is small at any @xmath50 .
it is important to study the @xmath50-dependence of the higher twists , as is done in ref .
@xcite . that @xmath69 is significantly larger than @xmath70 , and that @xmath71 , agrees well with the prediction of the two - scale model @xcite .
overall the net effect of higher twists is small , because of a cancellation between the twist 4 and twist 6 terms that are of similar sizes but opposite signs .
this trend has also been seen for higher twist analyses done on the unpolarized structure function @xmath72 @xcite .
this can be interpreted within a vector dominance framework : the oscillating signs arise from the development in series of the vector meson propagator @xmath73 where @xmath74 is the meson mass .
this work is supported by the u.s .
department of energy ( doe ) and the u.s . national science foundation .
the jefferson science associates operate the thomas jefferson national accelerator facility for the doe under contract de - ac05 - 84er40150 .
hermes collaboration : k. ackerstaff , _ et al .
_ , phys . lett .
* b 404 * , 383 ( 1997 ) ; * b 444 * , 531 ( 1998 ) ; a. airapetian , _ et al .
_ , phys . lett .
* b 442 * , 484 ( 1998 ) ; phys . rev .
lett . * 90 * , 092002 ( 2003 ) ; eur .
j. * c 26 * , 527 ( 2003 ) ; phys .
rev . * d 75 * , 012007 ( 2007 ) .
m. lacombe _
et al . , _ phys .
* c 21 * , 861 ( 1980 ) ; r. machleidt , k. holinde , c. elster , phys .
149 1 ( 1987 ) ; m.j .
zuilhof , j.a .
tjon , phys . rev .
* c 22 * , 2369 ( 1980 ) ; k. kotthoff , r. machleidt , d. schutte , nucl . phys . * a 264 * , 484 ( 1976 ) ; b. desplanques phys . lett .
* b 203 * , 200 ( 1988 ) . | we present the bjorken integral extracted from jefferson lab experiment eg1b for @xmath0 gev@xmath1 .
the integral is fit to extract the twist-4 element @xmath2 which appears to be relatively large and negative .
systematic studies of this higher twist analysis establish its legitimacy at @xmath3 around 1 gev@xmath4 .
we also performed an isospin decomposition of the generalized forward spin polarizability @xmath5 .
although its isovector part provides a reliable test of the calculation techniques of chiral perturbation theory , our data disagree with the calculations . |
we thank kaust research computing for supplying the computational resources for this study . n.s .
acknowledges sabic for financial support .
p. giannozzi , s. baroni , n. bonini , m. calandra , r. car , c. cavazzoni , d. ceresoli , g. l. chiarotti , m. cococcioni , i. dabo , a. dal corso , s. de gironcoli , s. fabris , g. fratesi , r. gebauer , u. gerstmann , c. gougoussis , a. kokalj , m. lazzeri , l. martin - samos , n. marzari , f. mauri , r. mazzarello , s. paolini , a. pasquarello , l. paulatto , c. sbraccia , s. scandolo , g. sclauzero , a. p. seitsonen , a. smogunov , p. umari , and r. m. wentzcovitch , j. phys . : condens .
matter * 21 * , 395502 ( 2009 ) .
se@xmath1 after relaxation : thickness @xmath32 ( black numbers ) and separation @xmath33 ( blue numbers ) .
all values are given in .
the inset addresses the optimization of the lattice parameter for a single ql.,scaledwidth=60.0% ] | the structural stability , electronic structure , and thermal transport properties of one to six quintuple layers ( qls ) of bi@xmath0se@xmath1 are investigated by van der waals density functional theory and semi - classical boltzmann theory
. the bandgap amounts to 0.41 ev for a single ql and reduces to 0.23 ev when the number of qls increases to six .
a single ql has a significantly higher thermoelectric figure of merit ( 0.27 ) than the bulk material ( 0.10 ) , which can be further enhanced to 0.30 by introducing 2.5% compressive strain .
positive phonon frequencies under strain indicate that the structural stability is maintained .
thermoelectric ( te ) materials have been extensively studied for key applications such as the conversion of waste heat into electricity .
they have huge potential to contribute to a solution of the energy and environment crisis .
the efficiency of te devices is determined by the dimensionless figure of merit @xmath2 , where @xmath3 is the electrical conductivity , @xmath4 is the seebeck coefficient , @xmath5 is the temperature , and @xmath6 is the thermal conductivity .
the latter comprises lattice ( @xmath7 ) and electronic ( @xmath8 ) contributions : @xmath9 . in order to achieve a high efficiency , the te material has to be a good electrical and poor thermal conductor and , at the same time , has to possess a high seebeck coefficient , see , for example , ref .
@xcite and the references therein .
nanostructuring , alloying , and the application of strain are effective and frequently used approaches @xcite to improve te properties , because the phonon scattering is enhanced and the thermal conductivity consequently is reduced .
the class of topological insulators has been proposed to be of particular potential for te materials @xcite .
a well known example is bismuth telluride ( bi@xmath0te@xmath1 ) , which shows a high figure of merit of about 1 @xcite , but also many other members of this family of materials , such as bi@xmath0se@xmath1 and sb@xmath0te@xmath1 , are topological insulators with an energy gap in the bulk and gapless surface states .
the high mobility of the surface electrons combined with the bulk energy gap give rise to the superior thermoelectric properties .
in particular , superlattices of te - based materials ( bi@xmath0te@xmath1/sb@xmath0te@xmath1 @xcite , pbse@xmath10te@xmath11/pbte @xcite ) and nanocrystalline bi@xmath12sb@xmath13te@xmath1 @xcite have been reported to exhibit a high @xmath14 .
however , the best performing materials contain te , which is an expensive element and of limited availability .
therefore , alternative materials ( for example based on se ) are required .
bi@xmath0se@xmath1 is a topological insulator with an energy gap of 0.3 ev in the bulk .
it has a rhombohedral crystal structure with lattice parameters of @xmath15 and @xmath16 ( space group @xmath17 ) @xcite .
the layered structure is based on slabs of five shifted se and bi atomic layers stacked along the @xmath18-axis with the sequence se(1)-bi - se(2)-bi - se(1 ) .
such a slab is called quintuple layer ( ql ) and has a thickness of about 1 nm . along the @xmath18-axis the qls are connected by weak van der waals ( vdw ) forces .
although the vdw interaction is relatively weak , it still plays a crucial role for the electronic and te properties .
various approaches have been proposed to improve the te properties of bulk bi@xmath0se@xmath1 , such as the introduction of point defects and doping of sb , cu , or ca atoms to tune the carrier concentration @xcite .
recent reports show that mechanical exfoliation of a single ql and of slabs of a few qls of bi@xmath0se@xmath1 is possible and that excellent te properties can be expected in such thin films @xcite .
sun and coworkers @xcite have shown that the te behavior of a single ql of bi@xmath0se@xmath1 is enhanced because of a reduced @xmath6 , as compared to the bulk value .
when the number of qls increases the bandgap of bi@xmath0se@xmath1 decreases and is finally closed at a thickness of eight qls @xcite .
this strong variation of the bandgap with the thickness will play a vital role for the tunability of the te properties . in this context , we investigate the effect of the thickness in terms of the number of qls on the electronic structure and transport behavior .
we will demonstrate a high figure of merit for a single ql of bi@xmath0se@xmath1 and will argue that the value can be further enhanced by tensile strain ( for which we show that the material remains stable ) .
we have carried out first principles calculations based on density functional theory , employing the generalized gradient approximation of the exchange correlation potential in the perdew - burke - ernzerhof flavor .
the vdw interaction is taken into account by means of the b97-d functional @xcite as implemented in the quantumespresso code @xcite .
all calculations are performed with a plane wave energy cutoff of 544 ev and a monkhorst - pack @xmath19 k - mesh for the brillouin zone integrations .
furthermore , we employ finer @xmath20 and @xmath21 k - meshes for calculating the density of states ( dos ) and transport properties , respectively .
all structures are optimized until an energy convergence of 10@xmath22 ev and a force convergence of 0.025 ev / are reached .
a 15 vacuum slab is used to prevent artificial interaction with periodic images due to the employed periodic boundary conditions .
the phonon frequencies are determined by density functional perturbation theory for evaluating the structural stability @xcite .
once self - consistency is achieved in the first principles calculations , the transport properties are calculated using the semi - classical boltzmann theory within the constant relaxation time approximation , as implemented in the boltztrap code @xcite .
this approach yields accurate results for various types of te materials @xcite .
the transport function is given by @xmath23 the electrical conductivity by @xmath24 and the seebeck coefficient by @xmath25 where @xmath26 is the fermi function , @xmath27 the scattering time , @xmath28 the fermi velocity , and @xmath29 the chemical potential . in order
to built supercells for one to six qls of bi@xmath0se@xmath1 , we start from a single ql using the bulk structural parameters and optimize the in - plane lattice constant , obtaining a slight elongation to @xmath30 . the thickness of the ql increases from the experimental bulk value of 6.966 @xcite to 7.096 , similar to the results in ref .
@xcite .
we then keep the @xmath31 lattice constant for all systems fixed at the value obtained for the single ql and relax in each case the atomic positions .
the resulting values for the layer thickness ( @xmath32 ) and separation between adjacent qls ( @xmath33 ) are summarized in fig . 1 .
while effects of the vdw correction expectedly are negligible for a single ql , our values for @xmath32 and @xmath33 for two and more qls are larger than previously reported @xcite .
we note that we use an optimized lattice constant , while these authors have employed the bulk value .
the effect of the vdw correction is significant and can not be ignored in the case of two and more qls in order to obtain correct results for @xmath32 and @xmath33 . as a consequence ,
our calculated energy gaps better reproduce the experiment @xcite than ref .
@xcite , see the band structures of the relaxed systems in fig .
2 .
this improved agreement is due to the larger @xmath34 ( as compared to @xmath35 ) .
our results show that the single ql has a bandgap of 0.41 ev , which decreases step by step to 0.23 ev for six qls , see fig .
3 .
the largest seebeck coefficient and figure of merit thus are expected for a single ql of bi@xmath0se@xmath1 .
we next apply both compressive and tensile strains of 2.5% and 5% to explore the effects on the te response . in the single ql case
the layer thickness is reduced to 6.725 and 6.920 for 5% and 2.5% tensile strain , respectively , and grows to 7.470 and 7.280 for 5% and 2.5% compressive strain .
particularly , at 2.5% tensile strain the value is almost equal to the bulk lattice constant .
the band structures of a single ql of bi@xmath0se@xmath1 under strain are depicted in fig . 4 together with the corresponding phonon dispersions : the bandgap reaches 0.60 ev for 5% compressive strain and decreases to 0.16 ev for 5% tensile strain .
2.5% tensile strain applied to a single ql of bi@xmath0se@xmath1 has about the same effect as an increase of the thickness to 4 qls .
this is just an example to demosntrate that the bandgap can be tune almost idependently by the strain and the number of qls , which is a great advantage from the technology point of view .
we obtain positive phonon frequencies in the whole range of strain under study for the single ql , which demonstrates a wide stability window .
there are 15 phonon modes at the center of the brillouin zone , 12 optical and 3 acoustical .
the optical modes consist of a@xmath36 and a@xmath37 out - of - plane modes and e@xmath38 and e@xmath39 in - plane modes .
nanostructured bi@xmath0se@xmath1 is subject to a red shift of the frequencies of the above modes due to phonon softening .
the out - of - plane phonon softening is explained by the finite size along the @xmath18-axis , resulting in a reduction of the restoring force . in an unstrained single ql of bi@xmath0se@xmath1 ,
the e@xmath39 and a@xmath37 frequencies are 129.3 @xmath40 and 174.8 @xmath40 , respectively , and thus close to the experimental bulk values of 131 @xmath40 and 174 @xmath40 @xcite .
strain effects on the a@xmath37 frequency are addressed in fig . 5 in comparison to the variation of the bandgap , in order to demonstrate the analogous behavior of these two quantities . under tensile strain
the frequency decreases , i.e. , the out - of - plane phonons soften , while in the case of compressive strain it increases .
the grneisen parameter for the a@xmath37 mode ( i.e. , the g mode at the @xmath41 point ) is given by @xmath42 , where @xmath43 is the difference of the frequencies with and without strain and @xmath44 is the frequency of the unstrained single ql .
we obtain positive values of 1.2 and 2.6 for 2.5% and 5% tensile strain , respectively , because the distance between the bi and se sublattices is reduced .
accordingly , negative values appear for compressive strain .
so far no experimental data are available for comparison , though an experimental confirmation of our predictions by raman spectroscopy would be desirable .
we now turn to the te properties of nanostructured bi@xmath0se@xmath1 with varying thickness . @xmath4 and @xmath45 are shown in fig . 6 as functions of the hole doping @xmath46 ( in @xmath47 ) . the highest seebeck coefficient is found for the single ql around @xmath48 @xmath47 . at low doping the value progressively reduces when the number of qls grows , because the bandgap decreases . above
@xmath49 @xmath47 the seebeck coefficient decreases in each case for increasing hole doping .
the reasons for the particularly high values found for the single ql are ( 1 ) the maximal bandgap and ( 2 ) the qualitatively different shape of the valence bands .
the boltztrap code obtains the electrical conductivity as @xmath50 .
we estimate @xmath51 for bi@xmath0se@xmath1 by comparing the calculated values of @xmath50 and @xmath4 with the experimental values of @xmath4 and @xmath3 at 300 k , which results in @xmath52 s. this value is used for all subsequent calculations of the te properties . a similar methodology has been used to establish the relaxation time in bi@xmath0te@xmath1 @xcite and has led to reasonable agreement with the experiment . the assumption that @xmath51 is independent of both the thickness and the strain is supported by previous theoretical work on sb@xmath0te@xmath1 @xcite .
experimentally , a thermal conductivity of @xmath53 w / km at a temperature of 300 k has been reported for nanostructured bi@xmath0se@xmath1 in ref .
@xcite and is used in the following . according to fig .
6 , the maximum of @xmath45 for the single ql is found at a hole doping of @xmath54 @xmath47 , which is below the degenerate limit .
it shifts to higher @xmath46 when the number of qls grows , except for the case of 2 qls .
can the value be further improved by strain ?
we find that the maximal seebeck coefficient shifts to higher and lower hole doping , respectively , for tensile and compressive strain . while @xmath45 is reduced by tensile strain as well as by strong compressive strain , a moderate compressive strain results in a significant enhancement around @xmath55 @xmath47 .
using the experimental @xmath53 w / km and calculated @xmath52 s we estimate a figure of merit of 0.27 for the single ql , which is remarkably more than the bulk value ( 0.10 ) . a moderate compressive strain ( 2.5% )
further enhances the value to 0.30 .
in addition , it can be expected that @xmath4 and thus the figure of merit grow with the temperature and that a decreasing @xmath51 for higher doping also has positive effects . in conclusion , we have established the dependence of the te properties of nanostructured bi@xmath0se@xmath1 on the thickness of the material and on applied compressive and tensile strain by density functional theory and semi - classical boltzmann theory . we have demonstrated that a careful description of structural properties , mainly the thickness of the qls and their separation , as determined by the van der waals interaction , are essential for obtaining quantitatively correct bandgaps .
we have shown that the bandgap decreases when the number of qls increases , which results in smaller seebeck coefficients .
the bandgap of a single ql of bi@xmath0se@xmath1 reaches a value of 0.6 ev for 5% compressive strain , whereas we obtain 0.16 ev for 5% tensile strain .
we have confirmed the structural stability of the systems under strain by means of phonon dispersions .
the highest figure of merit is found in the thinnest system ( single ql ) and our results demonstrate that it is further enhanced by compressive strain .
a value of 0.30 is achieved for a feasible strain of 2.5% , which is three times the bulk value .
strain engineering of nanostructured bi@xmath0se@xmath1 therefore turns out to be a highly efficient approach for creating high performance te materials . |
the cleo experiment , located on the cornell electron storage ring ( cesr ) , consists of a general - purpose particle physics detector @xcite used to reconstruct the products of symmetric @xmath1 collisions with centre - of - mass energies up to approximately 10 gev , corresponding to the @xmath2 resonance .
luminosity - increasing improvements in the cesr optics , which required the placement of superconducting quadrupole magnets nearer to the @xmath1 interaction region , precipitated upgrades to several inner subsystems of the detector prior to the cleo iii datataking run .
the main outer components of the cleo detector are a csi electromagnetic calorimeter , a 1.5 t solenoidal superconducting magnet , and muon detectors .
the upgraded inner subsystems consist of , in order of decreasing radius with respect to the @xmath1 beam line , the following four devices : a 230@xmath3400 channel ring - imaging erenkov ( rich ) detector for charged - hadron identification @xcite ; a new central drift chamber @xcite , with 9796 cells arranged in 47 layers , for charged - particle tracking ; a four - layer double - sided silicon microstrip vertex detector @xcite with 125@xmath3000 readout channels for precise tracking and decay vertexing ; and a thinly walled beryllium beam pipe @xcite to minimize tracking uncertainties due to multiple - scattering effects .
an active coolant - control farm , which is the focus of this article , was used to provide heat removal and mechanical stability ( via temperature control ) for the above four upgraded inner cleo subsystems . in the case of the rich detector ,
the principal heat sources were chains of viking @xcite front - end signal processing chips , with a combined power output of @xmath4360 w. the drift chamber s heat sources consisted of a total of @xmath4150 w of power dissipated by pre - amplifiers mounted on each of the two end plates ; temperature stability across the end plates was crucial to prevent wire breakage .
the silicon detector produced @xmath4488 w of power from 122 beo hybrid boards of front - end electronics chips @xcite , each dissipating 4 w ; spatial tracking resolution of the silicon layers depended on the alignment precision of the sensors and therefore the temperature stability .
the be beam pipe was itself passive , but required a design heat - removal capability up to the @xmath5 kw range due to the possibility of higher - order mode heating in the cesr @xmath1 collider . in this article
, we describe the approach taken in designing the cleo coolant - control farm , studies of the hydrocarbon pf-200ig and its suitability as a heat - transfer fluid , mechanical construction of the farm platforms , the sensor elements , the process - control and diagnostic systems , and the farm s performance in the cleo iii datataking run .
the principal challenge posed in the design of the cleo cooling system lay in the requirement that multiple detector subsystems , each with markedly different plumbing layouts , power loads , pressure tolerances , and physical locations , were to be actively cooled by a ` farm ' of independent coolant - control platforms governed and monitored by centralized control and diagnostic systems , respectively . rather than design separate custom coolant circuits tailored to the specifications of each detector subsystem , our approach was to found the system on a single simple generic design , one that possessed the flexibility to satisfy the cooling requirements of any of the subsystems without extensive modification .
the advantages of a scalable generic design were numerous .
each system had the same basic set of fittings and sensors , introducing economies of scale and easing spare - component inventories .
the modularity of the design was intended to allow for the rapid swapping of an entire coolant - control platform with a spare unit in case of a problem .
the offending platform could then be removed _ en masse _ from the radiation area for diagnosis and repair , without the accrual of excessive downtime .
technician training , serviceability , and the management of on - call experts were simpler in this farm paradigm ; separate cooling experts were not required for different detector subsystems , and the total number of experts needed was reduced .
our use of a modular farm of nearly identical cooling platforms lent itself well to a unified process - control system , which we describe in section [ sect : control ] .
scaling of the control electronics to accommodate additional coolant platforms was designed to be a facile matter of adding more input / output channels and updating the control logic .
all electronic sensors , including those not used for process - control variables , were read out by the control electronics . a separate diagnostic system , described in section [ sect : diagnostics ] , therefore had no need to interact with the farm sensors directly ; instead , all the diagnostic information was acquired through the control infrastructure .
the diagnostic system was designed to provide globally accessible , minute - by - minute performance information browsable on the world wide web ( www ) .
beneficial to the farm concept was a simple , rapid , and independent channel of communication between the coolant - control electronics and the detector subsystems proper .
instead of using a subsystem - based temperature reading , which would have required significant customization on each platform , the cooling platforms used the temperatures of their coolant supply as process variables , thereby precluding a need for the detector subsystems to communicate with any aspect of the cooling system .
communication in the other direction , however , namely from the process - control electronics to the detector subsystems , was possible in the form of interlocks that the cooling system could breach in the event of serious performance problems .
each cooling platform was designed to maintain a specific , user - defined , fixed coolant - supply set - point temperature at approximately a constant flow rate of heat - transfer fluid , with an active feedback system that automatically compensated for changes in subsystem thermal power load , ambient temperature and humidity changes , and variations in heat - sink water temperatures and flow rates .
supply set - point temperatures were remotely selectable , and were designed to remain at most within @xmath6 k of the requested temperature .
all the farm platforms were designed to deliver flow rates up to approximately 23 l / min and to handle heat loads of up to approximately 1 kw .
table [ tab : specs ] provides a summary of the principal operating parameters of a generic cleo coolant - control platform design ; subsystem - specific requirements are described in section [ sect : special ] below .
.summary of principal design parameters of a generic cleo cooling platform . [ cols="<,^,^",options="header " , ] the pf-200ig solvent was described by the manufacturer as non - corrosive to metals such as aluminium , copper , magnesium , and stainless steel @xcite .
motivated by the type of metal used to construct the cleo beam pipe , we undertook a study to examine the compatibility of pf-200ig with beryllium .
three beryllium plates with dimensions 5.3 cm @xmath7 2.5 cm @xmath7 0.3 cm were each coated on one side with a corrosion resistant primer ( br127 ) and immersed in volumes of de - ionized water , pf-200ig , and air ( the control sample ) , respectively .
following a period of 3 months at ambient temperature , pressure , and light exposure , there were no visibly discernible changes ; the samples were subsequently placed near cesr where they absorbed a radiation dose of @xmath425 krad . with still no visually apparent deterioration in the exposed or primer - coated sides of the beryllium plates
, we used a scanning electron microscope to examine the sample surfaces , micrographs of which are depicted in figure [ fig : sem ] .
we observed that the sample that was in contact with pf-200ig ( figure [ fig : sem](c ) ) exhibited less surface modification than the sample that was in contact with h@xmath8o ( figure [ fig : sem](b ) ) , as compared to the control sample ( figure [ fig : sem](a ) ) .
based on these observations , we concluded that pf-200ig was at least as compatible with beryllium as de - ionized water .
= 4.cm = 4.cm = 4.cm in addition to our studies with beryllium , we tested the compatibility of pf-200ig with other materials : cesium iodide , buna and viton elastomers , push - lok rubber hose , brass and stainless - steel fittings , copper , aluminium , plastic , and both polypropylene and nylon tubing .
our tests consisted of recording the masses and dimensions of the material samples and immersing them in containers of pf-200ig under ambient conditions for a period of @xmath48 weeks , whereupon we measured the mass changes and the fractions of linear swell .
for each of the materials tested , the observed changes in mass and size were negligible .
the mechanical design of the coolant - control platforms was driven primarily by the following criteria : reliability and serviceability ; modularity ; elevation and footprint ; mobility ; and ease of access to gauges , valves , filters , and reservoirs .
the limited space in the `` pit '' beneath the cleo detector , an approximately 16 m@xmath9 area of @xmath497 cm high crawl space , dictated a footprint of 76 cm @xmath7 76 cm and an elevation of @xmath487 cm for each of the coolant - control platforms , including the rich active - manifold platform .
each member of the farm was constructed on a rubber - footed skid of 0.6 cm thick aluminium , enabling a degree of mobility and easy access to platform facilities .
figure [ fig : platform_schematic ] depicts a schematic of the coolant flow circuit on board one of the platforms in the farm . upon leaving the reservoir due to suction from the pump ,
pf-200ig coolant reached the pump inlet by way of either a branch through the hot side of a heat exchanger or via one of two bypass shunts .
the cold side of the heat exchanger , the primary heat sink for the platform consisting of a 20-brazed - plate cetetherm ( cetetherm ab , ronneby , sweden ) honeycomb unit , was connected to a closed water circuit driven by a chiller system ( refer to figure [ fig : farm_schematic ] ) .
the amount of flow through the heat exchanger was regulated by a proportioning valve powered by a computer - controlled step motor ; the fraction that the valve was open constituted the control variable in the feedback system .
unregulated 12 vdc power supplies were mounted on each platform to power the proportioning - valve step motors .
= 14.cm the two heat - exchanger shunt branches each contained a ball valve and a normally - open solenoid valve .
one design goal of the bypass system was to guarantee that there be flow to the detector subsystem at all times , assuming that both solenoid valves could never be simultaneously closed while the proportioning valve was 0% open .
dual bypass branches were implemented in order to expand the dynamic range of the system using binary flow logic . for a given fixed fractional flow rate in the bypass branches ,
large variations in the subsystem s heat load meant that the proportioning valve alone would not provide enough compensation to maintain a stable set - point temperature . using solenoid valves to switch the two bypass branches ,
each with a fixed ball - valve setting , resulted in four discrete bypass flow configurations . with ball - valve settings appropriately chosen , the proportioning valve could provide full analog coverage for continuous flow changes intermediate to the four binary combinations of the two solenoid valves .
we note that an alternative to this scheme to maximize the dynamic range would be simply to use a second proportioning valve in the heat - exchanger bypass shunt ; we did not adopt this approach for reasons of reliability and economy .
once through the outlet port of the pump , the pf-200ig could either bypass back through the filter and into the reservoir or depart the platform for transport to the detector subsystem , as indicated in figure [ fig : platform_schematic ] .
the global rate of pf-200ig flow leaving the platform was configured by partially closing the ball valve on the bypass branch leading back to the reservoir . maximizing the fraction of flow through this bypass branch greatly assisted the feedback process by pre - cooling the temperature of the reservoir contents to near the set - point value .
the plumbing on the coolant - control platforms consisted primarily of brass 3/4 npt threaded pipe fittings connected using approximately 79 nipples ( sealed with loctite pst 567 ) per platform .
each platform also had @xmath47 ball valves , @xmath417 elbows , @xmath46 couplings , @xmath414 tees , and several bushings .
reservoirs were constructed from stainless steel , and the filter housings , also stainless steel , contained 75 @xmath10 m polypropylene filters .
the self - priming centrifugal pumps had stainless - steel casings , 750 w ( 2.2 kw for the drift - chamber platform ) single - phase electric motors , and a 34 m ( 45 m for the drift - chamber platform ) water head rating .
our determinations of these pump - head specifications took into account the flow requirements through the detector subsystems , the reduced density of pf-200ig ( refer to table [ tab : props ] ) , the diameters and elevations of the plumbing runs , and the need for adequate flow in the bypass branches for optimum temperature control of the heat - transfer fluid . throughout the plumbing of each platform ,
12 stainless - steel unions were used to aid in the servicing of different components .
an additional emergency bypass shunt containing a 690 kpa pressure relief valve linked the pump outlet port with the reservoir in the event of an overpressure situation .
refer to figure [ fig : platform_drawing ] for an assembly drawing of a typical coolant - control platform .
= 16.cm at the maximum 23 l / min flow rates required by the design , the vapour pressure of pf-200ig ( refer to table [ tab : props ] ) was low enough to ensure that the cavitation number was well in excess of the incipient cavitation value .
no pressurization of the coolant circuits beyond ambient was therefore deemed necessary to avoid cavitation , permitting us to vent any of the reservoirs to the atmosphere during operation . in practice
, we only vented the reservoir in the case of the beryllium beampipe cooling platform , where the differential pressure limit on the thin walls of the beryllium cooling channels was required not to exceed @xmath4203 kpa ( refer to section [ sect : special ] ) , a criterion that we also explicitly imposed on the system pressure near the outlet port of the pump on the platform proper . in lieu of the fixed ( 690 kpa ) pressure relief valve
, we installed an adjustable unit with the range 103 @xmath0 165 kpa .
as an extra precaution , the beampipe coolant - control platform also had a graphite rupture disc rated at 207 kpa . during prototype testing
, we observed that this disc would rupture at what appeared to be lower than the rated pressure , as indicated by a glycerin - damped analog visual gauge .
we ascribed this to transient pressure pulses from the pump s impeller and relief - valve oscillations ; we subsequently installed a water - hammer suppressor in an effort to minimize the effect of these pulses .
figure [ fig : farm_schematic ] depicts the main elements of the coolant - control farm , located in the pit beneath the cleo detector , and the flow of cooling fluids ( air , liquid water , and pf-200ig hydrocarbon ) between them .
wherever possible , brass quick - disconnect fittings were used to link up the liquid connections , which were insulated with ap armaflex .
a closed - circuit water chiller system provided water near a temperature of 282 k to the cold sides of the heat exchangers residing on each of the coolant - control platforms ( refer to figure [ fig : platform_schematic ] ) . in order to assist the water chillers , a heat exchanger to 280
k building water was inserted to pre - cool the returning coolant .
three air handlers , their compressor coils cooled using 280 k building water , maintained a flow of cool dry air across the farm platforms . also shown in figure [ fig : farm_schematic ] is the rich active - manifold platform , a dedicated module that consisted of two one - to - five manifolds , the supply manifold outfitted with five computer - controlled proportioning valves and the return manifold instrumented with five flow meters and transmitters .
on or near the coolant - control platforms , the relatively harsh environment due to dust , water , hydrocarbons , emi from pump and fan motors , and vibration demanded a design that made use of industrial sensor technologies . in order to minimize the amount of delicate electronics in the cleo pit , the analog - to - digital conversion of sensor signals took place in the control - system crate , described in section [ sect : control ] .
standard industrial 4@xmath020 ma transmitter technology was used to condition and send analog sensor signals from the farm platforms to the control system .
current transmitters had the advantages of greater noise immunity and an ability to send analog signals over relatively long distances .
in addition , the remote sensors in the cleo pit could be powered ( usually using @xmath43.5 ma of excitation current ) and read out using a single shielded two - wire current loop , with no local power supply requirements . because of the possibility for rigourous environmental conditions in the cleo pit , all sensors in the design were required to have sealed weatherproof enclosures that fulfilled the specifications of the nema-4x standard . as indicated in figure
[ fig : platform_schematic ] , each coolant - control platform was instrumented with two temperature sensors , one supply and one return , to provide a measurement of the temperature rise due to the combination of subsystem power load , frictional heating of the moving coolant , and ambient heat transfer .
the coolant supply temperature measurement was particularly critical since it served as the process variable in each of the main feedback control systems .
this required the use of temperature sensors that were accurate and stable to less than @xmath110.1 k with low noise and good linearity characteristics .
we investigated four temperature measurement technologies : thermocouples , resistance temperature detectors ( rtds ) , thermistors , and solid - state integrated circuits .
thermistors were too unreliable and nonlinear .
we used solid - state devices , based on the ad592 ( analog devices , inc . ) integrated circuit , in a prototype coolant - control platform , finding them to be attractive due to their relatively good linearity and well - defined characteristics , but difficult to insert into the coolant flow stream reliably and necessitating the design and production of a custom , weatherproof , diode - protected , 4@xmath020 ma transmitter circuit ( using , _
e.g. _ , the ad693 device ) .
we concluded that the ad592 integrated - circuit sensors were better suited for measuring temperatures at the surfaces of solid objects rather than inside flowing fluids .
we disfavoured a thermocouple solution because of the costs of the cold - junction compensation and linearization circuitry needed to achieve the desired accuracy .
our final design used tip - sensitive 100 @xmath12 ( industry standard iec 751 ) platinum rtds manufactured by minco products , inc . ; once combined with some relatively simple off - the - shelf linearization and 4@xmath020 ma transmitter circuitry , rtds were more linear , accurate , and sensitive than thermocouple implementations of comparable cost .
a weatherproof 316 stainless - steel thermowell assembly safely housed the rtd and the front - end electronics , allowed for the effective insertion of the rtd directly into the flow of coolant , and permitted the replacement of the platinum element without having to breach the volume of heat - transfer fluid .
in addition to the pairs of temperature sensors instrumenting each of the farm platforms ( as shown in figure [ fig : platform_schematic ] ) , the rich system had a dedicated array of 32 tip - sensitive 100 @xmath12 platinum rtds distributed azimuthally and on both the electron and positron sides of the detector in coolant - return manifolds located near the rich modules .
for these rtds , ` hockey - puck ' style @xmath13 ma transmitters were mounted together in a single array located atop the cleo detector ; shielded wire leads , 18.3 m in length , connected the rtds to these dedicated transmitters , which were calibrated to compensate for the net lead resistance .
a custom - built multiplexer circuit facilitated addressed read - out of any one of the 32 temperature transmitters at a time . each coolant - control platform had visual pressure and flow - rate gauges to aid in manual adjustments to the flow configuration . in addition , every platform transmitted two pressure measurements , _
i.e. _ , the pressure drop through the cleo subsystem and its plumbing network ( refer to figure [ fig : platform_schematic ] ) , to the control crate using two 0@xmath0689 kpa series 634e pressure sensors from dwyer instruments , inc .
flow rates of heat - transfer fluid were transmitted from density - compensated vane - type flow sensors manufactured by universal flow monitors , inc .
level transmitters in the coolant reservoirs were manufactured by omega engineering , inc . , and consisted of a magnet that was mounted inside a stainless - steel float that tracked up and down an insertion stem while setting a series of reed switches in a voltage divider resistor network .
the accuracies of the pressure , flow , and level sensors and transmitters used were @xmath414 kpa , @xmath41 l / min , and @xmath41.3 cm , respectively .
the heart of the coolant - control system consisted of a small logic controller ( slc ) module , a member of the slc 500 family of programmable controllers manufactured by allen - bradley company , inc .
, that resided in a dedicated 13-slot chassis with an integrated 1 a ( 24 vdc ) power supply module .
the slc 5/04 module used was capable of up to 4096 inputs plus 4096 outputs and had a memory of 32 k words .
the remainder of the process - control system consisted of three other module varieties mounted in the chassis : a 32-channel current - sourcing digital dc output module ( 1746-ob32 ) , three 4-channel 0@xmath020 ma analog output modules ( 1746-no4i ) with 14-bit resolution , and eight 4-channel @xmath1120 ma ( @xmath1110 vdc ) analog input modules ( 1746-ni4 ) with 16-bit resolution .
the 13-slot allen - bradley chassis was mounted inside a crate enclosure positioned in the experimental hall outside the main radiation area . also residing in the enclosure was a 24 vdc/12 a regulated power supply used to energize the entire set of @xmath13 ma sensors remotely deployed both on the coolant farm platforms underneath the cleo detector ( refer to figure [ fig : farm_schematic ] ) , @xmath435 m away , and in the array of 32 rich temperature transmitters on top of the detector , @xmath420 m away .
terminal blocks mounted on a din rail in the enclosure served to interconnect the 24 vdc power supply , the individual sensors , and the appropriate terminals of the analog input modules in the allen - bradley chassis . in the case of the 32 rich transmitter signals , since the array was multiplexed , only a single analog input channel was required .
the slc clocked through the 32 addresses in 0.25 s intervals by using five of the digital output channels in the 1746-ob32 module to switch the 24 vdc of the main sensor power supply .
the slc processor used a ladder - logic programming language in which subroutines were organized into ladders , their rungs each acting as if - then conditional statements . on the left side of every rung one or more
conditions were defined ; the corresponding right side executed one or more actions provided that all the conditions were met for the given rung .
editing of the ladder - logic code was achieved with rslogix 500 ( rockwell automation ) programming software running on a networked intel - based computer located in a clean computing environment and connected to the allen - bradley control crate by a 30 m rs-232 serial connection .
the ladder - logic code was compiled with the rslogix 500 software and was communicated to the slc 5/04 module using a dedicated utility ( rslinx ) . in a similar manner , this serial communication configuration had the capability to allow online edits , parameter adjustments , and diagnostic readout of inputs , outputs , and internal memory structures _ during _ programme execution in the slc .
the allen - bradley slc control code consisted of @xmath4260 rungs organized into a main ladder that cycled through a series of calls to 14 subroutine ladders .
for each of the beampipe , drift - chamber , rich , and silicon subsystems there were subroutines for platform sensor data acquisition , interlock decisions and output , and process control .
the rich system had two extra subroutines , one to read out the multiplexed rich temperature signals and one to perform process - control duties and flow - sensor data acquisition for the rich active - manifold platform . during every cycle of control code execution ,
an interlock decision was taken for each of the farm platforms .
the four criteria forming this decision consisted of a minimum coolant flow rate , minimum and maximum set - point temperatures , and a minimum level of heat - transfer fluid in the reservoir ( for leak detection ) . supply and
return pressure criteria were not included in the interlock decisions . for a given farm platform , if these conditions were all satisfied , the interlock ladder logic would use a channel in the 1746-ob32 output module to energize a normally open relay switch mounted to the din rail near the terminal blocks in the chassis enclosure .
the cooling - interlock relays were connected to a higher - level interlock crate that could switch off power to the subsystem electronics crates in the event of a failure .
the control system was designed such that there would also be an interlock breach if the allen - bradley slc had any interruption in power .
specific to the drift - chamber coolant - control platform , an additional interlock was used to reduce the possibility of cooling the chamber end plates unevenly , a situation that potentially posed deleterious consequences to the chamber s mechanical integrity .
if part or all of the drift - chamber electronics underwent an unexpected loss of power , the power to the centrifugal pump on the coolant - control platform was switched off by means of a relay on a 10-minute delay . in turn
, this would render the minimum - flow - rate criterion unsatisfied , thereby breaking the cooling interlock and removing power from all of the drift - chamber electronics crates . at the core of the process - control logic in each closed feedback loop
was a proportional integral derivative , or pid , instruction tuned to maintain a desired setting of an input process variable by computing appropriate real - time adjustments to an output control variable . for each coolant - control platform , the process variable consisted of an input from the platinum rtd sensor measuring the temperature of the heat - transfer fluid supplied to the detector subsystem ; the control variable consisted of the fractional opening of the heat - exchanger proportioning valve , as set by the step motor positioned using 4@xmath020 ma signals from the 1746-no4i analog output modules ( refer to figure [ fig : platform_schematic ] ) . in order to increase step - motor life ,
a deadband of 0.1 k was used in the pid algorithm ; in this way , the control variable was left unchanged once the process variable passed through the set point and until it was different from the set point by the deadband amount .
for reasons of reliability and flexibility , the diagnostic and process - control systems for the cleo coolant - control farm were kept relatively independent through the use of a two - tiered configuration .
notwithstanding , the same networked computer that was used to upload the compiled ladder - logic code into the allen - bradley slc , as described in section [ sect : control ] , served as the run - time implementation platform for the diagnostic software .
a graphical programming language was used to develop a user interface , also graphical , to the diagnostic system parameters .
the software was based on virtual instruments ( vis ) in the labview environment provided by national instruments corporation . during normal farm operation , _
i.e. _ , when the slc was in a ` standalone ' mode and assumed sole control of the farm , the rs-232 serial connection between the slc and the networked computer ( refer to section [ sect : control ] ) could be used by the labview software to read out regions of the slc memory for diagnostic purposes .
code from proprietary driver and vi libraries ( highwayview , from seg , watertown ma , usa ) transacted the data between the slc and the labview software .
although the diagnostic interface primarily read data from the slc for the purpose of displaying them to users , it did provide password - privileged cooling experts with the ability to adjust the platform set - point temperatures maintained by the control system .
such set - point changes were effected , again using a highwayview vi , by writing data into slc memory via the serial interface .
users of the local area network could therefore `` window in '' to the labview computer to view real - time diagnostics and , if necessary , make limited adjustments to the control parameters . diagnostic information was made available to other subsystem experts and data - acquisition personnel by hypertext transfer - protocol ( http ) servers running on the labview computer . for each platform in the farm , there was an http server capable of delivering streaming quasi - real - time images of the main labview front panel to a client web browser .
an example of one of these web - based front - panel displays is given in figure [ fig : labviewfp_subsystem ] .
other web accessible displays included special diagnostics for the rich temperature - sensor array ( section [ sect : sensors ] ) , rich active - manifold parameters ( section [ sect : mech ] ) , and virtual strip charts giving a graphical representation of the 12-hour history of some of the more important system parameters .
in addition , every minute , the labview machinery updated a detailed text - only html file containing an expert diagnostics digest that listed the status of all farm parameters .
this diagnostic file was written to public web space and was accessible remotely using the www .
in particular , the digest file format lent itself well to remote small - screened portable devices like personal digital assistants using mobile infrastructure software such as that available from avantgo , inc .
long - term archiving of the coolant - control farm operating history was achieved by taking half - hourly snapshots of the expert diagnostics digest file , compressing them , and storing them in a database .
maintenance of the farm platforms proved to be minimal , consisting of weekly visual inspections and annual cleanings and polypropylene filter replacements .
some minor coolant leaks , requiring occasional top ups , were detected by the observation of changes in reservoir level readings with time . in the beampipe and silicon - detector systems ,
which each had longer runs of push - lok rubber hose connecting the platforms to the subsystems , the pf-200ig acquired an orange colour , contrary to earlier compatibility studies ( refer to section [ sect : becompat ] ) .
we attributed this discolouration to the dissolution of a powdered protectant with which the inner surfaces of the push - lok rubber had been treated ; although no degradation of the hose was observed , we recommend using polypropylene or nylon tubing in future applications . operationally , the dynamic ranges of the proportioning valves alone precluded a need to switch the solenoids in the heat - exchanger bypass shunts in order to maintain pid control . with fixed ball - valve openings in the two shunt branches and a deadband setting of @xmath110.1 k in all the pid loops , supply temperatures were asserted with a stability of @xmath110.2 k with respect to the set - point values for the duration of the cleo iii datataking period .
this degree of control was exploited during cleo iii commissioning in pedestal sensitivity studies of the cleo beampipe pin - diode radiation monitors @xcite , for which the beampipe coolant - control platform was employed to vary the coolant supply temperature between 293 k and 299 k. the cleo coolant - control farm commenced operation in november 1999 with the beginning of the cleo iii commissioning period .
as described in section [ sect : special ] , prior to march 2000 the silicon farm platform was deployed remotely and was successfully operated and monitored by the central farm control and diagnostics systems , respectively . on 16 march 2000 , a building cooling - water pipe burst , flooding the cleo pit and submerging the farm platforms under @xmath425 cm of water ; all sensors , power supplies , and pumps , after drying and cleaning , survived the event . on a few other occasions ,
many of the farm s interlocks were exercised in power failures or due to temperature variations in the 290 k water supply that caused trips in the chiller compressors ( refer to figure [ fig : farm_schematic ] ) .
the allen - bradley slc 5/04 module ( described in section [ sect : control ] ) , which was equipped with a battery back - up system , was rendered immune to brief power glitches and , it was found , could be relied upon as though it were a firmware device .
the cleo iii detector began taking physics quality data in july 2000 for a run that ended in june 2001 , with a scheduled three - week down in september 2000 .
the coolant - control farm induced no cleo or cesr downtime in this period , during which a time - integrated luminosity of @xmath49 fb@xmath14 was accumulated .
we have described a novel approach to active particle - detector cooling that is based upon a farm of modular coolant - control platforms charged with the hydrocarbon solvent pf-200ig , uniquely used as a heat - transfer fluid @xcite . during the cleo iii datataking run , the farm provided reliable cooling support to the rich detector , the drift chamber , the silicon vertex detector , and the beryllium beam pipe , with a temperature stability that exceeded design specifications .
the cleo hydrocarbon coolant farm will see continued service in an upcoming programme of operation to explore the @xmath15 resonances , charm physics , and quantum chromodynamics ( cesr - c / cleo - c ) @xcite .
as mentioned in section [ sect : special ] , the drift - chamber platform , by virtue of its greater flow capacity , is being upgraded to provide additional cooling for passive permanent - magnet ( ndfeb ) quadrupole elements @xcite that are being installed prior to the @xmath15 resonance running period .
other aspects of the farm will remain the same .
we have shown that a centrally controlled and monitored farm of generic active coolant - control platforms , running the liquid hydrocarbon solvent pf-200ig as a heat - transfer fluid , can provide independent and regulated heat removal from several different subsystems in a particle - physics detector .
aspects of this design are applicable to future high - energy physics apparatus where flexibility , minimal maintenance , and the ability to monitor and operate detector and accelerator systems remotely will be progressively important .
we would like to thank k. powers and g. trutt of the wilson synchrotron laboratory , cornell university , for their excellent technical support .
this work was supported by the u.s .
national science foundation , the u.s .
department of energy , and the natural sciences and engineering research council of canada .
p - t technologies , a division of lps laboratories , 4647 hugh howell rd .
, p. o. box 105052 , tucker , georgia 30085 - 5052 , usa ; material safety data sheet and private communication .
note that the substance pf-200ig is a discontinued product ; possible replacement heat - transfer fluids are being considered , including pf-145hp , also manufactured by p - t technologies .
cleo collaboration , cleo - c and cesr - c taskforce , `` cleo - c and cesr - c : a new frontier of weak and strong interactions '' , cornell report no .
clns 01/1742 , 2001 ( unpublished ) ; l.k .
gibbons , [ hep - ex/0107079 ] . | we describe a novel approach to particle - detector cooling in which a modular farm of active coolant - control platforms provides independent and regulated heat removal from four recently upgraded subsystems of the cleo detector : the ring - imaging erenkov detector , the drift chamber , the silicon vertex detector , and the beryllium beam pipe .
we report on several aspects of the system : the suitability of using the aliphatic - hydrocarbon solvent pf-200ig as a heat - transfer fluid , the sensor elements and the mechanical design of the farm platforms , a control system that is founded upon a commercial programmable logic controller employed in industrial process - control applications , and a diagnostic system based on virtual instrumentation .
we summarize the system s performance and point out the potential application of the design to future high - energy physics apparatus .
clns 01/1754 , , , , , , , , , , , , , and .
cleo , detector , cooling , process control , hydrocarbon heat - transfer fluids 29.90.+r , 07.07.@xmath0a , 07.05.dz , 65.20.+w |
within qcd , wilson s operator product expansion ( ope ) @xcite finds its most straightforward application in lepton hadron deep inelastic scattering ( dis ) . the cross section for this process
is related to the time - ordered product of two currents , @xmath5 $ ] , which , at short distances , is expanded as @xmath6=\sum_k c_k(x)\hs o_k(0).\ ] ] the coefficient functions @xmath7 embody the short - distance structure of the amplitude , while @xmath8 are local operators whose matrix elements describe long - distance effects .
for dis the ope leads to an expansion of the forward compton amplitude which , by use of dispersion relations @xcite , gives predictions for the moments of the dis inclusive cross section .
unfortunately , for more general hard processes the straightforward use of the ope technique fails .
two ways out of this impasse have been proposed .
the first one @xcite relies on a careful study of collinear and soft singularities order by order in perturbation theory and aims at proving that such singularities can all be lumped into some universal ( i.e. hard - process independent ) functions .
this method has the advantage of being close to the physical parton picture for the hard process . as an alternative , an expansion in terms of _ cut vertices
_ was proposed by mueller @xcite as a more direct extension of the ope . although more formal , the method of cut vertices has the advantage of giving directly an expansion for the absorptive part of the amplitude , i.e. for the cross section , in terms of the convolution of a cut vertex with a coefficient function .
the cut vertex contains all the long - distance effects , and , as such , generalizes the matrix element of a local operator , while the coefficient function , as usual , is calculable in perturbation theory .
the cut vertex method has been used in ref.@xcite to deal with a variety of hard processes in qcd .
let us consider a deep inelastic semi - inclusive reaction in which a hadron in the final state is detected .
when the transverse momentum of the hadron is of order @xmath9 the usual current fragmentation mechanism holds @xcite . in the region
in which the transverse momentum of the produced hadron is much smaller than @xmath9 , the so - called target fragmentation region , the simple description based on parton densities and fragmentation functions fails .
for this reason , in order to describe the target fragmentation region as well , a new formulation of semi - inclusive hard processes in terms of _ fracture functions _ was proposed @xcite and later developed @xcite .
a fracture function @xmath10 gives the probability of extracting a parton @xmath11 with momentum fraction @xmath12 from a hadron @xmath13 while observing a final hadron @xmath14 in the target fragmentation region with longitudinal momentum fraction @xmath15 .
the possible relevance of such an object was already mentioned by feynman @xcite before the advent of qcd . in this paper
we will consider the case in which also the transverse momentum of the hadron @xmath14 ( or equivalently the momentum transfer @xmath16 ) is observed .
we will argue that , in the region @xmath17 , an expansion in terms of cut vertices can be given and that a @xmath0 dependent ( or extended ) fracture function can be defined in terms of a new kind of cut vertex .
we will show , within a @xmath18 toy - model field theory , that extended factorization is just a consequence of collinear power counting @xcite .
the paper is organized as follows . in sect .
2 we recall the definition of cut vertices and we obtain the expansion of the moments of the structure function in terms of matrix elements of local operators times coefficient functions by using cut vertices . in sect .
3 the cut vertex approach is extended to deal with semi - inclusive processes and a generalized cut vertex expansion is given . in sect .
4 we define extended fracture functions and their evolution equation is given and discussed . in sect .
5 we discuss the results and draw our conclusions .
in this section we will recall the definition of cut vertices in the simpler case of @xmath18 theory . this toy model , despite its simpler structure , shares several important properties with qcd .
it has a dimensionless , asymptotically free coupling constant and the diagrams with leading mass singularities have the same topology as in ( light cone gauge ) qcd . for these reasons
@xmath18 is an excellent theoretical framework for the study of factorization properties @xcite .
consider the inclusive deep inelastic process @xmath19 off the current @xmath20 .
we define as usual @xmath21 and @xmath12 as @xmath22 let us choose a frame in which @xmath23 with @xmath24 and @xmath25 . given a vector @xmath26
define @xmath27 . the structure function , defined as @xmath28 describes the interaction of the far off - shell current @xmath29 with an elementary quantum of momentum @xmath30 through the discontinuity of the forward scattering amplitude ( see fig .
1 ) . [ cols="^ " , ] [ 7 ] inverting the moments and expressing the extended fracture function in terms of the usual bjorken variable @xmath12 , one finds that @xmath31 obeys the simple homogeneous evolution equation @xmath32 where @xmath33 , defined as @xmath34 is the same dglap kernel which controls the evolution of the ordinary parton distribution functions .
this result looks particularly appealing since it means that the evolution of the extended fracture function follows the usual perturbative behaviour .
one may ask at this point how this result matches with the peculiar equation which drives the evolution of ordinary fracture functions @xcite : @xmath35 the evolution equation for @xmath36 contains two terms : a homogeneous term describing the non - perturbative production of a hadron coming from target fragmentation and an inhomogeneous term whose origin is the perturbative fragmentation due to initial state bremsstrahlung . as discussed in ref .
@xcite , the separation between perturbative and non - perturbative fragmentation introduces an arbitrary scale but the fracture function itself @xmath36 does not depend on it .
it can be shown @xcite that the evolution equation ( [ inomo ] ) can be derived by an explicit calculation using eq .
( [ eqomo ] ) together with jet calculus rules @xcite . by defining in fact the ordinary fracture function as an integral over @xmath0 up to a cut - off of order @xmath21 , e.g. @xmath37 with @xmath38 : @xmath39 the inhomogeneous term in the evolution equation is obtained by taking into account the @xmath21 dependence of the integration cut - off . moreover , the interplay between the scales @xmath21 and @xmath0 has a sizeable effect in terms of a new class of perturbative corrections of the form @xmath40 . such corrections are large and potentially dangerous in the region @xmath17 since they can ruin a reliable perturbative expansion .
those terms are naturally resummed into eq .
( [ defi ] ) .
for the extended fracture function these corrections do play an important role for understanding the dynamics of semi - inclusive processes in the kinematic region we have been considering here @xcite .
in this paper we have presented an extension of the cut vertex formalism to deal with semi - inclusive processes .
we have shown that an extended fracture function can be defined in terms of a new kind of cut vertex .
such an unintegrated , @xmath0-dependent fracture function obeys a dglap evolution equation with the same kernel as the one appearing in ordinary parton densities .
this result is particularly evident in the cut vertex approach and agrees with the analysis made in a different context in ref.@xcite . as a non - trivial check , the more complicated evolution equation for ordinary fracture functions @xcite can be obtained from the simpler one obeyed by extended fracture functions .
although our results have been obtained within the @xmath18 model , where a treatment based on mass singularities and infra - red power counting takes a much simpler form , we believe the extension to qcd to be harmless . indeed , the picture that emerges for semi - inclusive hadronic processes in the target fragmentation region appears to be fully consistent with the intuitive qcd - corrected parton picture .
in addition , a new class of logarithmic contributions appears at @xmath41 .
extended fracture functions could give an additional tool to handle this interesting regime which certainly deserves further study @xcite .
we finally note that the relation between ordinary and extended fracture functions is quite analogous to the one between the experimentally defined / measured diffractive structure functions @xmath42 and @xmath43 @xcite which depend upon observing large rapidity gaps .
relating such quantities to fracture functions is another interesting issue for future investigation .
wilson , phys . rev .
* d179 * ( 1969 ) 1499 .
n. christ , b. hasslacher and a.h .
mueller , phys . rev .
* d6 * ( 1972 ) 3543 .
d. amati , r. petronzio and g. veneziano , nucl .
b140 * ( 1978 ) 54 , * b146 * ( 1978 ) 29 ; r.k .
ellis , h. georgi , m. machacek , h.d .
politzer and g. ross , nucl .
b152 * ( 1979 ) 285 .
mueller , phys . rev .
* d18 * ( 1978 ) 3705 ; phys . rep . * 73 * ( 1981 ) 237 .
s. gupta and a.h .
mueller , phys . rev . *
d20 * ( 1979 ) 118 .
g. altarelli , r.k .
ellis , g. martinelli and s.y .
pi , nucl . phys .
* b160 * ( 1979 ) 301 .
l. trentadue and g. veneziano , phys . lett .
* b323 * ( 1994 ) 201 .
d. graudenz , nucl . phys .
* b432 * ( 1994 ) 351 .
d. de florian and r. sassot , phys . rev .
* d56 * ( 1997 ) 426 ; + g.m .
shore and g. veneziano , hep - ph 9709213 .
feynman , in _ photon hadron interactions _
, w.a . benjamin ed .
a. berera and d.e .
soper , phys . rev .
* d53 * ( 1996 ) 6162 .
g. sterman , phys . rev .
* d17 * ( 1978 ) 2773 , 2789 .
collins , d.e .
soper and g. sterman , in _ perturbative quantum chromodynamics _ , ed .
mueller ( world scientific , singapore , 1989 ) 1 .
taylor , phys . lett .
* b73 * ( 1978 ) 85 ; y. kazama and y.p .
yao , phys .
* 41 * ( 1978 ) 611 , phys
d19 * ( 1979 ) 3111 ; t. kubota , nucl . phys .
* b165 * ( 1980 ) 277 ; l. baulieu , e.g. floratos and c. kounnas , phys . rev .
* d23 * ( 1981 ) 2464 .
l. baulieu , e.g. floratos and c. kounnas , nucl .
* b166 * ( 1980 ) 321 .
t. munehisa , prog .
* 67 * ( 1982 ) 882 .
m. grazzini , uprf-97 - 09 , hep - ph/9709312 .
g. camici , m. grazzini and l. trentadue , _ in preparation_. k. konishi , a. ukawa and g. veneziano , nucl . phys .
* b157 * ( 1979 ) 45 .
h1 collaboration , t. ahmed et al .
, phys . lett . *
b348 * ( 1995 ) 681 ; zeus collaboration , m. derrick et al .
, z. phys .
* c68 * ( 1995 ) 569 . | using a generalized cut vertex expansion we introduce the concept of an extended fracture function for the description of semi - inclusive deep inelastic processes in the target fragmentation region .
extended fracture functions are shown to obey a standard homogeneous dglap - type equation which , upon integration over @xmath0 , becomes the usual inhomogeneous evolution equation for ordinary fracture functions .
cern - th/97 - 248 + uprf-97 - 011 + september 1997 + fracture functions from cut vertices + 1.0 cm m. grazzini@xmath1 , l. trentadue@xmath1 and g. veneziano@xmath2 + .5 cm @xmath3 _ dipartimento di fisica , universit di parma and + infn gruppo collegato di parma , 43100 parma , italy + @xmath4 _ theory division , cern , ch-1211 , geneva , switzerland _ _
+ pacs 13.85.ni |
many elliptical galaxies are not ism - free , which contain a considerable amount of cool dust ( e.g. knapp et al .
1989 ; goudfrooij & de jong 1995 ; temi et al .
2004 ; temi et al .
2007 ) . early _ spitzer _
irs observations have found that even polycyclic aromatic hydrocarbons ( pahs ) are present in elliptical galaxies ( kaneda et al .
the spectra show that the pah 11.3 @xmath0 m feature is notably strong relative to usually the strongest 7.7 @xmath0 m feature , which might be typical in evolved systems such as elliptical galaxies , i.e. the dominance of neutral pahs over ionized ones due to very soft radiation fields from evolved stars .
in addition to the pah features , a series of h@xmath1 rotational lines as well as ionic fine - structure lines ( e.g. [ neii ] and [ neiii ] ) are detected from a large fraction of the elliptical galaxies , implying the presence of warm molecular and ionized gases @xcite . their presence seems to be inconsistent with the dominance of neutral pahs irradiated by soft radiation field .
what are the origin and fate of the pahs in elliptical galaxies ? how are they related with the warm molecular and ionized gases as well as the cool dust
? how do their properties change from the center to outer regions ?
spatial information is indispensable to disentangle various competing components and tackle these issues . to pursue follow - on studies of the early _ spitzer _ discoveries , we performed deep spectral mapping observations of elliptical galaxies with the _ spitzer _ irs . in this letter , we report the result of ngc 4589 , a nearby e2 elliptical galaxy with a minor - axis optical dust lane @xcite .
ngc 4589 has a complex stellar rotation field , although its morphology shows the smooth optical profile following the de vaucouleurs law @xcite , which supports that the galaxy is a relatively old merger remnant .
recently , a core - collapse supernova ( sn ib 2005cz ) was observed in this genuine elliptical galaxy @xcite ; sn ib 2005cz is likely to be the end product of one of young stars , which were produced by the galaxy merger about @xmath2 years ago @xcite .
ngc 4589 is an x - ray - emitting elliptical galaxy with a low - ionization nuclear emission - line region ( liner ) nucleus @xcite .
our earlier irs observation detected the pah features from the galaxy in a staring mode but not in a mapping mode , and therefore we could not spatially resolve them @xcite .
after the _ spitzer _ detection of the pah features , their spatial distributions were estimated from the _ akari _ 11 @xmath0 m and 15 @xmath0 m band images , while that of the far - infrared ( ir ) dust was obtained from the _ akari _ 90 @xmath0 m image @xcite .
they all show significant deviations from the smooth stellar distribution , from which kaneda et al .
( 2008b ) suggested that the pahs and far - ir dust are mostly of an interstellar origin , rather than of a stellar origin . in this letter , from the _ spitzer_/irs spectral mapping of ngc 4589 , we clearly find that the pah 11.3 @xmath0 m emission comes from the minor - axis dust lane , while the pah 17 @xmath0 m emission is distributed differently from the pah 11.3 @xmath0 m emission , and more similarly to the far - ir dust emission .
we observed in a spectral mapping mode with the irs ( houck et al .
2004 ) onboard _ spitzer _
@xcite on may 67 , 2009 .
the total observation time was 7.7 hours .
the data were taken in part of our guest observers ( go5 ) program ( pi : h. k. ; program i d : 50369 ) . starting with the basic calibrated data ( bcd ; pipeline version s18.7.0 ) reduced by the _ spitzer _ science center ( ssc ) , we extracted spectral image data by the cubism software ( version 1.7 ) following standard procedures .
figure 1a shows the slit positions of the short - low ( sl ; 5.214.5 @xmath0 m ) and long - low ( ll ; 1436 @xmath0 m ) modules .
we made a @xmath3 grid for sl with a ramp duration of 60 s for each exposure , and a @xmath4 grid for ll with a ramp duration of 30 s for each .
we shifted the slit in directions perpendicular and parallel to its length by @xmath5 and @xmath6 , respectively , for sl , and by @xmath7 and @xmath8 , respectively , for ll .
the 50 % overlap of the slit aperture in the perpendicular direction is crucial for reliable estimation of the slit - loss correction factors for extended sources , which we could not obtain in our earlier irs staring observations .
the sl and ll slit directions are almost perpendicular to each other , resulting in the coverage of @xmath9 area around the center of the galaxy by both modules .
we repeated the same spectral mapping observations two and three times for sl and ll , respectively . from the spectral image data
, we created the spectra of the central @xmath10 region .
figure 1b shows the resultant spectra , where the background spectra were created from eight slit - aperture data at the ends of the spectral mapping area for sl and four slit - aperture data for ll , and subtracted . in fig.1b , we clearly see the prominent pah 11.3 @xmath0 m feature and the pah 17 @xmath0 m broad feature , together with several emission lines such as h@xmath11s(3 ) at 9.7 @xmath0 m , [ neii ] at 12.8 @xmath0 m , [ neiii ] at 15.6 @xmath0 m , h@xmath11s(1 ) at 17.0 @xmath0 m , [ siii ] at 34.8 @xmath0 m , and the mid - ir dust continuum emission at wavelengths longer than 30 @xmath0 m .
photospheric emission from old stars appears to dominate in the continuum emission at wavelengths shorter than 20 @xmath0 m , judging from the slope of the continuum .
hereafter the bin sizes of spectral images are set to be @xmath5 for sl and @xmath12 for ll .
we applied smoothing with a boxcar kernel of 3 pixels in width ( @xmath13 for sl and @xmath14 for ll ) for every spectral map . in table 1
, we list the wavelength ranges used for extracting each spectral component , together with the flux from the central @xmath15 region .
the [ neii ] and h@xmath11s(1 ) line emissions are blended with the underlying pah 12.7 @xmath0 m and 17 @xmath0 m features , respectively . lrrrr @xmath16 @xmath0 m continuum&@xmath16& &47 & + h@xmath11s(3 ) & @xmath17&@xmath18 , @xmath19&2.6 & + pah 11.3 @xmath0m&@xmath20&@xmath21 , @xmath22&11 & + pah 12.7 @xmath0m&@xmath23&@xmath24&2.5 & + @xmath25neii@xmath26 $ ] & @xmath27&@xmath24&1.7&blended with pah 12.7
@xmath0 m + @xmath25neiii@xmath26 $ ] & @xmath28&@xmath29 , @xmath30&0.7 & + h@xmath11s(1 ) & @xmath31&@xmath29 , @xmath30 , @xmath32&2.4&blended with pah 17 @xmath0 m + pah 17 @xmath0m&@xmath33&@xmath29 , @xmath30 , @xmath32&2.9 & + @xmath34 @xmath0 m continuum&@xmath35& &18 & + @xmath25siii@xmath26 $ ] & @xmath36&@xmath35&3.0 & +
figure 2 shows the contour maps of the pah 11.3 @xmath0 m and 17 @xmath0 m emission features overlaid on the contour maps of the @xmath16 @xmath0 m and @xmath34 @xmath0 m continuum emissions , respectively .
the @xmath16 @xmath0 m continuum emission shows a smooth stellar distribution , which is consistent with the _ 2mass _ image in fig.1a .
however , the pah 11.3 @xmath0 m emission exhibits a distinctly elongated distribution , posessing an excellent spatial coincidence with the minor - axis optical dust lane ( see fig.2 of mllenhoff & bender 1989 ) .
thus we for the first time obtain direct evidence that the pah emission comes predominantly from the optical dust lane of the elliptical galaxy .
the pah 17 @xmath0 m feature shows a spatial distribution different from that of the pah 11.3 @xmath0 m emission , apparently extending to the southeast and the west .
since the pah 17 @xmath0 m emission has a relatively small feature - to - continuum ratio , the distribution may be sensitive to uncertainties in the background subtraction . as plotted in fig.2b
, however , the @xmath37 @xmath0 m background continuum shows a smooth elliptical distribution entirely different from the pah 17 @xmath0 m feature and similar to the @xmath16 @xmath0 m continuum , supporting that the stellar emission dominates in the continuum at @xmath38 20 @xmath0 m .
thus the above result is not affected by the uncertainties in the background subtraction .
both pah distributions in fig.2 are consistent with the earlier estimates based on the _ akari _ multi - band images ( see fig.3 of kaneda et al .
the @xmath34 @xmath0 m continuum emitted by very small dust grains also shows a hint of extension to the southeast , which is , at least , not of an elliptical shape as the stellar continuum emissions show .
the 11.3 @xmath0 m feature is attributed to a c - h out - of - plane bending mode mainly arising from neutral pahs @xcite .
the 17 @xmath0 m broad feature is interpreted as the emission of larger pahs with a few thousand carbon atoms in c c c in- and out - of - plane bending modes @xcite . in general , larger pahs can be excited by softer radiation field , which might cause the observed different distributions .
however such soft radiation as optical emission from old stars is hard enough to excite pahs with the number of c atoms of @xmath39 @xcite , which are main emitters of the 11.3 @xmath0 m feature .
the different distributions are therefore difficult to be explained only by variations in the radiation hardness .
it is likely that the pahs responsible for the 17 @xmath0 m feature belong to a different population from those for the 11.3 @xmath0 m feature .
+ in fig.3 , we show the contour maps of other spectral components , although their s / ns are rather low as compared to the pah 11.3 @xmath0 m emission in fig.2 .
the robust results obtained from these maps are as follows : ( 1 ) the molecular hydrogen line emissions come from the dust lane , similarly to the pah 11.3 @xmath0 m feature .
( 2 ) the [ neii ] line emission shows a more compact distribution near the nucleus than the pah 12.7 @xmath0 m feature . ( 3 ) the [ siii ] line emission shows a spatial distribution extending to the southeast and the west , similarly to the pah 17 @xmath0 m feature .
the presence of the h@xmath11 lines implies that there is far - uv emission from early - type stars or shocks providing collisional heating .
the spatial coincidence with the strong pah 11.3 @xmath0 m feature that reflects soft radiation field supports the latter scenario for heating the molecular gas .
sofue & wakamatsu ( 1993 ) concluded that the molecular gas is probably situated only in the nucleus , since the velocity dispersion of the co line , 105 km s@xmath40 , is significantly lower than the rotation velocity of the dust lane ( 200 km s@xmath40 ; mllenhoff & bender 1989 ) .
however we find that the molecular gas is indeed situated in the dust lane . in fig.3f , the x - ray image retrieved from the _ chandra _ data archive is superposed on the [ siii ] contour map , where the hot plasma also appears to extend from the nucleus to the west and the southeast .
the early irs study found a positive correlation between the [ siii ] line and the x - ray luminosity from a sample of x - ray - emitting dusty elliptical galaxies @xcite .
the spatial correlation in fig.3f , together with the luminosity correlation , may suggest relatively abundant gas - phase si in the hot plasma through sputtering destruction of dust grains where si is depleted .
the highly - ionized gas responsible for the [ neii ] and [ neiii ] lines is likely to be associated with a low - luminosity agn or liner nucleus @xcite , which is consistent with the relatively compact distribution of the [ neii ] line emission as compared to that of the pah 12.7 @xmath0 m feature .
alternatively , the highly - ionized gas might be associated with the above hot plasma , since there is a hint of spatial correspondence in the low - level brightness for the [ neiii ] line emission . from the difference in spatial distribution between the pah features and these lines , we can
, at least , deduce that the properties of the pahs are not significantly regulated by hard radiation from the liner nucleus . in fig.4
, we superpose the contour maps of the pah 11.3 @xmath0 m and 17 @xmath0 m features onto the _ akari _ 90 @xmath0 m image of ngc 4589 @xcite . though their spatial resolutions are different , the distribution of the far - ir dust emission is extended in similar directions to that of the pah 17 @xmath0 m feature , both of which are apparently different from the stellar distribution . since ngc 4589 is thought to be a @xmath2-year - old merger remnant , the distributed dust is likely to have been brought in by the merger , and so are the large pahs emitting the 17 @xmath0 m feature .
considering relatively short destruction timescales of the pahs in the hot plasma , the large pahs might have be produced later through the fragmentation of large carbonaceous grains @xcite .
then , part of the dust and gas are likely to have fallen into the center , which are settled into a ring or a disk ( i.e. the dust lane ) .
the distinctive spatial distribution of the pah 11.3 @xmath0 m emission perfectly following the dust lane indicates that the pahs responsible for the 11.3 @xmath0 m feature are not directly brought in by the merger but secondary products created at later evolutionary stages of the brought - in ism .
the similarity in the distribution between the pah 11.3 @xmath0 m feature and the h@xmath11 line emission suggests that the pahs may be created through grain surface chemistry in the same manner as h@xmath11 formation in the dense gas region . in the absence of hard radiation field ,
the pahs can well exist in superhydrogenated states , which even act as catalysts for interstellar h@xmath11 formation @xcite .
with the _ spitzer_/irs , we performed deep spectral mapping observations of ngc 4589 , a merger remnant with a minor - axis optical dust lane . as a result , we obtain clear evidence that the pah 11.3 @xmath0 m emission comes predominantly from the optical dust lane of the galaxy .
we also detect molecular hydrogen line emissions from the dust lane .
the distribution of the pah 17 @xmath0 m emission is extended more widely in directions similar to that of the far - ir dust continuum emission . from their distinctive distributions
, we suggest that the pahs emitting the 11.3 @xmath0 m feature are secondary products through the evolution of the ism brought in by the merger , while the distributed dust and large pahs emitting the 17 @xmath0 m feature are relics of the merger .
this work is based on observations made with the _ spitzer _ space telescope , which is operated by the jet propulsion laboratory , california institute of technology under nasa contract 1407 .
we would like to thank the irs team and the ssc for their dedicated work in generating the bcd .
we are grateful to the sings team for providing the cubism software .
we would also express many thanks to the anonymous referee for giving us useful comments .
this work has also made use of data obtained from the _ chandra _ data archive .
this research is supported by the grants - in - aid for the scientific research nos .
19740114 and 21740139 , and the nagoya university global coe program , `` quest for fundamental principles in the universe : from particles to the solar system and the cosmos '' , both from the ministry of education , culture , sports , science and technology of japan .
allamandola , l. j. , tielens , a. g. g. m. , & barker , j. r. 1989 , , 71 , 733 draine , b. t. & li , a. 2007 , , 657 , 810 dressel , l. l. & wilson , a. s. 1985 , , 291 , 668 goudfrooij , p. & de jong , t. 1995 , , 298 , 784 hakobyan , a. a. , et al . 2008 , , 488 , 523 kaneda , h. , onaka , t. , & sakon , i. 2005 , , 632 , l83 kaneda , h. , onaka , t. , sakon , i. , kitayama , t. , okada , y. , & suzuki , t. 2008a , 684 , 270 kaneda , h. , suzuki , t. , onaka , t. , okada , y. , & sakon , i. 2008b , , 60 , s467 kawabata , k. s. , et al . 2010 , nature , in press ( arxiv:0906.2811v2 ) knapp , g. r. , guhathakurta , p. , kim , d .- w . , & jura , m. 1989 , , 70 , 329 houck , j. r. , et al .
2004 , , 154 , 18 mllenhoff , c. & bender , r. 1989 , , 214 , 61 onaka , t. , matsumoto , h. , sakon , i. , & kaneda , h. 2010 , , in press ( arxiv:1002.2253 ) osullivan , e. , forbes , d. a. , & ponman , t. j. 2001 , mnras , 328 , 461 peeters , e. , mattioda , a. l. , hudgins , d. m. , & allamandola , l. j. 2004 , , 617 , l65 rauls , e. & hornek@xmath41r , l. 2008 , , 679 , 531 schutte , w. a. , tielens , a. g. g. m. , & allamandola , l. j. 1993 , , 415 , 397 smith , j. d. t. , et al .
2007 , , 656 , 770 sofue , y. & wakamatsu , k. 1993 , pasj , 45 , 529 temi , p. , brighenti , f. , & mathews , w. g. 2007 , , 660 , 1215 temi , p. , brighenti , f. , mathews , w. g. , & bregman , j. d. 2004 , , 151 , 237 tielens , a. g. g. m. 2008 , , 46 , 289 van kerckhoven , c. , et al .
2000 , , 357,1013 werner , m. w. , et al .
2004 , , 154 , 1 willner , s. p. , elvis , m. , fabbiano , g. , lawrence , a. , & ward , m. j. 1985 , , 299 , 443 | _ spitzer _ and _ akari _ observations have found that polycyclic aromatic hydrocarbons ( pahs ) are present in nearby elliptical galaxies , but their spatial distributions are still unknown . in order to investigate their distributions , we performed deep spectral mapping observations of the pah - detected elliptical galaxy ngc 4589 , a merger remnant with a minor - axis optical dust lane . as a result , we obtain clear evidence that the pah 11.3 @xmath0 m emission comes predominantly from the dust lane of the galaxy .
we also detect molecular hydrogen line emissions from the dust lane .
the pah 17 @xmath0 m emission is distributed differently from the pah 11.3 @xmath0 m emission , and more similarly to the dust continuum emission . from their distinctive distributions , we suggest that the pahs responsible for the 11.3 @xmath0 m feature are secondary products through the evolution of the ism brought in by the merger . |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.